diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18678 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.977871655602495,
+  "eval_steps": 500,
+  "global_step": 3100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.0536,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1533827903669656e-06,
+      "loss": 1.8044,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.413030972429927e-06,
+      "loss": 1.6811,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.306765580733931e-06,
+      "loss": 1.8336,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.7437,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.566413762796893e-06,
+      "loss": 1.8636,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.045309775610838e-06,
+      "loss": 1.8633,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4601483711008965e-06,
+      "loss": 1.7105,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.826061944859854e-06,
+      "loss": 1.5974,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.153382790366967e-06,
+      "loss": 1.9246,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.449480512024893e-06,
+      "loss": 2.0995,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.71979655316386e-06,
+      "loss": 1.8925,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.968463205835412e-06,
+      "loss": 1.8551,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.198692565977805e-06,
+      "loss": 1.7944,
+      "step": 14
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.413030972429927e-06,
+      "loss": 1.6847,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.613531161467863e-06,
+      "loss": 1.7153,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.801872138612942e-06,
+      "loss": 1.6666,
+      "step": 17
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.979444735226819e-06,
+      "loss": 1.8019,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.147414002175753e-06,
+      "loss": 1.6039,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.306765580733932e-06,
+      "loss": 1.8896,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.458340748040766e-06,
+      "loss": 1.6843,
+      "step": 21
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.602863302391858e-06,
+      "loss": 1.6808,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.7409604673319e-06,
+      "loss": 1.8264,
+      "step": 23
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.873179343530824e-06,
+      "loss": 2.0215,
+      "step": 24
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 1.7486,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 1.5332,
+      "step": 26
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.996747967479676e-06,
+      "loss": 1.8167,
+      "step": 27
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.993495934959351e-06,
+      "loss": 1.5671,
+      "step": 28
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.990243902439025e-06,
+      "loss": 1.939,
+      "step": 29
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9869918699187e-06,
+      "loss": 1.591,
+      "step": 30
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.983739837398375e-06,
+      "loss": 1.7627,
+      "step": 31
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.98048780487805e-06,
+      "loss": 1.9192,
+      "step": 32
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977235772357724e-06,
+      "loss": 1.6786,
+      "step": 33
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.973983739837399e-06,
+      "loss": 1.8899,
+      "step": 34
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.970731707317074e-06,
+      "loss": 1.6923,
+      "step": 35
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.967479674796748e-06,
+      "loss": 1.5348,
+      "step": 36
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.964227642276424e-06,
+      "loss": 1.5424,
+      "step": 37
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.960975609756098e-06,
+      "loss": 1.687,
+      "step": 38
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.957723577235773e-06,
+      "loss": 1.611,
+      "step": 39
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.954471544715448e-06,
+      "loss": 1.7493,
+      "step": 40
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.951219512195124e-06,
+      "loss": 1.6585,
+      "step": 41
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.947967479674797e-06,
+      "loss": 1.619,
+      "step": 42
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.944715447154472e-06,
+      "loss": 1.8075,
+      "step": 43
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.941463414634147e-06,
+      "loss": 1.6401,
+      "step": 44
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.938211382113821e-06,
+      "loss": 1.6874,
+      "step": 45
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.934959349593498e-06,
+      "loss": 1.6465,
+      "step": 46
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931707317073171e-06,
+      "loss": 1.8225,
+      "step": 47
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928455284552846e-06,
+      "loss": 1.6644,
+      "step": 48
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.925203252032522e-06,
+      "loss": 1.6024,
+      "step": 49
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.921951219512195e-06,
+      "loss": 1.5707,
+      "step": 50
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91869918699187e-06,
+      "loss": 1.9898,
+      "step": 51
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.915447154471545e-06,
+      "loss": 1.615,
+      "step": 52
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.91219512195122e-06,
+      "loss": 1.6191,
+      "step": 53
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.908943089430894e-06,
+      "loss": 1.476,
+      "step": 54
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.905691056910571e-06,
+      "loss": 1.3852,
+      "step": 55
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.902439024390245e-06,
+      "loss": 1.5717,
+      "step": 56
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.89918699186992e-06,
+      "loss": 1.6057,
+      "step": 57
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.895934959349595e-06,
+      "loss": 1.6961,
+      "step": 58
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.892682926829268e-06,
+      "loss": 1.8449,
+      "step": 59
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.889430894308944e-06,
+      "loss": 1.7218,
+      "step": 60
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.886178861788619e-06,
+      "loss": 1.8627,
+      "step": 61
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.882926829268294e-06,
+      "loss": 1.8129,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.879674796747967e-06,
+      "loss": 1.7398,
+      "step": 63
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.876422764227643e-06,
+      "loss": 1.9181,
+      "step": 64
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.873170731707318e-06,
+      "loss": 1.5574,
+      "step": 65
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.869918699186993e-06,
+      "loss": 1.6007,
+      "step": 66
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.6972,
+      "step": 67
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.863414634146342e-06,
+      "loss": 1.7126,
+      "step": 68
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.860162601626017e-06,
+      "loss": 1.806,
+      "step": 69
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.856910569105692e-06,
+      "loss": 1.7234,
+      "step": 70
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.853658536585367e-06,
+      "loss": 1.4732,
+      "step": 71
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.85040650406504e-06,
+      "loss": 1.4904,
+      "step": 72
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.847154471544716e-06,
+      "loss": 1.8472,
+      "step": 73
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.843902439024391e-06,
+      "loss": 1.5605,
+      "step": 74
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.840650406504065e-06,
+      "loss": 1.7829,
+      "step": 75
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.837398373983741e-06,
+      "loss": 1.7221,
+      "step": 76
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.834146341463415e-06,
+      "loss": 1.4963,
+      "step": 77
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.83089430894309e-06,
+      "loss": 1.8138,
+      "step": 78
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.827642276422765e-06,
+      "loss": 1.8428,
+      "step": 79
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.824390243902439e-06,
+      "loss": 1.5902,
+      "step": 80
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.821138211382114e-06,
+      "loss": 1.8195,
+      "step": 81
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.81788617886179e-06,
+      "loss": 1.8778,
+      "step": 82
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.814634146341464e-06,
+      "loss": 1.614,
+      "step": 83
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.811382113821138e-06,
+      "loss": 1.5085,
+      "step": 84
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.808130081300815e-06,
+      "loss": 1.7795,
+      "step": 85
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.804878048780488e-06,
+      "loss": 1.765,
+      "step": 86
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.801626016260163e-06,
+      "loss": 1.7471,
+      "step": 87
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.798373983739839e-06,
+      "loss": 1.8205,
+      "step": 88
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.795121951219512e-06,
+      "loss": 1.7067,
+      "step": 89
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.791869918699187e-06,
+      "loss": 1.7872,
+      "step": 90
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.788617886178862e-06,
+      "loss": 1.6857,
+      "step": 91
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.785365853658538e-06,
+      "loss": 1.5487,
+      "step": 92
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.782113821138213e-06,
+      "loss": 1.5343,
+      "step": 93
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.778861788617886e-06,
+      "loss": 1.5696,
+      "step": 94
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.775609756097562e-06,
+      "loss": 1.7968,
+      "step": 95
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.772357723577237e-06,
+      "loss": 1.5593,
+      "step": 96
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.769105691056912e-06,
+      "loss": 1.795,
+      "step": 97
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.765853658536585e-06,
+      "loss": 1.6514,
+      "step": 98
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.762601626016262e-06,
+      "loss": 1.6499,
+      "step": 99
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.759349593495936e-06,
+      "loss": 1.8725,
+      "step": 100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.8506,
+      "step": 101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.752845528455286e-06,
+      "loss": 1.8272,
+      "step": 102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.74959349593496e-06,
+      "loss": 1.578,
+      "step": 103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.746341463414635e-06,
+      "loss": 1.6211,
+      "step": 104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.74308943089431e-06,
+      "loss": 1.607,
+      "step": 105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.739837398373985e-06,
+      "loss": 1.7362,
+      "step": 106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.736585365853659e-06,
+      "loss": 1.656,
+      "step": 107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 1.5239,
+      "step": 108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.730081300813009e-06,
+      "loss": 1.9875,
+      "step": 109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.726829268292684e-06,
+      "loss": 1.4154,
+      "step": 110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.72357723577236e-06,
+      "loss": 1.6282,
+      "step": 111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.720325203252033e-06,
+      "loss": 1.4583,
+      "step": 112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.717073170731708e-06,
+      "loss": 1.6865,
+      "step": 113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.713821138211383e-06,
+      "loss": 1.7281,
+      "step": 114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.710569105691058e-06,
+      "loss": 1.7251,
+      "step": 115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.707317073170732e-06,
+      "loss": 1.6696,
+      "step": 116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.704065040650407e-06,
+      "loss": 1.763,
+      "step": 117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.700813008130082e-06,
+      "loss": 1.8399,
+      "step": 118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.697560975609756e-06,
+      "loss": 1.3993,
+      "step": 119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.694308943089433e-06,
+      "loss": 1.5465,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.691056910569106e-06,
+      "loss": 1.6184,
+      "step": 121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.687804878048781e-06,
+      "loss": 1.6158,
+      "step": 122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.684552845528457e-06,
+      "loss": 1.7232,
+      "step": 123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.681300813008132e-06,
+      "loss": 1.6743,
+      "step": 124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.678048780487805e-06,
+      "loss": 1.8448,
+      "step": 125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.67479674796748e-06,
+      "loss": 1.6451,
+      "step": 126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.671544715447156e-06,
+      "loss": 1.659,
+      "step": 127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.668292682926829e-06,
+      "loss": 1.4637,
+      "step": 128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.665040650406506e-06,
+      "loss": 1.386,
+      "step": 129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.66178861788618e-06,
+      "loss": 1.7403,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.658536585365855e-06,
+      "loss": 1.685,
+      "step": 131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.65528455284553e-06,
+      "loss": 1.7384,
+      "step": 132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.652032520325203e-06,
+      "loss": 1.6809,
+      "step": 133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.648780487804879e-06,
+      "loss": 1.736,
+      "step": 134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.645528455284554e-06,
+      "loss": 1.5045,
+      "step": 135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.642276422764229e-06,
+      "loss": 1.6128,
+      "step": 136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.639024390243902e-06,
+      "loss": 1.6891,
+      "step": 137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.635772357723578e-06,
+      "loss": 1.6903,
+      "step": 138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.632520325203253e-06,
+      "loss": 1.689,
+      "step": 139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.629268292682928e-06,
+      "loss": 1.6451,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.626016260162603e-06,
+      "loss": 1.607,
+      "step": 141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.622764227642277e-06,
+      "loss": 1.7186,
+      "step": 142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.619512195121952e-06,
+      "loss": 1.6419,
+      "step": 143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.616260162601627e-06,
+      "loss": 1.7664,
+      "step": 144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.613008130081302e-06,
+      "loss": 1.7533,
+      "step": 145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.609756097560976e-06,
+      "loss": 1.488,
+      "step": 146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.606504065040651e-06,
+      "loss": 1.7294,
+      "step": 147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.603252032520326e-06,
+      "loss": 1.4712,
+      "step": 148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.9081,
+      "step": 149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.596747967479676e-06,
+      "loss": 1.4043,
+      "step": 150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.59349593495935e-06,
+      "loss": 1.4853,
+      "step": 151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.590243902439025e-06,
+      "loss": 1.4879,
+      "step": 152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.5869918699187e-06,
+      "loss": 1.5951,
+      "step": 153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.583739837398375e-06,
+      "loss": 1.5946,
+      "step": 154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.580487804878049e-06,
+      "loss": 1.2488,
+      "step": 155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.577235772357724e-06,
+      "loss": 1.5455,
+      "step": 156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.5739837398374e-06,
+      "loss": 1.463,
+      "step": 157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.570731707317073e-06,
+      "loss": 1.3317,
+      "step": 158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56747967479675e-06,
+      "loss": 1.7321,
+      "step": 159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.564227642276423e-06,
+      "loss": 1.6294,
+      "step": 160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.560975609756098e-06,
+      "loss": 1.571,
+      "step": 161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.557723577235774e-06,
+      "loss": 1.79,
+      "step": 162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.554471544715447e-06,
+      "loss": 1.4929,
+      "step": 163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551219512195122e-06,
+      "loss": 1.6054,
+      "step": 164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547967479674797e-06,
+      "loss": 1.4543,
+      "step": 165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544715447154473e-06,
+      "loss": 1.7564,
+      "step": 166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.541463414634146e-06,
+      "loss": 1.827,
+      "step": 167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.538211382113823e-06,
+      "loss": 1.5357,
+      "step": 168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.534959349593497e-06,
+      "loss": 1.3583,
+      "step": 169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.531707317073172e-06,
+      "loss": 1.4905,
+      "step": 170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.528455284552847e-06,
+      "loss": 1.3308,
+      "step": 171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.52520325203252e-06,
+      "loss": 1.749,
+      "step": 172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.521951219512196e-06,
+      "loss": 1.6352,
+      "step": 173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.51869918699187e-06,
+      "loss": 1.5218,
+      "step": 174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.515447154471546e-06,
+      "loss": 1.7816,
+      "step": 175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 1.7469,
+      "step": 176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.508943089430895e-06,
+      "loss": 1.6206,
+      "step": 177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.50569105691057e-06,
+      "loss": 1.6619,
+      "step": 178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.502439024390245e-06,
+      "loss": 1.6729,
+      "step": 179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.49918699186992e-06,
+      "loss": 1.5492,
+      "step": 180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.495934959349594e-06,
+      "loss": 1.5338,
+      "step": 181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.492682926829269e-06,
+      "loss": 1.6483,
+      "step": 182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.489430894308944e-06,
+      "loss": 1.5496,
+      "step": 183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.48617886178862e-06,
+      "loss": 1.5383,
+      "step": 184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.482926829268293e-06,
+      "loss": 1.5945,
+      "step": 185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.479674796747968e-06,
+      "loss": 1.5497,
+      "step": 186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.476422764227643e-06,
+      "loss": 1.5349,
+      "step": 187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.473170731707317e-06,
+      "loss": 1.781,
+      "step": 188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.469918699186993e-06,
+      "loss": 1.5969,
+      "step": 189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.466666666666667e-06,
+      "loss": 1.588,
+      "step": 190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.463414634146342e-06,
+      "loss": 1.4773,
+      "step": 191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.460162601626017e-06,
+      "loss": 1.6899,
+      "step": 192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.456910569105693e-06,
+      "loss": 1.6527,
+      "step": 193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.453658536585366e-06,
+      "loss": 1.5902,
+      "step": 194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.450406504065041e-06,
+      "loss": 1.2522,
+      "step": 195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.447154471544716e-06,
+      "loss": 1.7773,
+      "step": 196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.44390243902439e-06,
+      "loss": 1.5562,
+      "step": 197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.440650406504067e-06,
+      "loss": 1.4486,
+      "step": 198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.43739837398374e-06,
+      "loss": 1.7037,
+      "step": 199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.434146341463415e-06,
+      "loss": 1.8581,
+      "step": 200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 1.743,
+      "step": 201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.427642276422764e-06,
+      "loss": 1.516,
+      "step": 202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.42439024390244e-06,
+      "loss": 1.6526,
+      "step": 203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.421138211382114e-06,
+      "loss": 1.6481,
+      "step": 204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.41788617886179e-06,
+      "loss": 1.8172,
+      "step": 205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.414634146341463e-06,
+      "loss": 1.6312,
+      "step": 206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.41138211382114e-06,
+      "loss": 1.7242,
+      "step": 207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.408130081300814e-06,
+      "loss": 1.6819,
+      "step": 208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.404878048780489e-06,
+      "loss": 1.6015,
+      "step": 209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.401626016260164e-06,
+      "loss": 1.695,
+      "step": 210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.398373983739837e-06,
+      "loss": 1.6964,
+      "step": 211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.395121951219513e-06,
+      "loss": 1.2685,
+      "step": 212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.391869918699188e-06,
+      "loss": 1.8155,
+      "step": 213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.388617886178863e-06,
+      "loss": 1.5999,
+      "step": 214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.385365853658536e-06,
+      "loss": 1.8224,
+      "step": 215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.382113821138212e-06,
+      "loss": 1.4813,
+      "step": 216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.378861788617887e-06,
+      "loss": 1.2376,
+      "step": 217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.375609756097562e-06,
+      "loss": 1.7143,
+      "step": 218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.372357723577237e-06,
+      "loss": 1.73,
+      "step": 219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.36910569105691e-06,
+      "loss": 1.7165,
+      "step": 220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.365853658536586e-06,
+      "loss": 1.7497,
+      "step": 221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.362601626016261e-06,
+      "loss": 1.5315,
+      "step": 222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.359349593495936e-06,
+      "loss": 1.5823,
+      "step": 223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.356097560975611e-06,
+      "loss": 1.6749,
+      "step": 224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.352845528455285e-06,
+      "loss": 1.724,
+      "step": 225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.34959349593496e-06,
+      "loss": 1.6257,
+      "step": 226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.346341463414635e-06,
+      "loss": 1.766,
+      "step": 227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.34308943089431e-06,
+      "loss": 1.7834,
+      "step": 228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.339837398373984e-06,
+      "loss": 1.5571,
+      "step": 229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.336585365853659e-06,
+      "loss": 1.5665,
+      "step": 230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.8923,
+      "step": 231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.33008130081301e-06,
+      "loss": 1.8982,
+      "step": 232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.326829268292685e-06,
+      "loss": 1.6174,
+      "step": 233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.323577235772358e-06,
+      "loss": 1.6751,
+      "step": 234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.320325203252033e-06,
+      "loss": 1.412,
+      "step": 235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.317073170731709e-06,
+      "loss": 1.567,
+      "step": 236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.313821138211384e-06,
+      "loss": 1.6949,
+      "step": 237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.310569105691057e-06,
+      "loss": 1.5671,
+      "step": 238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.307317073170732e-06,
+      "loss": 1.7493,
+      "step": 239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.304065040650408e-06,
+      "loss": 1.7601,
+      "step": 240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.300813008130081e-06,
+      "loss": 1.6387,
+      "step": 241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.297560975609758e-06,
+      "loss": 1.5297,
+      "step": 242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.294308943089431e-06,
+      "loss": 1.4356,
+      "step": 243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.291056910569107e-06,
+      "loss": 1.7097,
+      "step": 244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.287804878048782e-06,
+      "loss": 1.4814,
+      "step": 245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.284552845528455e-06,
+      "loss": 1.7551,
+      "step": 246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.28130081300813e-06,
+      "loss": 1.685,
+      "step": 247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.278048780487806e-06,
+      "loss": 1.3929,
+      "step": 248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.274796747967481e-06,
+      "loss": 1.662,
+      "step": 249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.271544715447154e-06,
+      "loss": 1.8795,
+      "step": 250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.5654,
+      "step": 251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265040650406505e-06,
+      "loss": 1.8497,
+      "step": 252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.26178861788618e-06,
+      "loss": 1.3939,
+      "step": 253
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.258536585365855e-06,
+      "loss": 1.3955,
+      "step": 254
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.255284552845529e-06,
+      "loss": 1.534,
+      "step": 255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.252032520325204e-06,
+      "loss": 1.8848,
+      "step": 256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.248780487804879e-06,
+      "loss": 1.7198,
+      "step": 257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.245528455284554e-06,
+      "loss": 1.5618,
+      "step": 258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.242276422764228e-06,
+      "loss": 1.4964,
+      "step": 259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.239024390243903e-06,
+      "loss": 1.792,
+      "step": 260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.235772357723578e-06,
+      "loss": 1.789,
+      "step": 261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.232520325203253e-06,
+      "loss": 1.6814,
+      "step": 262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.229268292682928e-06,
+      "loss": 1.5439,
+      "step": 263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.226016260162602e-06,
+      "loss": 1.6299,
+      "step": 264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.222764227642277e-06,
+      "loss": 1.5562,
+      "step": 265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.219512195121952e-06,
+      "loss": 1.5332,
+      "step": 266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.216260162601627e-06,
+      "loss": 1.7804,
+      "step": 267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.213008130081301e-06,
+      "loss": 1.5276,
+      "step": 268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209756097560976e-06,
+      "loss": 1.6706,
+      "step": 269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.206504065040651e-06,
+      "loss": 1.6385,
+      "step": 270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.203252032520325e-06,
+      "loss": 1.4157,
+      "step": 271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 1.5792,
+      "step": 272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.196747967479675e-06,
+      "loss": 1.6301,
+      "step": 273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.19349593495935e-06,
+      "loss": 1.7528,
+      "step": 274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.190243902439026e-06,
+      "loss": 1.6925,
+      "step": 275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.1869918699187e-06,
+      "loss": 1.7756,
+      "step": 276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.183739837398374e-06,
+      "loss": 1.787,
+      "step": 277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.18048780487805e-06,
+      "loss": 1.5087,
+      "step": 278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.177235772357725e-06,
+      "loss": 1.6119,
+      "step": 279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.173983739837398e-06,
+      "loss": 1.3762,
+      "step": 280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.170731707317075e-06,
+      "loss": 1.8892,
+      "step": 281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.167479674796748e-06,
+      "loss": 1.6768,
+      "step": 282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.164227642276424e-06,
+      "loss": 1.6514,
+      "step": 283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.160975609756099e-06,
+      "loss": 1.6304,
+      "step": 284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.157723577235772e-06,
+      "loss": 1.414,
+      "step": 285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.154471544715448e-06,
+      "loss": 1.5245,
+      "step": 286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.151219512195123e-06,
+      "loss": 1.5003,
+      "step": 287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.147967479674798e-06,
+      "loss": 1.6958,
+      "step": 288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.144715447154471e-06,
+      "loss": 1.7058,
+      "step": 289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.141463414634148e-06,
+      "loss": 1.7121,
+      "step": 290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.138211382113822e-06,
+      "loss": 1.7719,
+      "step": 291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.134959349593497e-06,
+      "loss": 1.5376,
+      "step": 292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.131707317073172e-06,
+      "loss": 1.7174,
+      "step": 293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.128455284552846e-06,
+      "loss": 1.6176,
+      "step": 294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.12520325203252e-06,
+      "loss": 1.6354,
+      "step": 295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.121951219512196e-06,
+      "loss": 1.7872,
+      "step": 296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.118699186991871e-06,
+      "loss": 1.6343,
+      "step": 297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.115447154471545e-06,
+      "loss": 1.7884,
+      "step": 298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.11219512195122e-06,
+      "loss": 1.683,
+      "step": 299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.108943089430895e-06,
+      "loss": 1.6817,
+      "step": 300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 1.5559,
+      "step": 301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.102439024390245e-06,
+      "loss": 1.6468,
+      "step": 302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.099186991869919e-06,
+      "loss": 1.798,
+      "step": 303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.095934959349594e-06,
+      "loss": 1.5261,
+      "step": 304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.09268292682927e-06,
+      "loss": 1.7569,
+      "step": 305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.089430894308944e-06,
+      "loss": 1.6516,
+      "step": 306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.086178861788618e-06,
+      "loss": 1.5875,
+      "step": 307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.082926829268293e-06,
+      "loss": 1.5958,
+      "step": 308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.079674796747968e-06,
+      "loss": 1.3016,
+      "step": 309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.076422764227642e-06,
+      "loss": 1.5502,
+      "step": 310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.073170731707319e-06,
+      "loss": 1.7509,
+      "step": 311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.069918699186992e-06,
+      "loss": 1.7824,
+      "step": 312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.6807,
+      "step": 313
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.063414634146343e-06,
+      "loss": 1.5343,
+      "step": 314
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.060162601626016e-06,
+      "loss": 1.4126,
+      "step": 315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.056910569105691e-06,
+      "loss": 1.4971,
+      "step": 316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.053658536585366e-06,
+      "loss": 1.7488,
+      "step": 317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.050406504065042e-06,
+      "loss": 1.2876,
+      "step": 318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.047154471544715e-06,
+      "loss": 1.6428,
+      "step": 319
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.043902439024392e-06,
+      "loss": 1.5369,
+      "step": 320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.040650406504065e-06,
+      "loss": 1.6086,
+      "step": 321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.03739837398374e-06,
+      "loss": 1.7492,
+      "step": 322
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.034146341463416e-06,
+      "loss": 1.5592,
+      "step": 323
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.03089430894309e-06,
+      "loss": 1.4994,
+      "step": 324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.027642276422765e-06,
+      "loss": 1.3149,
+      "step": 325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 1.5925,
+      "step": 326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.021138211382115e-06,
+      "loss": 1.5064,
+      "step": 327
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.017886178861788e-06,
+      "loss": 1.4235,
+      "step": 328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.014634146341464e-06,
+      "loss": 1.6144,
+      "step": 329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.011382113821139e-06,
+      "loss": 1.3013,
+      "step": 330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.008130081300814e-06,
+      "loss": 1.1626,
+      "step": 331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.004878048780489e-06,
+      "loss": 1.6555,
+      "step": 332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.001626016260163e-06,
+      "loss": 1.3684,
+      "step": 333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.998373983739838e-06,
+      "loss": 1.4345,
+      "step": 334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.995121951219513e-06,
+      "loss": 1.676,
+      "step": 335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.991869918699188e-06,
+      "loss": 1.4694,
+      "step": 336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.988617886178862e-06,
+      "loss": 1.3038,
+      "step": 337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.985365853658537e-06,
+      "loss": 1.7949,
+      "step": 338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.982113821138212e-06,
+      "loss": 1.3557,
+      "step": 339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.978861788617886e-06,
+      "loss": 1.329,
+      "step": 340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.975609756097562e-06,
+      "loss": 1.6158,
+      "step": 341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.972357723577236e-06,
+      "loss": 1.4086,
+      "step": 342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.969105691056911e-06,
+      "loss": 1.5873,
+      "step": 343
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.965853658536586e-06,
+      "loss": 1.4768,
+      "step": 344
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.962601626016261e-06,
+      "loss": 1.5341,
+      "step": 345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.959349593495935e-06,
+      "loss": 1.4283,
+      "step": 346
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.95609756097561e-06,
+      "loss": 1.5595,
+      "step": 347
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.952845528455285e-06,
+      "loss": 1.4362,
+      "step": 348
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.94959349593496e-06,
+      "loss": 1.5893,
+      "step": 349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.946341463414636e-06,
+      "loss": 1.5102,
+      "step": 350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 1.5763,
+      "step": 351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.939837398373984e-06,
+      "loss": 1.5823,
+      "step": 352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.93658536585366e-06,
+      "loss": 1.4033,
+      "step": 353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.933333333333333e-06,
+      "loss": 1.5851,
+      "step": 354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.93008130081301e-06,
+      "loss": 1.7889,
+      "step": 355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.926829268292683e-06,
+      "loss": 1.269,
+      "step": 356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.923577235772359e-06,
+      "loss": 1.3806,
+      "step": 357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.920325203252034e-06,
+      "loss": 1.4277,
+      "step": 358
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.917073170731709e-06,
+      "loss": 1.4258,
+      "step": 359
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.913821138211382e-06,
+      "loss": 1.6508,
+      "step": 360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.910569105691058e-06,
+      "loss": 1.344,
+      "step": 361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.907317073170733e-06,
+      "loss": 1.7584,
+      "step": 362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.904065040650406e-06,
+      "loss": 1.6729,
+      "step": 363
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.900813008130083e-06,
+      "loss": 1.432,
+      "step": 364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.897560975609757e-06,
+      "loss": 1.3929,
+      "step": 365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.894308943089432e-06,
+      "loss": 1.5528,
+      "step": 366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.891056910569107e-06,
+      "loss": 1.5668,
+      "step": 367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.88780487804878e-06,
+      "loss": 1.461,
+      "step": 368
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.884552845528456e-06,
+      "loss": 1.4908,
+      "step": 369
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.881300813008131e-06,
+      "loss": 1.4185,
+      "step": 370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.878048780487806e-06,
+      "loss": 1.3597,
+      "step": 371
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.87479674796748e-06,
+      "loss": 1.3553,
+      "step": 372
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.871544715447157e-06,
+      "loss": 1.5445,
+      "step": 373
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.86829268292683e-06,
+      "loss": 1.3625,
+      "step": 374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.865040650406505e-06,
+      "loss": 1.3585,
+      "step": 375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.86178861788618e-06,
+      "loss": 1.6084,
+      "step": 376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.858536585365854e-06,
+      "loss": 1.6486,
+      "step": 377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.855284552845529e-06,
+      "loss": 1.6923,
+      "step": 378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.852032520325204e-06,
+      "loss": 1.5218,
+      "step": 379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.84878048780488e-06,
+      "loss": 1.735,
+      "step": 380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.845528455284553e-06,
+      "loss": 1.4707,
+      "step": 381
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.842276422764228e-06,
+      "loss": 1.5065,
+      "step": 382
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.839024390243903e-06,
+      "loss": 1.4732,
+      "step": 383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.835772357723578e-06,
+      "loss": 1.5967,
+      "step": 384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.832520325203254e-06,
+      "loss": 1.4879,
+      "step": 385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.829268292682927e-06,
+      "loss": 1.5413,
+      "step": 386
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.826016260162602e-06,
+      "loss": 1.4199,
+      "step": 387
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.822764227642278e-06,
+      "loss": 1.5369,
+      "step": 388
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.819512195121953e-06,
+      "loss": 1.3381,
+      "step": 389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.816260162601626e-06,
+      "loss": 1.3409,
+      "step": 390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.813008130081301e-06,
+      "loss": 1.2085,
+      "step": 391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.809756097560977e-06,
+      "loss": 1.5043,
+      "step": 392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.80650406504065e-06,
+      "loss": 1.5956,
+      "step": 393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.803252032520327e-06,
+      "loss": 1.4618,
+      "step": 394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.8e-06,
+      "loss": 1.5296,
+      "step": 395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.796747967479676e-06,
+      "loss": 1.5756,
+      "step": 396
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.79349593495935e-06,
+      "loss": 1.6844,
+      "step": 397
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.790243902439024e-06,
+      "loss": 1.3689,
+      "step": 398
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.7869918699187e-06,
+      "loss": 1.4538,
+      "step": 399
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.783739837398375e-06,
+      "loss": 1.7367,
+      "step": 400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.53,
+      "step": 401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.777235772357723e-06,
+      "loss": 1.502,
+      "step": 402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.7739837398374e-06,
+      "loss": 1.4,
+      "step": 403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.770731707317074e-06,
+      "loss": 1.4918,
+      "step": 404
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.767479674796749e-06,
+      "loss": 1.6176,
+      "step": 405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.764227642276424e-06,
+      "loss": 1.5716,
+      "step": 406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.760975609756098e-06,
+      "loss": 1.2413,
+      "step": 407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.757723577235773e-06,
+      "loss": 1.4993,
+      "step": 408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.754471544715448e-06,
+      "loss": 1.555,
+      "step": 409
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.751219512195123e-06,
+      "loss": 1.3041,
+      "step": 410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.747967479674797e-06,
+      "loss": 1.483,
+      "step": 411
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.744715447154472e-06,
+      "loss": 1.4658,
+      "step": 412
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.741463414634147e-06,
+      "loss": 1.5955,
+      "step": 413
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.738211382113822e-06,
+      "loss": 1.6142,
+      "step": 414
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.734959349593497e-06,
+      "loss": 1.5653,
+      "step": 415
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.731707317073171e-06,
+      "loss": 1.5375,
+      "step": 416
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.728455284552846e-06,
+      "loss": 1.3436,
+      "step": 417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.725203252032521e-06,
+      "loss": 1.2776,
+      "step": 418
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.721951219512196e-06,
+      "loss": 1.7424,
+      "step": 419
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.71869918699187e-06,
+      "loss": 1.2306,
+      "step": 420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.715447154471545e-06,
+      "loss": 1.1992,
+      "step": 421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.71219512195122e-06,
+      "loss": 1.6503,
+      "step": 422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.708943089430894e-06,
+      "loss": 1.5325,
+      "step": 423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.70569105691057e-06,
+      "loss": 1.3696,
+      "step": 424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.702439024390244e-06,
+      "loss": 1.6183,
+      "step": 425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.69918699186992e-06,
+      "loss": 1.4681,
+      "step": 426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.695934959349595e-06,
+      "loss": 1.6487,
+      "step": 427
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.69268292682927e-06,
+      "loss": 1.3393,
+      "step": 428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.689430894308943e-06,
+      "loss": 1.3022,
+      "step": 429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.686178861788618e-06,
+      "loss": 1.4968,
+      "step": 430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.682926829268294e-06,
+      "loss": 1.5164,
+      "step": 431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.679674796747967e-06,
+      "loss": 1.5876,
+      "step": 432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.676422764227644e-06,
+      "loss": 1.2498,
+      "step": 433
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.673170731707317e-06,
+      "loss": 1.3834,
+      "step": 434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.669918699186993e-06,
+      "loss": 1.5345,
+      "step": 435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.4276,
+      "step": 436
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.663414634146341e-06,
+      "loss": 1.18,
+      "step": 437
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.660162601626017e-06,
+      "loss": 1.3438,
+      "step": 438
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.656910569105692e-06,
+      "loss": 1.5515,
+      "step": 439
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.653658536585367e-06,
+      "loss": 1.5289,
+      "step": 440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.65040650406504e-06,
+      "loss": 1.376,
+      "step": 441
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.647154471544717e-06,
+      "loss": 1.5586,
+      "step": 442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.64390243902439e-06,
+      "loss": 1.4523,
+      "step": 443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.640650406504066e-06,
+      "loss": 1.5361,
+      "step": 444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.637398373983741e-06,
+      "loss": 1.2965,
+      "step": 445
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.634146341463415e-06,
+      "loss": 1.5903,
+      "step": 446
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.63089430894309e-06,
+      "loss": 1.3478,
+      "step": 447
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.627642276422765e-06,
+      "loss": 1.5397,
+      "step": 448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.62439024390244e-06,
+      "loss": 1.4469,
+      "step": 449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.621138211382114e-06,
+      "loss": 1.3215,
+      "step": 450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.617886178861789e-06,
+      "loss": 1.2262,
+      "step": 451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.614634146341464e-06,
+      "loss": 1.4903,
+      "step": 452
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.61138211382114e-06,
+      "loss": 1.2247,
+      "step": 453
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.608130081300814e-06,
+      "loss": 1.3248,
+      "step": 454
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.604878048780488e-06,
+      "loss": 1.6522,
+      "step": 455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.601626016260163e-06,
+      "loss": 1.5639,
+      "step": 456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.598373983739838e-06,
+      "loss": 1.5375,
+      "step": 457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.595121951219513e-06,
+      "loss": 1.3746,
+      "step": 458
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.591869918699187e-06,
+      "loss": 1.4423,
+      "step": 459
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.588617886178862e-06,
+      "loss": 1.7226,
+      "step": 460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.585365853658537e-06,
+      "loss": 1.4439,
+      "step": 461
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.58211382113821e-06,
+      "loss": 1.3816,
+      "step": 462
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.578861788617888e-06,
+      "loss": 1.1823,
+      "step": 463
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.575609756097561e-06,
+      "loss": 1.3571,
+      "step": 464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.572357723577236e-06,
+      "loss": 1.5047,
+      "step": 465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.569105691056912e-06,
+      "loss": 1.441,
+      "step": 466
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.565853658536587e-06,
+      "loss": 1.4761,
+      "step": 467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.56260162601626e-06,
+      "loss": 1.6058,
+      "step": 468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.559349593495935e-06,
+      "loss": 1.5795,
+      "step": 469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.55609756097561e-06,
+      "loss": 1.3098,
+      "step": 470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.552845528455284e-06,
+      "loss": 1.3605,
+      "step": 471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.549593495934961e-06,
+      "loss": 1.3805,
+      "step": 472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.546341463414634e-06,
+      "loss": 1.3127,
+      "step": 473
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.54308943089431e-06,
+      "loss": 1.7115,
+      "step": 474
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.539837398373985e-06,
+      "loss": 1.5059,
+      "step": 475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 1.6556,
+      "step": 476
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.3742,
+      "step": 477
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.530081300813009e-06,
+      "loss": 1.4849,
+      "step": 478
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.526829268292684e-06,
+      "loss": 1.5063,
+      "step": 479
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.523577235772359e-06,
+      "loss": 1.4223,
+      "step": 480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.520325203252033e-06,
+      "loss": 1.558,
+      "step": 481
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.517073170731708e-06,
+      "loss": 1.6228,
+      "step": 482
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.513821138211383e-06,
+      "loss": 1.6235,
+      "step": 483
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.510569105691058e-06,
+      "loss": 1.6669,
+      "step": 484
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.507317073170732e-06,
+      "loss": 1.4977,
+      "step": 485
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.504065040650408e-06,
+      "loss": 1.4615,
+      "step": 486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.500813008130082e-06,
+      "loss": 1.5797,
+      "step": 487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.497560975609757e-06,
+      "loss": 1.6213,
+      "step": 488
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.494308943089432e-06,
+      "loss": 1.1375,
+      "step": 489
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.491056910569106e-06,
+      "loss": 1.4343,
+      "step": 490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.487804878048781e-06,
+      "loss": 1.2965,
+      "step": 491
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.484552845528456e-06,
+      "loss": 1.5125,
+      "step": 492
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.481300813008131e-06,
+      "loss": 1.387,
+      "step": 493
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.478048780487805e-06,
+      "loss": 1.4299,
+      "step": 494
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.47479674796748e-06,
+      "loss": 1.3791,
+      "step": 495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.471544715447155e-06,
+      "loss": 1.5965,
+      "step": 496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.46829268292683e-06,
+      "loss": 1.1611,
+      "step": 497
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.465040650406506e-06,
+      "loss": 1.4392,
+      "step": 498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.461788617886179e-06,
+      "loss": 1.2923,
+      "step": 499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.458536585365854e-06,
+      "loss": 1.4154,
+      "step": 500
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.6067962646484375,
+      "eval_runtime": 141.8581,
+      "eval_samples_per_second": 4.272,
+      "eval_steps_per_second": 4.272,
+      "step": 500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 1.6002,
+      "step": 501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.452032520325205e-06,
+      "loss": 1.2813,
+      "step": 502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.448780487804878e-06,
+      "loss": 1.5729,
+      "step": 503
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.445528455284553e-06,
+      "loss": 1.3391,
+      "step": 504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.442276422764229e-06,
+      "loss": 1.7376,
+      "step": 505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.439024390243902e-06,
+      "loss": 1.4396,
+      "step": 506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.435772357723579e-06,
+      "loss": 1.4824,
+      "step": 507
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.432520325203252e-06,
+      "loss": 1.32,
+      "step": 508
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.429268292682928e-06,
+      "loss": 1.4105,
+      "step": 509
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.426016260162603e-06,
+      "loss": 1.6573,
+      "step": 510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.422764227642278e-06,
+      "loss": 1.3721,
+      "step": 511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.419512195121951e-06,
+      "loss": 1.4455,
+      "step": 512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.416260162601627e-06,
+      "loss": 1.422,
+      "step": 513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.413008130081302e-06,
+      "loss": 1.358,
+      "step": 514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.409756097560975e-06,
+      "loss": 1.4902,
+      "step": 515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.406504065040652e-06,
+      "loss": 1.323,
+      "step": 516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.403252032520326e-06,
+      "loss": 1.4881,
+      "step": 517
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.6092,
+      "step": 518
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.396747967479676e-06,
+      "loss": 1.1993,
+      "step": 519
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.39349593495935e-06,
+      "loss": 1.3538,
+      "step": 520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.390243902439025e-06,
+      "loss": 1.7321,
+      "step": 521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.3869918699187e-06,
+      "loss": 1.5392,
+      "step": 522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.383739837398375e-06,
+      "loss": 1.5762,
+      "step": 523
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.380487804878049e-06,
+      "loss": 1.4268,
+      "step": 524
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.377235772357725e-06,
+      "loss": 1.5569,
+      "step": 525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.373983739837399e-06,
+      "loss": 1.4795,
+      "step": 526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370731707317074e-06,
+      "loss": 1.4702,
+      "step": 527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36747967479675e-06,
+      "loss": 1.7304,
+      "step": 528
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.364227642276423e-06,
+      "loss": 1.3331,
+      "step": 529
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.360975609756098e-06,
+      "loss": 1.6886,
+      "step": 530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.357723577235773e-06,
+      "loss": 1.569,
+      "step": 531
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.354471544715448e-06,
+      "loss": 1.2473,
+      "step": 532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.351219512195122e-06,
+      "loss": 1.6269,
+      "step": 533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.347967479674797e-06,
+      "loss": 1.3735,
+      "step": 534
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.344715447154472e-06,
+      "loss": 1.5432,
+      "step": 535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.341463414634147e-06,
+      "loss": 1.3322,
+      "step": 536
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.338211382113823e-06,
+      "loss": 1.6241,
+      "step": 537
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.334959349593496e-06,
+      "loss": 1.4373,
+      "step": 538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.331707317073171e-06,
+      "loss": 1.2256,
+      "step": 539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.328455284552847e-06,
+      "loss": 1.1923,
+      "step": 540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.325203252032522e-06,
+      "loss": 1.5781,
+      "step": 541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.321951219512195e-06,
+      "loss": 1.1594,
+      "step": 542
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.31869918699187e-06,
+      "loss": 1.5167,
+      "step": 543
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.315447154471546e-06,
+      "loss": 1.4168,
+      "step": 544
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.312195121951219e-06,
+      "loss": 1.4213,
+      "step": 545
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.308943089430896e-06,
+      "loss": 1.4967,
+      "step": 546
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.30569105691057e-06,
+      "loss": 1.4551,
+      "step": 547
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.302439024390245e-06,
+      "loss": 1.7946,
+      "step": 548
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.29918699186992e-06,
+      "loss": 1.4268,
+      "step": 549
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.295934959349595e-06,
+      "loss": 1.5279,
+      "step": 550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.5608,
+      "step": 551
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.289430894308944e-06,
+      "loss": 1.2482,
+      "step": 552
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.286178861788619e-06,
+      "loss": 1.3781,
+      "step": 553
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.282926829268292e-06,
+      "loss": 1.3366,
+      "step": 554
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.27967479674797e-06,
+      "loss": 1.4554,
+      "step": 555
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.276422764227643e-06,
+      "loss": 1.1645,
+      "step": 556
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.273170731707318e-06,
+      "loss": 1.5263,
+      "step": 557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.269918699186993e-06,
+      "loss": 1.6176,
+      "step": 558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.5259,
+      "step": 559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.263414634146342e-06,
+      "loss": 1.6063,
+      "step": 560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.260162601626017e-06,
+      "loss": 1.5718,
+      "step": 561
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.256910569105692e-06,
+      "loss": 1.585,
+      "step": 562
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.253658536585366e-06,
+      "loss": 1.204,
+      "step": 563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.25040650406504e-06,
+      "loss": 1.5214,
+      "step": 564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.247154471544716e-06,
+      "loss": 1.256,
+      "step": 565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.243902439024391e-06,
+      "loss": 1.5181,
+      "step": 566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.240650406504066e-06,
+      "loss": 1.4303,
+      "step": 567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.23739837398374e-06,
+      "loss": 1.6173,
+      "step": 568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.234146341463415e-06,
+      "loss": 1.5245,
+      "step": 569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.23089430894309e-06,
+      "loss": 1.4251,
+      "step": 570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.227642276422765e-06,
+      "loss": 1.6013,
+      "step": 571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.224390243902439e-06,
+      "loss": 1.4144,
+      "step": 572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.221138211382114e-06,
+      "loss": 1.4822,
+      "step": 573
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.21788617886179e-06,
+      "loss": 1.2498,
+      "step": 574
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.214634146341463e-06,
+      "loss": 1.4035,
+      "step": 575
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 8.21138211382114e-06,
+      "loss": 1.4348,
+      "step": 576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.208130081300813e-06,
+      "loss": 1.2738,
+      "step": 577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.204878048780488e-06,
+      "loss": 1.3894,
+      "step": 578
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 8.201626016260164e-06,
+      "loss": 1.3909,
+      "step": 579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.198373983739839e-06,
+      "loss": 1.4416,
+      "step": 580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.195121951219512e-06,
+      "loss": 1.4567,
+      "step": 581
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 8.191869918699187e-06,
+      "loss": 1.5784,
+      "step": 582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.188617886178863e-06,
+      "loss": 1.4794,
+      "step": 583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.185365853658536e-06,
+      "loss": 1.5093,
+      "step": 584
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.182113821138213e-06,
+      "loss": 1.4111,
+      "step": 585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.178861788617886e-06,
+      "loss": 1.6598,
+      "step": 586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.175609756097562e-06,
+      "loss": 1.344,
+      "step": 587
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.172357723577237e-06,
+      "loss": 1.3105,
+      "step": 588
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.16910569105691e-06,
+      "loss": 1.6021,
+      "step": 589
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.165853658536585e-06,
+      "loss": 1.469,
+      "step": 590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 8.16260162601626e-06,
+      "loss": 1.5578,
+      "step": 591
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.159349593495936e-06,
+      "loss": 1.4479,
+      "step": 592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.15609756097561e-06,
+      "loss": 1.5824,
+      "step": 593
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.152845528455286e-06,
+      "loss": 1.2874,
+      "step": 594
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.14959349593496e-06,
+      "loss": 1.6409,
+      "step": 595
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.146341463414635e-06,
+      "loss": 1.4922,
+      "step": 596
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.14308943089431e-06,
+      "loss": 1.5295,
+      "step": 597
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.139837398373984e-06,
+      "loss": 1.4093,
+      "step": 598
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.136585365853659e-06,
+      "loss": 1.6339,
+      "step": 599
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.133333333333334e-06,
+      "loss": 1.4266,
+      "step": 600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 1.5529,
+      "step": 601
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.126829268292683e-06,
+      "loss": 1.3609,
+      "step": 602
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.123577235772358e-06,
+      "loss": 1.6272,
+      "step": 603
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.120325203252033e-06,
+      "loss": 1.3487,
+      "step": 604
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.117073170731708e-06,
+      "loss": 1.4477,
+      "step": 605
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.113821138211383e-06,
+      "loss": 1.5086,
+      "step": 606
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.110569105691057e-06,
+      "loss": 1.3573,
+      "step": 607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.107317073170734e-06,
+      "loss": 1.3851,
+      "step": 608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.104065040650407e-06,
+      "loss": 1.4334,
+      "step": 609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.100813008130082e-06,
+      "loss": 1.3445,
+      "step": 610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.097560975609758e-06,
+      "loss": 1.7443,
+      "step": 611
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.094308943089431e-06,
+      "loss": 1.6259,
+      "step": 612
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.091056910569106e-06,
+      "loss": 1.4475,
+      "step": 613
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.087804878048781e-06,
+      "loss": 1.4172,
+      "step": 614
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.084552845528457e-06,
+      "loss": 1.3567,
+      "step": 615
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.08130081300813e-06,
+      "loss": 1.4151,
+      "step": 616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.078048780487805e-06,
+      "loss": 1.5765,
+      "step": 617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.07479674796748e-06,
+      "loss": 1.6123,
+      "step": 618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.071544715447156e-06,
+      "loss": 1.1772,
+      "step": 619
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.068292682926831e-06,
+      "loss": 1.6215,
+      "step": 620
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.065040650406504e-06,
+      "loss": 1.4718,
+      "step": 621
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.06178861788618e-06,
+      "loss": 1.1363,
+      "step": 622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.058536585365855e-06,
+      "loss": 1.2548,
+      "step": 623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.05528455284553e-06,
+      "loss": 1.3291,
+      "step": 624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.052032520325203e-06,
+      "loss": 1.32,
+      "step": 625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 1.3368,
+      "step": 626
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.045528455284554e-06,
+      "loss": 1.3999,
+      "step": 627
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.042276422764227e-06,
+      "loss": 1.317,
+      "step": 628
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 8.039024390243904e-06,
+      "loss": 1.2394,
+      "step": 629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.035772357723578e-06,
+      "loss": 1.3729,
+      "step": 630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.032520325203253e-06,
+      "loss": 1.2583,
+      "step": 631
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 8.029268292682928e-06,
+      "loss": 1.3256,
+      "step": 632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.026016260162602e-06,
+      "loss": 1.361,
+      "step": 633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.022764227642277e-06,
+      "loss": 1.0694,
+      "step": 634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 8.019512195121952e-06,
+      "loss": 1.4032,
+      "step": 635
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.016260162601627e-06,
+      "loss": 1.2993,
+      "step": 636
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.0130081300813e-06,
+      "loss": 1.4113,
+      "step": 637
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.009756097560977e-06,
+      "loss": 1.3584,
+      "step": 638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.006504065040651e-06,
+      "loss": 1.2335,
+      "step": 639
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.003252032520326e-06,
+      "loss": 1.2672,
+      "step": 640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.5265,
+      "step": 641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.996747967479675e-06,
+      "loss": 1.29,
+      "step": 642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.99349593495935e-06,
+      "loss": 1.2354,
+      "step": 643
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.990243902439025e-06,
+      "loss": 1.4025,
+      "step": 644
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.9869918699187e-06,
+      "loss": 1.2242,
+      "step": 645
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.983739837398374e-06,
+      "loss": 1.4565,
+      "step": 646
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.980487804878049e-06,
+      "loss": 1.4118,
+      "step": 647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.977235772357724e-06,
+      "loss": 1.5411,
+      "step": 648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.9739837398374e-06,
+      "loss": 1.3918,
+      "step": 649
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.970731707317075e-06,
+      "loss": 1.309,
+      "step": 650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.967479674796748e-06,
+      "loss": 1.3004,
+      "step": 651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.964227642276423e-06,
+      "loss": 1.3834,
+      "step": 652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.960975609756098e-06,
+      "loss": 1.1848,
+      "step": 653
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.957723577235774e-06,
+      "loss": 1.3983,
+      "step": 654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.954471544715447e-06,
+      "loss": 1.3415,
+      "step": 655
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.951219512195122e-06,
+      "loss": 1.4793,
+      "step": 656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.947967479674798e-06,
+      "loss": 1.076,
+      "step": 657
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.944715447154471e-06,
+      "loss": 1.1463,
+      "step": 658
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.941463414634148e-06,
+      "loss": 1.3898,
+      "step": 659
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.938211382113821e-06,
+      "loss": 1.2726,
+      "step": 660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.934959349593497e-06,
+      "loss": 1.3014,
+      "step": 661
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.931707317073172e-06,
+      "loss": 1.2351,
+      "step": 662
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.928455284552847e-06,
+      "loss": 1.1233,
+      "step": 663
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.92520325203252e-06,
+      "loss": 1.3581,
+      "step": 664
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.921951219512196e-06,
+      "loss": 1.5989,
+      "step": 665
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.91869918699187e-06,
+      "loss": 1.0798,
+      "step": 666
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.915447154471544e-06,
+      "loss": 1.3955,
+      "step": 667
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.912195121951221e-06,
+      "loss": 1.3645,
+      "step": 668
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.908943089430895e-06,
+      "loss": 1.2955,
+      "step": 669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.90569105691057e-06,
+      "loss": 1.5259,
+      "step": 670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.902439024390245e-06,
+      "loss": 1.1694,
+      "step": 671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.899186991869919e-06,
+      "loss": 1.3757,
+      "step": 672
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.895934959349594e-06,
+      "loss": 1.2343,
+      "step": 673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.892682926829269e-06,
+      "loss": 1.4036,
+      "step": 674
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.889430894308944e-06,
+      "loss": 1.4886,
+      "step": 675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.886178861788618e-06,
+      "loss": 1.2545,
+      "step": 676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.882926829268294e-06,
+      "loss": 1.3849,
+      "step": 677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.879674796747968e-06,
+      "loss": 1.449,
+      "step": 678
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.876422764227643e-06,
+      "loss": 1.37,
+      "step": 679
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.873170731707318e-06,
+      "loss": 1.1809,
+      "step": 680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.869918699186992e-06,
+      "loss": 1.1049,
+      "step": 681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.3176,
+      "step": 682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.863414634146342e-06,
+      "loss": 1.2981,
+      "step": 683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.860162601626017e-06,
+      "loss": 1.1863,
+      "step": 684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.856910569105691e-06,
+      "loss": 1.4437,
+      "step": 685
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.853658536585366e-06,
+      "loss": 1.2973,
+      "step": 686
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.850406504065041e-06,
+      "loss": 1.2121,
+      "step": 687
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.847154471544716e-06,
+      "loss": 1.2568,
+      "step": 688
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.843902439024392e-06,
+      "loss": 1.221,
+      "step": 689
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.840650406504065e-06,
+      "loss": 1.4176,
+      "step": 690
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.83739837398374e-06,
+      "loss": 1.1732,
+      "step": 691
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.834146341463415e-06,
+      "loss": 1.1603,
+      "step": 692
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.83089430894309e-06,
+      "loss": 1.2562,
+      "step": 693
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.827642276422764e-06,
+      "loss": 1.1951,
+      "step": 694
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.82439024390244e-06,
+      "loss": 1.3775,
+      "step": 695
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.821138211382115e-06,
+      "loss": 1.2736,
+      "step": 696
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.817886178861788e-06,
+      "loss": 1.3632,
+      "step": 697
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.814634146341465e-06,
+      "loss": 1.3347,
+      "step": 698
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.811382113821138e-06,
+      "loss": 1.2729,
+      "step": 699
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.808130081300814e-06,
+      "loss": 1.4577,
+      "step": 700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.5629,
+      "step": 701
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.801626016260164e-06,
+      "loss": 1.1883,
+      "step": 702
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.798373983739837e-06,
+      "loss": 1.2007,
+      "step": 703
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.795121951219513e-06,
+      "loss": 1.0999,
+      "step": 704
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.791869918699188e-06,
+      "loss": 1.3069,
+      "step": 705
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.788617886178861e-06,
+      "loss": 1.437,
+      "step": 706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.785365853658538e-06,
+      "loss": 1.3514,
+      "step": 707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.782113821138212e-06,
+      "loss": 1.1724,
+      "step": 708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.778861788617887e-06,
+      "loss": 1.202,
+      "step": 709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.775609756097562e-06,
+      "loss": 1.5265,
+      "step": 710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.772357723577236e-06,
+      "loss": 1.5743,
+      "step": 711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.76910569105691e-06,
+      "loss": 1.3205,
+      "step": 712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.765853658536586e-06,
+      "loss": 1.3839,
+      "step": 713
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.762601626016261e-06,
+      "loss": 1.3082,
+      "step": 714
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.759349593495935e-06,
+      "loss": 1.2327,
+      "step": 715
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.75609756097561e-06,
+      "loss": 1.2383,
+      "step": 716
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.752845528455285e-06,
+      "loss": 1.3538,
+      "step": 717
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.74959349593496e-06,
+      "loss": 1.2992,
+      "step": 718
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.746341463414635e-06,
+      "loss": 1.3524,
+      "step": 719
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.743089430894309e-06,
+      "loss": 1.4946,
+      "step": 720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.739837398373984e-06,
+      "loss": 1.4023,
+      "step": 721
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.73658536585366e-06,
+      "loss": 1.474,
+      "step": 722
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.1706,
+      "step": 723
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.730081300813008e-06,
+      "loss": 1.3084,
+      "step": 724
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.726829268292683e-06,
+      "loss": 1.4063,
+      "step": 725
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.723577235772358e-06,
+      "loss": 1.3535,
+      "step": 726
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.720325203252033e-06,
+      "loss": 1.0598,
+      "step": 727
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.717073170731709e-06,
+      "loss": 1.3855,
+      "step": 728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.713821138211382e-06,
+      "loss": 1.3196,
+      "step": 729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.710569105691057e-06,
+      "loss": 1.4757,
+      "step": 730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.707317073170732e-06,
+      "loss": 1.171,
+      "step": 731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.704065040650408e-06,
+      "loss": 1.5208,
+      "step": 732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.700813008130083e-06,
+      "loss": 1.39,
+      "step": 733
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.697560975609756e-06,
+      "loss": 1.26,
+      "step": 734
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.694308943089432e-06,
+      "loss": 1.4742,
+      "step": 735
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.691056910569107e-06,
+      "loss": 1.1885,
+      "step": 736
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.687804878048782e-06,
+      "loss": 1.3186,
+      "step": 737
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.684552845528455e-06,
+      "loss": 1.2698,
+      "step": 738
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.68130081300813e-06,
+      "loss": 1.2798,
+      "step": 739
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.678048780487806e-06,
+      "loss": 1.311,
+      "step": 740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.67479674796748e-06,
+      "loss": 1.2981,
+      "step": 741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.671544715447156e-06,
+      "loss": 1.4004,
+      "step": 742
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.66829268292683e-06,
+      "loss": 1.4679,
+      "step": 743
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.665040650406505e-06,
+      "loss": 1.324,
+      "step": 744
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.66178861788618e-06,
+      "loss": 1.3271,
+      "step": 745
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.658536585365855e-06,
+      "loss": 1.4546,
+      "step": 746
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.655284552845529e-06,
+      "loss": 1.2878,
+      "step": 747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.652032520325204e-06,
+      "loss": 1.5134,
+      "step": 748
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.648780487804879e-06,
+      "loss": 1.6176,
+      "step": 749
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.645528455284553e-06,
+      "loss": 1.4697,
+      "step": 750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.64227642276423e-06,
+      "loss": 1.3675,
+      "step": 751
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.639024390243903e-06,
+      "loss": 1.4007,
+      "step": 752
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.635772357723578e-06,
+      "loss": 1.2617,
+      "step": 753
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.632520325203253e-06,
+      "loss": 1.2239,
+      "step": 754
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.629268292682928e-06,
+      "loss": 1.2468,
+      "step": 755
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.626016260162602e-06,
+      "loss": 1.3658,
+      "step": 756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.622764227642277e-06,
+      "loss": 1.2836,
+      "step": 757
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.619512195121952e-06,
+      "loss": 1.4195,
+      "step": 758
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.616260162601627e-06,
+      "loss": 1.0597,
+      "step": 759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.613008130081302e-06,
+      "loss": 1.2532,
+      "step": 760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.609756097560976e-06,
+      "loss": 1.2008,
+      "step": 761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.6065040650406506e-06,
+      "loss": 1.3477,
+      "step": 762
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.603252032520327e-06,
+      "loss": 1.4744,
+      "step": 763
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 1.1077,
+      "step": 764
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.596747967479675e-06,
+      "loss": 1.258,
+      "step": 765
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.5934959349593505e-06,
+      "loss": 1.6319,
+      "step": 766
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.590243902439025e-06,
+      "loss": 1.4949,
+      "step": 767
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.586991869918699e-06,
+      "loss": 1.2363,
+      "step": 768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.583739837398375e-06,
+      "loss": 1.4886,
+      "step": 769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.5804878048780495e-06,
+      "loss": 1.2699,
+      "step": 770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.577235772357724e-06,
+      "loss": 1.2417,
+      "step": 771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 7.573983739837399e-06,
+      "loss": 1.4455,
+      "step": 772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.570731707317074e-06,
+      "loss": 1.4472,
+      "step": 773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.5674796747967486e-06,
+      "loss": 1.1961,
+      "step": 774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 7.564227642276424e-06,
+      "loss": 1.492,
+      "step": 775
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 1.6875,
+      "step": 776
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.557723577235772e-06,
+      "loss": 1.139,
+      "step": 777
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 7.5544715447154484e-06,
+      "loss": 1.3865,
+      "step": 778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.551219512195123e-06,
+      "loss": 1.4373,
+      "step": 779
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.547967479674797e-06,
+      "loss": 1.4863,
+      "step": 780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.544715447154472e-06,
+      "loss": 1.4408,
+      "step": 781
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.541463414634147e-06,
+      "loss": 1.2452,
+      "step": 782
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.538211382113822e-06,
+      "loss": 1.1683,
+      "step": 783
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.534959349593497e-06,
+      "loss": 1.1629,
+      "step": 784
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.531707317073171e-06,
+      "loss": 1.4936,
+      "step": 785
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.528455284552846e-06,
+      "loss": 1.3447,
+      "step": 786
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.525203252032522e-06,
+      "loss": 1.3882,
+      "step": 787
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.521951219512196e-06,
+      "loss": 1.5082,
+      "step": 788
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.51869918699187e-06,
+      "loss": 1.4121,
+      "step": 789
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.515447154471546e-06,
+      "loss": 1.2164,
+      "step": 790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.51219512195122e-06,
+      "loss": 1.326,
+      "step": 791
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.508943089430894e-06,
+      "loss": 1.332,
+      "step": 792
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.50569105691057e-06,
+      "loss": 1.485,
+      "step": 793
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.502439024390245e-06,
+      "loss": 1.4878,
+      "step": 794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.499186991869919e-06,
+      "loss": 1.3608,
+      "step": 795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.495934959349594e-06,
+      "loss": 1.3076,
+      "step": 796
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.4926829268292685e-06,
+      "loss": 1.4848,
+      "step": 797
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.489430894308944e-06,
+      "loss": 1.488,
+      "step": 798
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.486178861788619e-06,
+      "loss": 1.3955,
+      "step": 799
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.482926829268293e-06,
+      "loss": 1.3853,
+      "step": 800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 1.3082,
+      "step": 801
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.476422764227644e-06,
+      "loss": 1.4004,
+      "step": 802
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.473170731707318e-06,
+      "loss": 1.1345,
+      "step": 803
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.469918699186992e-06,
+      "loss": 1.4272,
+      "step": 804
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.5545,
+      "step": 805
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.463414634146342e-06,
+      "loss": 1.597,
+      "step": 806
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.460162601626016e-06,
+      "loss": 1.3394,
+      "step": 807
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.456910569105692e-06,
+      "loss": 1.3585,
+      "step": 808
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.4536585365853665e-06,
+      "loss": 1.3431,
+      "step": 809
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.450406504065041e-06,
+      "loss": 1.4112,
+      "step": 810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.447154471544716e-06,
+      "loss": 1.2675,
+      "step": 811
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.443902439024391e-06,
+      "loss": 1.296,
+      "step": 812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.4406504065040656e-06,
+      "loss": 1.3006,
+      "step": 813
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.437398373983741e-06,
+      "loss": 1.3702,
+      "step": 814
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.434146341463415e-06,
+      "loss": 1.3133,
+      "step": 815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.430894308943089e-06,
+      "loss": 1.23,
+      "step": 816
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.4276422764227655e-06,
+      "loss": 1.4371,
+      "step": 817
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.42439024390244e-06,
+      "loss": 1.0529,
+      "step": 818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.421138211382114e-06,
+      "loss": 1.2735,
+      "step": 819
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.417886178861789e-06,
+      "loss": 1.3434,
+      "step": 820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.414634146341464e-06,
+      "loss": 1.1519,
+      "step": 821
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.411382113821138e-06,
+      "loss": 1.0074,
+      "step": 822
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.408130081300814e-06,
+      "loss": 1.3005,
+      "step": 823
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.404878048780488e-06,
+      "loss": 1.4737,
+      "step": 824
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.401626016260163e-06,
+      "loss": 1.4338,
+      "step": 825
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.398373983739838e-06,
+      "loss": 1.175,
+      "step": 826
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.395121951219513e-06,
+      "loss": 1.4578,
+      "step": 827
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.391869918699187e-06,
+      "loss": 1.3808,
+      "step": 828
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.388617886178863e-06,
+      "loss": 1.4883,
+      "step": 829
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.385365853658537e-06,
+      "loss": 1.1138,
+      "step": 830
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.382113821138211e-06,
+      "loss": 1.1484,
+      "step": 831
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.378861788617887e-06,
+      "loss": 1.5129,
+      "step": 832
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.375609756097562e-06,
+      "loss": 1.1886,
+      "step": 833
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.372357723577236e-06,
+      "loss": 1.3183,
+      "step": 834
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.369105691056911e-06,
+      "loss": 1.2466,
+      "step": 835
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.3658536585365855e-06,
+      "loss": 1.4346,
+      "step": 836
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 7.36260162601626e-06,
+      "loss": 1.1375,
+      "step": 837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.359349593495936e-06,
+      "loss": 1.4055,
+      "step": 838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.35609756097561e-06,
+      "loss": 1.146,
+      "step": 839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.3528455284552846e-06,
+      "loss": 1.2631,
+      "step": 840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.349593495934961e-06,
+      "loss": 1.3435,
+      "step": 841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.346341463414635e-06,
+      "loss": 1.574,
+      "step": 842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.343089430894309e-06,
+      "loss": 1.2833,
+      "step": 843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.3398373983739845e-06,
+      "loss": 1.3218,
+      "step": 844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.336585365853659e-06,
+      "loss": 1.2237,
+      "step": 845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.5563,
+      "step": 846
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.330081300813009e-06,
+      "loss": 1.3506,
+      "step": 847
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.3268292682926835e-06,
+      "loss": 1.3707,
+      "step": 848
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.323577235772358e-06,
+      "loss": 1.2903,
+      "step": 849
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.320325203252033e-06,
+      "loss": 1.4918,
+      "step": 850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.4208,
+      "step": 851
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 7.3138211382113826e-06,
+      "loss": 1.3251,
+      "step": 852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.310569105691058e-06,
+      "loss": 1.2379,
+      "step": 853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.307317073170732e-06,
+      "loss": 1.3171,
+      "step": 854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.3040650406504064e-06,
+      "loss": 1.5917,
+      "step": 855
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.3008130081300825e-06,
+      "loss": 1.2904,
+      "step": 856
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.297560975609757e-06,
+      "loss": 1.3906,
+      "step": 857
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.294308943089431e-06,
+      "loss": 1.2194,
+      "step": 858
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.291056910569106e-06,
+      "loss": 1.4132,
+      "step": 859
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.287804878048781e-06,
+      "loss": 1.1204,
+      "step": 860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.284552845528457e-06,
+      "loss": 1.2652,
+      "step": 861
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.281300813008131e-06,
+      "loss": 1.2934,
+      "step": 862
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.278048780487805e-06,
+      "loss": 1.1168,
+      "step": 863
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.2747967479674806e-06,
+      "loss": 1.1844,
+      "step": 864
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.271544715447155e-06,
+      "loss": 1.5539,
+      "step": 865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.268292682926829e-06,
+      "loss": 1.4761,
+      "step": 866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.265040650406505e-06,
+      "loss": 1.4416,
+      "step": 867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.26178861788618e-06,
+      "loss": 1.1777,
+      "step": 868
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.258536585365854e-06,
+      "loss": 1.5521,
+      "step": 869
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.25528455284553e-06,
+      "loss": 1.3197,
+      "step": 870
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.252032520325204e-06,
+      "loss": 1.5776,
+      "step": 871
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.248780487804879e-06,
+      "loss": 1.6328,
+      "step": 872
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.245528455284554e-06,
+      "loss": 1.2583,
+      "step": 873
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.242276422764228e-06,
+      "loss": 1.321,
+      "step": 874
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.2390243902439025e-06,
+      "loss": 1.3356,
+      "step": 875
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.2357723577235786e-06,
+      "loss": 1.3062,
+      "step": 876
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.232520325203253e-06,
+      "loss": 1.3825,
+      "step": 877
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.229268292682927e-06,
+      "loss": 1.5639,
+      "step": 878
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.226016260162602e-06,
+      "loss": 1.1933,
+      "step": 879
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.222764227642277e-06,
+      "loss": 1.4793,
+      "step": 880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.219512195121952e-06,
+      "loss": 1.5361,
+      "step": 881
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.216260162601627e-06,
+      "loss": 1.6394,
+      "step": 882
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.2130081300813015e-06,
+      "loss": 1.3396,
+      "step": 883
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.209756097560976e-06,
+      "loss": 1.4382,
+      "step": 884
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.206504065040652e-06,
+      "loss": 1.2569,
+      "step": 885
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.203252032520326e-06,
+      "loss": 1.5506,
+      "step": 886
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.2029,
+      "step": 887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.196747967479676e-06,
+      "loss": 1.5553,
+      "step": 888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.19349593495935e-06,
+      "loss": 1.2231,
+      "step": 889
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.190243902439024e-06,
+      "loss": 1.2941,
+      "step": 890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.1869918699187e-06,
+      "loss": 1.3191,
+      "step": 891
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.183739837398375e-06,
+      "loss": 1.346,
+      "step": 892
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.180487804878049e-06,
+      "loss": 1.3875,
+      "step": 893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.177235772357724e-06,
+      "loss": 1.1806,
+      "step": 894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.173983739837399e-06,
+      "loss": 1.1736,
+      "step": 895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.170731707317074e-06,
+      "loss": 1.3844,
+      "step": 896
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.167479674796749e-06,
+      "loss": 1.1793,
+      "step": 897
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.164227642276423e-06,
+      "loss": 1.4713,
+      "step": 898
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.160975609756098e-06,
+      "loss": 1.4122,
+      "step": 899
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.157723577235774e-06,
+      "loss": 1.5265,
+      "step": 900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 1.3023,
+      "step": 901
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.151219512195122e-06,
+      "loss": 1.3309,
+      "step": 902
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.1479674796747976e-06,
+      "loss": 1.1222,
+      "step": 903
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.144715447154472e-06,
+      "loss": 1.2537,
+      "step": 904
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.141463414634146e-06,
+      "loss": 1.4281,
+      "step": 905
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.138211382113822e-06,
+      "loss": 1.3172,
+      "step": 906
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.134959349593497e-06,
+      "loss": 1.4982,
+      "step": 907
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.131707317073171e-06,
+      "loss": 1.4308,
+      "step": 908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.128455284552846e-06,
+      "loss": 1.2351,
+      "step": 909
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.125203252032521e-06,
+      "loss": 1.4459,
+      "step": 910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.121951219512196e-06,
+      "loss": 1.4483,
+      "step": 911
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.118699186991871e-06,
+      "loss": 1.4458,
+      "step": 912
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.115447154471545e-06,
+      "loss": 1.4197,
+      "step": 913
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.1121951219512195e-06,
+      "loss": 1.2328,
+      "step": 914
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.1089430894308956e-06,
+      "loss": 1.3312,
+      "step": 915
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.10569105691057e-06,
+      "loss": 1.2779,
+      "step": 916
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.102439024390244e-06,
+      "loss": 1.2505,
+      "step": 917
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.0991869918699194e-06,
+      "loss": 1.3826,
+      "step": 918
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.095934959349594e-06,
+      "loss": 1.2925,
+      "step": 919
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.092682926829268e-06,
+      "loss": 1.4488,
+      "step": 920
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.089430894308944e-06,
+      "loss": 1.489,
+      "step": 921
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.0861788617886185e-06,
+      "loss": 1.4431,
+      "step": 922
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.082926829268293e-06,
+      "loss": 1.4236,
+      "step": 923
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.079674796747968e-06,
+      "loss": 1.5587,
+      "step": 924
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.076422764227643e-06,
+      "loss": 1.399,
+      "step": 925
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 1.4993,
+      "step": 926
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.069918699186993e-06,
+      "loss": 1.4357,
+      "step": 927
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.066666666666667e-06,
+      "loss": 1.2067,
+      "step": 928
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.063414634146341e-06,
+      "loss": 1.3453,
+      "step": 929
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.0601626016260174e-06,
+      "loss": 1.2631,
+      "step": 930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.056910569105692e-06,
+      "loss": 1.5131,
+      "step": 931
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.053658536585366e-06,
+      "loss": 1.2649,
+      "step": 932
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.050406504065041e-06,
+      "loss": 1.2892,
+      "step": 933
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.047154471544716e-06,
+      "loss": 1.3005,
+      "step": 934
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.043902439024391e-06,
+      "loss": 1.3426,
+      "step": 935
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.040650406504066e-06,
+      "loss": 1.1382,
+      "step": 936
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.03739837398374e-06,
+      "loss": 1.157,
+      "step": 937
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.034146341463415e-06,
+      "loss": 1.4394,
+      "step": 938
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.030894308943091e-06,
+      "loss": 1.3353,
+      "step": 939
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.027642276422765e-06,
+      "loss": 1.1097,
+      "step": 940
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.024390243902439e-06,
+      "loss": 1.3169,
+      "step": 941
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0211382113821146e-06,
+      "loss": 1.3715,
+      "step": 942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017886178861789e-06,
+      "loss": 1.1851,
+      "step": 943
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014634146341463e-06,
+      "loss": 1.1243,
+      "step": 944
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.011382113821139e-06,
+      "loss": 1.3708,
+      "step": 945
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008130081300814e-06,
+      "loss": 1.1303,
+      "step": 946
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.004878048780488e-06,
+      "loss": 1.1304,
+      "step": 947
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.001626016260163e-06,
+      "loss": 1.1161,
+      "step": 948
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9983739837398375e-06,
+      "loss": 1.2576,
+      "step": 949
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.995121951219513e-06,
+      "loss": 1.1403,
+      "step": 950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.991869918699188e-06,
+      "loss": 1.162,
+      "step": 951
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.988617886178862e-06,
+      "loss": 1.325,
+      "step": 952
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9853658536585365e-06,
+      "loss": 1.1005,
+      "step": 953
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.9821138211382126e-06,
+      "loss": 1.1801,
+      "step": 954
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.978861788617887e-06,
+      "loss": 1.1385,
+      "step": 955
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.975609756097561e-06,
+      "loss": 1.347,
+      "step": 956
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.9723577235772364e-06,
+      "loss": 0.952,
+      "step": 957
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.969105691056911e-06,
+      "loss": 1.2789,
+      "step": 958
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.965853658536585e-06,
+      "loss": 1.3284,
+      "step": 959
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.962601626016261e-06,
+      "loss": 1.2914,
+      "step": 960
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.9593495934959355e-06,
+      "loss": 1.0652,
+      "step": 961
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.95609756097561e-06,
+      "loss": 1.2755,
+      "step": 962
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.952845528455285e-06,
+      "loss": 1.171,
+      "step": 963
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.94959349593496e-06,
+      "loss": 1.2474,
+      "step": 964
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.9463414634146345e-06,
+      "loss": 1.2498,
+      "step": 965
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.94308943089431e-06,
+      "loss": 1.3245,
+      "step": 966
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.939837398373984e-06,
+      "loss": 1.0161,
+      "step": 967
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.936585365853658e-06,
+      "loss": 1.1666,
+      "step": 968
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.0657,
+      "step": 969
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.930081300813009e-06,
+      "loss": 1.2206,
+      "step": 970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.926829268292683e-06,
+      "loss": 1.0813,
+      "step": 971
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.923577235772358e-06,
+      "loss": 1.1919,
+      "step": 972
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.920325203252033e-06,
+      "loss": 1.2482,
+      "step": 973
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.917073170731707e-06,
+      "loss": 1.1188,
+      "step": 974
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.913821138211383e-06,
+      "loss": 0.9906,
+      "step": 975
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.910569105691057e-06,
+      "loss": 1.2239,
+      "step": 976
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.907317073170732e-06,
+      "loss": 1.4231,
+      "step": 977
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.904065040650407e-06,
+      "loss": 1.2347,
+      "step": 978
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.900813008130082e-06,
+      "loss": 1.3025,
+      "step": 979
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.897560975609756e-06,
+      "loss": 1.1643,
+      "step": 980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.894308943089432e-06,
+      "loss": 1.5787,
+      "step": 981
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.891056910569106e-06,
+      "loss": 1.197,
+      "step": 982
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.88780487804878e-06,
+      "loss": 1.1107,
+      "step": 983
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.884552845528456e-06,
+      "loss": 1.1726,
+      "step": 984
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.881300813008131e-06,
+      "loss": 1.3131,
+      "step": 985
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.878048780487805e-06,
+      "loss": 1.1074,
+      "step": 986
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.87479674796748e-06,
+      "loss": 1.2034,
+      "step": 987
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.8715447154471545e-06,
+      "loss": 1.2221,
+      "step": 988
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.8682926829268305e-06,
+      "loss": 1.1808,
+      "step": 989
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.865040650406505e-06,
+      "loss": 1.1999,
+      "step": 990
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.861788617886179e-06,
+      "loss": 1.0814,
+      "step": 991
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.858536585365854e-06,
+      "loss": 1.37,
+      "step": 992
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.85528455284553e-06,
+      "loss": 1.4058,
+      "step": 993
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.852032520325204e-06,
+      "loss": 1.2208,
+      "step": 994
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.848780487804879e-06,
+      "loss": 1.0659,
+      "step": 995
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.8455284552845534e-06,
+      "loss": 1.2105,
+      "step": 996
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.842276422764228e-06,
+      "loss": 1.3592,
+      "step": 997
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.839024390243904e-06,
+      "loss": 1.1901,
+      "step": 998
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.835772357723578e-06,
+      "loss": 1.3219,
+      "step": 999
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.8325203252032525e-06,
+      "loss": 1.1611,
+      "step": 1000
+    },
+    {
+      "epoch": 3.22,
+      "eval_loss": 1.6613833904266357,
+      "eval_runtime": 143.9891,
+      "eval_samples_per_second": 4.209,
+      "eval_steps_per_second": 4.209,
+      "step": 1000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.2657,
+      "step": 1001
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.826016260162602e-06,
+      "loss": 1.2322,
+      "step": 1002
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.822764227642276e-06,
+      "loss": 1.2459,
+      "step": 1003
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.819512195121952e-06,
+      "loss": 1.1359,
+      "step": 1004
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.816260162601627e-06,
+      "loss": 1.1745,
+      "step": 1005
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.813008130081301e-06,
+      "loss": 1.2101,
+      "step": 1006
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.809756097560976e-06,
+      "loss": 1.2457,
+      "step": 1007
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 6.8065040650406514e-06,
+      "loss": 1.3701,
+      "step": 1008
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.803252032520326e-06,
+      "loss": 0.9807,
+      "step": 1009
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.1459,
+      "step": 1010
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.796747967479675e-06,
+      "loss": 1.1951,
+      "step": 1011
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.79349593495935e-06,
+      "loss": 1.1861,
+      "step": 1012
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.790243902439026e-06,
+      "loss": 1.0895,
+      "step": 1013
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.7869918699187e-06,
+      "loss": 1.1592,
+      "step": 1014
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.783739837398374e-06,
+      "loss": 1.3009,
+      "step": 1015
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.7804878048780495e-06,
+      "loss": 1.2371,
+      "step": 1016
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.777235772357724e-06,
+      "loss": 1.383,
+      "step": 1017
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.773983739837399e-06,
+      "loss": 1.2262,
+      "step": 1018
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.770731707317074e-06,
+      "loss": 1.1455,
+      "step": 1019
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.767479674796749e-06,
+      "loss": 1.0679,
+      "step": 1020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.764227642276423e-06,
+      "loss": 1.3603,
+      "step": 1021
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.760975609756099e-06,
+      "loss": 1.0226,
+      "step": 1022
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.757723577235773e-06,
+      "loss": 1.1112,
+      "step": 1023
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.754471544715448e-06,
+      "loss": 1.4014,
+      "step": 1024
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.751219512195123e-06,
+      "loss": 1.2159,
+      "step": 1025
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.747967479674797e-06,
+      "loss": 1.3411,
+      "step": 1026
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7447154471544715e-06,
+      "loss": 1.2595,
+      "step": 1027
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7414634146341475e-06,
+      "loss": 1.1438,
+      "step": 1028
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.738211382113822e-06,
+      "loss": 1.0971,
+      "step": 1029
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.734959349593496e-06,
+      "loss": 1.2523,
+      "step": 1030
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.731707317073171e-06,
+      "loss": 1.4866,
+      "step": 1031
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.728455284552846e-06,
+      "loss": 1.1862,
+      "step": 1032
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.725203252032521e-06,
+      "loss": 1.1817,
+      "step": 1033
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.721951219512196e-06,
+      "loss": 1.089,
+      "step": 1034
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.7186991869918704e-06,
+      "loss": 1.3383,
+      "step": 1035
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.715447154471545e-06,
+      "loss": 1.1175,
+      "step": 1036
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.712195121951221e-06,
+      "loss": 1.3579,
+      "step": 1037
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.708943089430895e-06,
+      "loss": 1.3161,
+      "step": 1038
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.7056910569105695e-06,
+      "loss": 1.2585,
+      "step": 1039
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.702439024390245e-06,
+      "loss": 1.1202,
+      "step": 1040
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.699186991869919e-06,
+      "loss": 1.3008,
+      "step": 1041
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.695934959349593e-06,
+      "loss": 1.1501,
+      "step": 1042
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.692682926829269e-06,
+      "loss": 1.0098,
+      "step": 1043
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.689430894308944e-06,
+      "loss": 1.2538,
+      "step": 1044
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.686178861788618e-06,
+      "loss": 1.3385,
+      "step": 1045
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.682926829268293e-06,
+      "loss": 1.3002,
+      "step": 1046
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.6796747967479684e-06,
+      "loss": 1.3446,
+      "step": 1047
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.676422764227643e-06,
+      "loss": 1.2232,
+      "step": 1048
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.673170731707318e-06,
+      "loss": 1.0826,
+      "step": 1049
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.669918699186992e-06,
+      "loss": 1.1711,
+      "step": 1050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.2756,
+      "step": 1051
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.663414634146343e-06,
+      "loss": 1.3283,
+      "step": 1052
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.660162601626017e-06,
+      "loss": 1.2222,
+      "step": 1053
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.656910569105691e-06,
+      "loss": 1.2166,
+      "step": 1054
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.6536585365853665e-06,
+      "loss": 1.4098,
+      "step": 1055
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.650406504065041e-06,
+      "loss": 1.3237,
+      "step": 1056
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.647154471544715e-06,
+      "loss": 1.3674,
+      "step": 1057
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.643902439024391e-06,
+      "loss": 1.3996,
+      "step": 1058
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.640650406504066e-06,
+      "loss": 1.2052,
+      "step": 1059
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.63739837398374e-06,
+      "loss": 1.2108,
+      "step": 1060
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.634146341463415e-06,
+      "loss": 1.0826,
+      "step": 1061
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.63089430894309e-06,
+      "loss": 1.1624,
+      "step": 1062
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.627642276422765e-06,
+      "loss": 1.1387,
+      "step": 1063
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.62439024390244e-06,
+      "loss": 1.1379,
+      "step": 1064
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.621138211382114e-06,
+      "loss": 1.0995,
+      "step": 1065
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.6178861788617885e-06,
+      "loss": 1.235,
+      "step": 1066
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.6146341463414645e-06,
+      "loss": 1.1541,
+      "step": 1067
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.611382113821139e-06,
+      "loss": 1.3567,
+      "step": 1068
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.608130081300813e-06,
+      "loss": 1.1845,
+      "step": 1069
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.604878048780488e-06,
+      "loss": 1.1251,
+      "step": 1070
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.601626016260163e-06,
+      "loss": 1.2585,
+      "step": 1071
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.598373983739838e-06,
+      "loss": 1.0045,
+      "step": 1072
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.595121951219513e-06,
+      "loss": 1.1859,
+      "step": 1073
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.5918699186991875e-06,
+      "loss": 1.4056,
+      "step": 1074
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.588617886178862e-06,
+      "loss": 1.2296,
+      "step": 1075
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 1.3136,
+      "step": 1076
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.582113821138212e-06,
+      "loss": 1.0881,
+      "step": 1077
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.5788617886178865e-06,
+      "loss": 1.1818,
+      "step": 1078
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.575609756097562e-06,
+      "loss": 1.2093,
+      "step": 1079
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.572357723577236e-06,
+      "loss": 1.061,
+      "step": 1080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.56910569105691e-06,
+      "loss": 1.0953,
+      "step": 1081
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.565853658536586e-06,
+      "loss": 1.1923,
+      "step": 1082
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.562601626016261e-06,
+      "loss": 1.2087,
+      "step": 1083
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.559349593495935e-06,
+      "loss": 1.4782,
+      "step": 1084
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.55609756097561e-06,
+      "loss": 1.2143,
+      "step": 1085
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.552845528455285e-06,
+      "loss": 1.2427,
+      "step": 1086
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.54959349593496e-06,
+      "loss": 1.1979,
+      "step": 1087
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.546341463414635e-06,
+      "loss": 1.094,
+      "step": 1088
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.543089430894309e-06,
+      "loss": 1.2291,
+      "step": 1089
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.539837398373984e-06,
+      "loss": 1.207,
+      "step": 1090
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.53658536585366e-06,
+      "loss": 1.3683,
+      "step": 1091
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.533333333333334e-06,
+      "loss": 1.2324,
+      "step": 1092
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.530081300813008e-06,
+      "loss": 1.2633,
+      "step": 1093
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.5268292682926836e-06,
+      "loss": 1.3169,
+      "step": 1094
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.523577235772358e-06,
+      "loss": 1.3235,
+      "step": 1095
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.520325203252032e-06,
+      "loss": 1.1122,
+      "step": 1096
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.517073170731708e-06,
+      "loss": 1.1915,
+      "step": 1097
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.513821138211383e-06,
+      "loss": 1.2246,
+      "step": 1098
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.510569105691057e-06,
+      "loss": 1.4952,
+      "step": 1099
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.507317073170732e-06,
+      "loss": 1.1352,
+      "step": 1100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 1.0845,
+      "step": 1101
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.500813008130082e-06,
+      "loss": 1.2807,
+      "step": 1102
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.497560975609757e-06,
+      "loss": 1.3159,
+      "step": 1103
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.494308943089431e-06,
+      "loss": 1.3593,
+      "step": 1104
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.4910569105691055e-06,
+      "loss": 1.0639,
+      "step": 1105
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.4878048780487815e-06,
+      "loss": 1.1625,
+      "step": 1106
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.484552845528456e-06,
+      "loss": 0.9431,
+      "step": 1107
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.48130081300813e-06,
+      "loss": 1.1634,
+      "step": 1108
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.478048780487805e-06,
+      "loss": 1.0886,
+      "step": 1109
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.47479674796748e-06,
+      "loss": 1.1194,
+      "step": 1110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.471544715447154e-06,
+      "loss": 1.0131,
+      "step": 1111
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.46829268292683e-06,
+      "loss": 1.2008,
+      "step": 1112
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.4650406504065045e-06,
+      "loss": 1.3865,
+      "step": 1113
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.461788617886179e-06,
+      "loss": 1.2803,
+      "step": 1114
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.458536585365854e-06,
+      "loss": 1.1345,
+      "step": 1115
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.455284552845529e-06,
+      "loss": 1.3843,
+      "step": 1116
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.452032520325204e-06,
+      "loss": 1.0555,
+      "step": 1117
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.448780487804879e-06,
+      "loss": 1.2365,
+      "step": 1118
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.445528455284553e-06,
+      "loss": 1.2278,
+      "step": 1119
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 6.442276422764229e-06,
+      "loss": 1.3309,
+      "step": 1120
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.439024390243903e-06,
+      "loss": 1.0142,
+      "step": 1121
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.435772357723578e-06,
+      "loss": 1.1222,
+      "step": 1122
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.432520325203253e-06,
+      "loss": 1.0934,
+      "step": 1123
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.429268292682927e-06,
+      "loss": 1.1245,
+      "step": 1124
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.426016260162602e-06,
+      "loss": 1.0694,
+      "step": 1125
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.422764227642278e-06,
+      "loss": 1.2918,
+      "step": 1126
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.419512195121952e-06,
+      "loss": 0.9973,
+      "step": 1127
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.416260162601626e-06,
+      "loss": 1.1723,
+      "step": 1128
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.4130081300813015e-06,
+      "loss": 1.1776,
+      "step": 1129
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.409756097560977e-06,
+      "loss": 1.2771,
+      "step": 1130
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.406504065040651e-06,
+      "loss": 1.2805,
+      "step": 1131
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.403252032520326e-06,
+      "loss": 0.9268,
+      "step": 1132
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.4026,
+      "step": 1133
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.396747967479675e-06,
+      "loss": 1.2264,
+      "step": 1134
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.393495934959351e-06,
+      "loss": 1.3098,
+      "step": 1135
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.390243902439025e-06,
+      "loss": 1.3694,
+      "step": 1136
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.3869918699187e-06,
+      "loss": 1.1043,
+      "step": 1137
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.383739837398375e-06,
+      "loss": 1.2205,
+      "step": 1138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.380487804878049e-06,
+      "loss": 1.1898,
+      "step": 1139
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.3772357723577235e-06,
+      "loss": 1.0788,
+      "step": 1140
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.3739837398373995e-06,
+      "loss": 1.135,
+      "step": 1141
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.370731707317074e-06,
+      "loss": 1.2373,
+      "step": 1142
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.367479674796748e-06,
+      "loss": 1.1192,
+      "step": 1143
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.364227642276423e-06,
+      "loss": 1.2122,
+      "step": 1144
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.3609756097560986e-06,
+      "loss": 1.3549,
+      "step": 1145
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.357723577235773e-06,
+      "loss": 1.0125,
+      "step": 1146
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.354471544715448e-06,
+      "loss": 1.2783,
+      "step": 1147
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.351219512195122e-06,
+      "loss": 1.0579,
+      "step": 1148
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.347967479674797e-06,
+      "loss": 1.0793,
+      "step": 1149
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.344715447154473e-06,
+      "loss": 1.1068,
+      "step": 1150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.1228,
+      "step": 1151
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.3382113821138215e-06,
+      "loss": 1.3829,
+      "step": 1152
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.334959349593497e-06,
+      "loss": 1.3064,
+      "step": 1153
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.331707317073171e-06,
+      "loss": 1.2586,
+      "step": 1154
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.328455284552845e-06,
+      "loss": 1.0042,
+      "step": 1155
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.325203252032521e-06,
+      "loss": 1.0902,
+      "step": 1156
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.321951219512196e-06,
+      "loss": 1.0343,
+      "step": 1157
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.31869918699187e-06,
+      "loss": 1.2152,
+      "step": 1158
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.315447154471546e-06,
+      "loss": 1.4384,
+      "step": 1159
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.31219512195122e-06,
+      "loss": 1.0673,
+      "step": 1160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.308943089430895e-06,
+      "loss": 1.2978,
+      "step": 1161
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.30569105691057e-06,
+      "loss": 1.4085,
+      "step": 1162
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.302439024390244e-06,
+      "loss": 1.1129,
+      "step": 1163
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.299186991869919e-06,
+      "loss": 1.4531,
+      "step": 1164
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.295934959349595e-06,
+      "loss": 1.3212,
+      "step": 1165
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.292682926829269e-06,
+      "loss": 1.3373,
+      "step": 1166
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.289430894308943e-06,
+      "loss": 0.9972,
+      "step": 1167
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.2861788617886185e-06,
+      "loss": 1.162,
+      "step": 1168
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.282926829268293e-06,
+      "loss": 1.3537,
+      "step": 1169
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.279674796747968e-06,
+      "loss": 1.3495,
+      "step": 1170
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.276422764227643e-06,
+      "loss": 0.9845,
+      "step": 1171
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.2731707317073176e-06,
+      "loss": 1.1474,
+      "step": 1172
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.269918699186992e-06,
+      "loss": 1.2666,
+      "step": 1173
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.266666666666668e-06,
+      "loss": 1.3012,
+      "step": 1174
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.263414634146342e-06,
+      "loss": 1.4606,
+      "step": 1175
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.260162601626017e-06,
+      "loss": 1.2515,
+      "step": 1176
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.256910569105692e-06,
+      "loss": 1.2432,
+      "step": 1177
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.253658536585366e-06,
+      "loss": 1.1543,
+      "step": 1178
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.2504065040650405e-06,
+      "loss": 1.3621,
+      "step": 1179
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.2471544715447165e-06,
+      "loss": 1.2284,
+      "step": 1180
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.243902439024391e-06,
+      "loss": 1.3843,
+      "step": 1181
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.240650406504065e-06,
+      "loss": 1.3898,
+      "step": 1182
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.23739837398374e-06,
+      "loss": 1.1366,
+      "step": 1183
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.234146341463415e-06,
+      "loss": 1.3341,
+      "step": 1184
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.23089430894309e-06,
+      "loss": 1.3732,
+      "step": 1185
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.227642276422765e-06,
+      "loss": 1.0965,
+      "step": 1186
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.2243902439024394e-06,
+      "loss": 1.2176,
+      "step": 1187
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.221138211382114e-06,
+      "loss": 1.1016,
+      "step": 1188
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.21788617886179e-06,
+      "loss": 1.3749,
+      "step": 1189
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.214634146341464e-06,
+      "loss": 1.2878,
+      "step": 1190
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.2113821138211385e-06,
+      "loss": 1.1914,
+      "step": 1191
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.208130081300814e-06,
+      "loss": 1.1613,
+      "step": 1192
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.204878048780488e-06,
+      "loss": 1.3497,
+      "step": 1193
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.201626016260162e-06,
+      "loss": 1.2401,
+      "step": 1194
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.198373983739838e-06,
+      "loss": 1.149,
+      "step": 1195
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.195121951219513e-06,
+      "loss": 1.124,
+      "step": 1196
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.191869918699187e-06,
+      "loss": 1.2402,
+      "step": 1197
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.188617886178862e-06,
+      "loss": 1.2365,
+      "step": 1198
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.185365853658537e-06,
+      "loss": 1.314,
+      "step": 1199
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.182113821138212e-06,
+      "loss": 1.2765,
+      "step": 1200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 1.2201,
+      "step": 1201
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.175609756097561e-06,
+      "loss": 1.1931,
+      "step": 1202
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.172357723577236e-06,
+      "loss": 1.1752,
+      "step": 1203
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.169105691056912e-06,
+      "loss": 1.1651,
+      "step": 1204
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.165853658536586e-06,
+      "loss": 1.2096,
+      "step": 1205
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.16260162601626e-06,
+      "loss": 1.3962,
+      "step": 1206
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.1593495934959355e-06,
+      "loss": 1.3824,
+      "step": 1207
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.15609756097561e-06,
+      "loss": 1.1957,
+      "step": 1208
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.152845528455284e-06,
+      "loss": 1.1053,
+      "step": 1209
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.14959349593496e-06,
+      "loss": 1.0156,
+      "step": 1210
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.1463414634146346e-06,
+      "loss": 1.4006,
+      "step": 1211
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.143089430894309e-06,
+      "loss": 1.1296,
+      "step": 1212
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.139837398373984e-06,
+      "loss": 1.3357,
+      "step": 1213
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.136585365853659e-06,
+      "loss": 1.4407,
+      "step": 1214
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.0873,
+      "step": 1215
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.130081300813009e-06,
+      "loss": 1.2407,
+      "step": 1216
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.126829268292683e-06,
+      "loss": 1.0922,
+      "step": 1217
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1235772357723575e-06,
+      "loss": 1.272,
+      "step": 1218
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.1203252032520335e-06,
+      "loss": 1.0886,
+      "step": 1219
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.117073170731708e-06,
+      "loss": 1.2364,
+      "step": 1220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.113821138211382e-06,
+      "loss": 1.1882,
+      "step": 1221
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.110569105691057e-06,
+      "loss": 1.3471,
+      "step": 1222
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.107317073170732e-06,
+      "loss": 1.176,
+      "step": 1223
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.104065040650407e-06,
+      "loss": 1.3021,
+      "step": 1224
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.100813008130082e-06,
+      "loss": 1.2732,
+      "step": 1225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 1.3303,
+      "step": 1226
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.094308943089431e-06,
+      "loss": 1.1978,
+      "step": 1227
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.091056910569107e-06,
+      "loss": 1.3248,
+      "step": 1228
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.087804878048781e-06,
+      "loss": 1.1954,
+      "step": 1229
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.0845528455284555e-06,
+      "loss": 1.2543,
+      "step": 1230
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.081300813008131e-06,
+      "loss": 1.2889,
+      "step": 1231
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.078048780487805e-06,
+      "loss": 1.3431,
+      "step": 1232
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.074796747967479e-06,
+      "loss": 1.303,
+      "step": 1233
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.071544715447155e-06,
+      "loss": 1.1532,
+      "step": 1234
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.06829268292683e-06,
+      "loss": 1.464,
+      "step": 1235
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.065040650406504e-06,
+      "loss": 1.3516,
+      "step": 1236
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.061788617886179e-06,
+      "loss": 1.2639,
+      "step": 1237
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.058536585365854e-06,
+      "loss": 1.2268,
+      "step": 1238
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.055284552845529e-06,
+      "loss": 1.0185,
+      "step": 1239
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.052032520325204e-06,
+      "loss": 1.2658,
+      "step": 1240
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.048780487804878e-06,
+      "loss": 1.297,
+      "step": 1241
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.045528455284553e-06,
+      "loss": 1.2631,
+      "step": 1242
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.042276422764229e-06,
+      "loss": 1.2362,
+      "step": 1243
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.039024390243903e-06,
+      "loss": 1.2008,
+      "step": 1244
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.035772357723578e-06,
+      "loss": 1.2248,
+      "step": 1245
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.0325203252032525e-06,
+      "loss": 0.9257,
+      "step": 1246
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 6.029268292682927e-06,
+      "loss": 1.1249,
+      "step": 1247
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.026016260162603e-06,
+      "loss": 1.1536,
+      "step": 1248
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.022764227642277e-06,
+      "loss": 1.0131,
+      "step": 1249
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 6.019512195121952e-06,
+      "loss": 0.9605,
+      "step": 1250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.016260162601627e-06,
+      "loss": 1.0759,
+      "step": 1251
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.013008130081301e-06,
+      "loss": 0.9553,
+      "step": 1252
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.009756097560976e-06,
+      "loss": 1.0583,
+      "step": 1253
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.0065040650406515e-06,
+      "loss": 0.9929,
+      "step": 1254
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.003252032520326e-06,
+      "loss": 1.0566,
+      "step": 1255
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6e-06,
+      "loss": 0.9254,
+      "step": 1256
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.996747967479676e-06,
+      "loss": 1.0721,
+      "step": 1257
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.9934959349593505e-06,
+      "loss": 0.9717,
+      "step": 1258
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.990243902439025e-06,
+      "loss": 0.9898,
+      "step": 1259
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.9869918699187e-06,
+      "loss": 1.0405,
+      "step": 1260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.983739837398374e-06,
+      "loss": 1.3534,
+      "step": 1261
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.980487804878049e-06,
+      "loss": 1.0363,
+      "step": 1262
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.977235772357725e-06,
+      "loss": 1.1055,
+      "step": 1263
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.973983739837399e-06,
+      "loss": 1.1993,
+      "step": 1264
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.9707317073170734e-06,
+      "loss": 0.9142,
+      "step": 1265
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.967479674796749e-06,
+      "loss": 1.1987,
+      "step": 1266
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.964227642276423e-06,
+      "loss": 1.0954,
+      "step": 1267
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.960975609756098e-06,
+      "loss": 1.2091,
+      "step": 1268
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.957723577235773e-06,
+      "loss": 1.1212,
+      "step": 1269
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.954471544715448e-06,
+      "loss": 1.1564,
+      "step": 1270
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.951219512195122e-06,
+      "loss": 1.1198,
+      "step": 1271
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.947967479674798e-06,
+      "loss": 1.0272,
+      "step": 1272
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.944715447154472e-06,
+      "loss": 1.2623,
+      "step": 1273
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.941463414634147e-06,
+      "loss": 1.3364,
+      "step": 1274
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.938211382113822e-06,
+      "loss": 1.109,
+      "step": 1275
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.934959349593496e-06,
+      "loss": 1.2003,
+      "step": 1276
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.931707317073171e-06,
+      "loss": 1.0284,
+      "step": 1277
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.928455284552847e-06,
+      "loss": 1.1626,
+      "step": 1278
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.925203252032521e-06,
+      "loss": 1.2776,
+      "step": 1279
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.921951219512195e-06,
+      "loss": 1.2909,
+      "step": 1280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.9186991869918705e-06,
+      "loss": 1.2091,
+      "step": 1281
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.915447154471546e-06,
+      "loss": 1.0834,
+      "step": 1282
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.91219512195122e-06,
+      "loss": 1.0771,
+      "step": 1283
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.908943089430895e-06,
+      "loss": 1.0411,
+      "step": 1284
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.9056910569105695e-06,
+      "loss": 1.0266,
+      "step": 1285
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.902439024390244e-06,
+      "loss": 1.0965,
+      "step": 1286
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.89918699186992e-06,
+      "loss": 1.0904,
+      "step": 1287
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.895934959349594e-06,
+      "loss": 1.3399,
+      "step": 1288
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.892682926829269e-06,
+      "loss": 1.1701,
+      "step": 1289
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 5.889430894308944e-06,
+      "loss": 1.1237,
+      "step": 1290
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.886178861788618e-06,
+      "loss": 1.1577,
+      "step": 1291
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.8829268292682924e-06,
+      "loss": 1.0144,
+      "step": 1292
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.8796747967479685e-06,
+      "loss": 1.0652,
+      "step": 1293
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.876422764227643e-06,
+      "loss": 1.0704,
+      "step": 1294
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.873170731707317e-06,
+      "loss": 1.1192,
+      "step": 1295
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.869918699186992e-06,
+      "loss": 1.1754,
+      "step": 1296
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.1538,
+      "step": 1297
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.863414634146342e-06,
+      "loss": 0.913,
+      "step": 1298
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.860162601626017e-06,
+      "loss": 1.2208,
+      "step": 1299
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.856910569105691e-06,
+      "loss": 1.0298,
+      "step": 1300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.8683,
+      "step": 1301
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.850406504065042e-06,
+      "loss": 0.9575,
+      "step": 1302
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.847154471544716e-06,
+      "loss": 1.254,
+      "step": 1303
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.8439024390243904e-06,
+      "loss": 0.9902,
+      "step": 1304
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.840650406504066e-06,
+      "loss": 1.238,
+      "step": 1305
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.83739837398374e-06,
+      "loss": 1.0735,
+      "step": 1306
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.834146341463415e-06,
+      "loss": 1.0255,
+      "step": 1307
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.83089430894309e-06,
+      "loss": 1.1252,
+      "step": 1308
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.827642276422765e-06,
+      "loss": 1.0589,
+      "step": 1309
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.824390243902439e-06,
+      "loss": 1.097,
+      "step": 1310
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.821138211382115e-06,
+      "loss": 1.0057,
+      "step": 1311
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.817886178861789e-06,
+      "loss": 1.1281,
+      "step": 1312
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.814634146341464e-06,
+      "loss": 1.0399,
+      "step": 1313
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.811382113821139e-06,
+      "loss": 1.1834,
+      "step": 1314
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.808130081300813e-06,
+      "loss": 1.0898,
+      "step": 1315
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.804878048780488e-06,
+      "loss": 1.2561,
+      "step": 1316
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.801626016260164e-06,
+      "loss": 1.1021,
+      "step": 1317
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.798373983739838e-06,
+      "loss": 1.3042,
+      "step": 1318
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.795121951219512e-06,
+      "loss": 1.2167,
+      "step": 1319
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7918699186991875e-06,
+      "loss": 0.939,
+      "step": 1320
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.788617886178862e-06,
+      "loss": 1.0567,
+      "step": 1321
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.785365853658537e-06,
+      "loss": 1.0452,
+      "step": 1322
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.782113821138212e-06,
+      "loss": 1.0248,
+      "step": 1323
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.7788617886178865e-06,
+      "loss": 1.1581,
+      "step": 1324
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.775609756097561e-06,
+      "loss": 1.0629,
+      "step": 1325
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.772357723577237e-06,
+      "loss": 1.1443,
+      "step": 1326
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.769105691056911e-06,
+      "loss": 1.1666,
+      "step": 1327
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.765853658536586e-06,
+      "loss": 1.0928,
+      "step": 1328
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.762601626016261e-06,
+      "loss": 1.1136,
+      "step": 1329
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.759349593495935e-06,
+      "loss": 1.124,
+      "step": 1330
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.7560975609756095e-06,
+      "loss": 1.1792,
+      "step": 1331
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.7528455284552855e-06,
+      "loss": 1.1074,
+      "step": 1332
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.74959349593496e-06,
+      "loss": 1.1843,
+      "step": 1333
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.746341463414634e-06,
+      "loss": 1.3069,
+      "step": 1334
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.743089430894309e-06,
+      "loss": 1.0718,
+      "step": 1335
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.7398373983739845e-06,
+      "loss": 1.2497,
+      "step": 1336
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.736585365853659e-06,
+      "loss": 1.2793,
+      "step": 1337
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 1.0235,
+      "step": 1338
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.730081300813008e-06,
+      "loss": 1.1484,
+      "step": 1339
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.726829268292683e-06,
+      "loss": 1.1677,
+      "step": 1340
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.723577235772359e-06,
+      "loss": 1.2075,
+      "step": 1341
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.720325203252033e-06,
+      "loss": 1.2569,
+      "step": 1342
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.7170731707317075e-06,
+      "loss": 1.1348,
+      "step": 1343
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.713821138211383e-06,
+      "loss": 1.2936,
+      "step": 1344
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.710569105691057e-06,
+      "loss": 1.2871,
+      "step": 1345
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.707317073170731e-06,
+      "loss": 1.0735,
+      "step": 1346
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.704065040650407e-06,
+      "loss": 1.0016,
+      "step": 1347
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.700813008130082e-06,
+      "loss": 0.9385,
+      "step": 1348
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.697560975609756e-06,
+      "loss": 1.2066,
+      "step": 1349
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.694308943089431e-06,
+      "loss": 1.1172,
+      "step": 1350
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.691056910569106e-06,
+      "loss": 1.1327,
+      "step": 1351
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.687804878048781e-06,
+      "loss": 0.9313,
+      "step": 1352
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.684552845528456e-06,
+      "loss": 1.1189,
+      "step": 1353
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.68130081300813e-06,
+      "loss": 0.9958,
+      "step": 1354
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.678048780487805e-06,
+      "loss": 1.0439,
+      "step": 1355
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.674796747967481e-06,
+      "loss": 1.2137,
+      "step": 1356
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.671544715447155e-06,
+      "loss": 1.0797,
+      "step": 1357
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.668292682926829e-06,
+      "loss": 1.052,
+      "step": 1358
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.6650406504065045e-06,
+      "loss": 1.0593,
+      "step": 1359
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.661788617886179e-06,
+      "loss": 1.0434,
+      "step": 1360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.658536585365853e-06,
+      "loss": 1.3231,
+      "step": 1361
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.655284552845529e-06,
+      "loss": 1.1211,
+      "step": 1362
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.6520325203252035e-06,
+      "loss": 1.2072,
+      "step": 1363
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.648780487804878e-06,
+      "loss": 1.0603,
+      "step": 1364
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.645528455284554e-06,
+      "loss": 1.2229,
+      "step": 1365
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.642276422764228e-06,
+      "loss": 1.1257,
+      "step": 1366
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.639024390243903e-06,
+      "loss": 0.957,
+      "step": 1367
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 5.635772357723578e-06,
+      "loss": 1.0924,
+      "step": 1368
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.632520325203252e-06,
+      "loss": 1.1182,
+      "step": 1369
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.6292682926829265e-06,
+      "loss": 1.1589,
+      "step": 1370
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.6260162601626025e-06,
+      "loss": 1.0026,
+      "step": 1371
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.622764227642277e-06,
+      "loss": 0.9912,
+      "step": 1372
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.619512195121952e-06,
+      "loss": 1.279,
+      "step": 1373
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.616260162601626e-06,
+      "loss": 1.1666,
+      "step": 1374
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.613008130081301e-06,
+      "loss": 0.9102,
+      "step": 1375
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 1.1439,
+      "step": 1376
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.606504065040651e-06,
+      "loss": 1.1706,
+      "step": 1377
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.603252032520325e-06,
+      "loss": 1.0788,
+      "step": 1378
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.9954,
+      "step": 1379
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.596747967479676e-06,
+      "loss": 1.1444,
+      "step": 1380
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.59349593495935e-06,
+      "loss": 0.8948,
+      "step": 1381
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.590243902439025e-06,
+      "loss": 1.0902,
+      "step": 1382
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.5869918699187e-06,
+      "loss": 1.2826,
+      "step": 1383
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 5.583739837398374e-06,
+      "loss": 1.3139,
+      "step": 1384
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.58048780487805e-06,
+      "loss": 1.0528,
+      "step": 1385
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.577235772357724e-06,
+      "loss": 1.1904,
+      "step": 1386
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.573983739837399e-06,
+      "loss": 1.1637,
+      "step": 1387
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.570731707317074e-06,
+      "loss": 1.0404,
+      "step": 1388
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.567479674796748e-06,
+      "loss": 0.8807,
+      "step": 1389
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.5642276422764226e-06,
+      "loss": 1.1282,
+      "step": 1390
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.560975609756099e-06,
+      "loss": 1.0252,
+      "step": 1391
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.557723577235773e-06,
+      "loss": 1.0999,
+      "step": 1392
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.554471544715447e-06,
+      "loss": 1.0255,
+      "step": 1393
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.551219512195123e-06,
+      "loss": 0.9867,
+      "step": 1394
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.547967479674798e-06,
+      "loss": 1.2341,
+      "step": 1395
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.544715447154472e-06,
+      "loss": 0.8869,
+      "step": 1396
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.541463414634147e-06,
+      "loss": 1.3785,
+      "step": 1397
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.5382113821138215e-06,
+      "loss": 1.0284,
+      "step": 1398
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.534959349593496e-06,
+      "loss": 1.0371,
+      "step": 1399
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.531707317073172e-06,
+      "loss": 1.0703,
+      "step": 1400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 1.0345,
+      "step": 1401
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.5252032520325206e-06,
+      "loss": 0.9989,
+      "step": 1402
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.521951219512196e-06,
+      "loss": 1.2392,
+      "step": 1403
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.51869918699187e-06,
+      "loss": 1.1197,
+      "step": 1404
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.515447154471545e-06,
+      "loss": 1.0841,
+      "step": 1405
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.5121951219512205e-06,
+      "loss": 1.2022,
+      "step": 1406
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.508943089430895e-06,
+      "loss": 1.134,
+      "step": 1407
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.505691056910569e-06,
+      "loss": 1.0421,
+      "step": 1408
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.502439024390245e-06,
+      "loss": 1.0369,
+      "step": 1409
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.4991869918699195e-06,
+      "loss": 1.1463,
+      "step": 1410
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.495934959349594e-06,
+      "loss": 1.0583,
+      "step": 1411
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.492682926829269e-06,
+      "loss": 1.2942,
+      "step": 1412
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.489430894308943e-06,
+      "loss": 1.007,
+      "step": 1413
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.486178861788618e-06,
+      "loss": 1.0757,
+      "step": 1414
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.482926829268294e-06,
+      "loss": 1.0156,
+      "step": 1415
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.479674796747968e-06,
+      "loss": 1.0921,
+      "step": 1416
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.476422764227642e-06,
+      "loss": 1.1842,
+      "step": 1417
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.473170731707318e-06,
+      "loss": 1.1599,
+      "step": 1418
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.469918699186993e-06,
+      "loss": 1.1192,
+      "step": 1419
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.466666666666667e-06,
+      "loss": 0.9996,
+      "step": 1420
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.463414634146342e-06,
+      "loss": 1.1386,
+      "step": 1421
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.460162601626017e-06,
+      "loss": 0.9959,
+      "step": 1422
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.456910569105691e-06,
+      "loss": 0.9882,
+      "step": 1423
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.453658536585367e-06,
+      "loss": 1.1396,
+      "step": 1424
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.450406504065041e-06,
+      "loss": 1.0513,
+      "step": 1425
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.447154471544716e-06,
+      "loss": 0.9922,
+      "step": 1426
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.443902439024391e-06,
+      "loss": 0.9583,
+      "step": 1427
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.440650406504065e-06,
+      "loss": 1.151,
+      "step": 1428
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.4373983739837396e-06,
+      "loss": 0.9857,
+      "step": 1429
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 5.434146341463416e-06,
+      "loss": 0.9978,
+      "step": 1430
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.43089430894309e-06,
+      "loss": 1.2528,
+      "step": 1431
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.427642276422764e-06,
+      "loss": 0.9702,
+      "step": 1432
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.4243902439024395e-06,
+      "loss": 1.1763,
+      "step": 1433
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.421138211382115e-06,
+      "loss": 1.2789,
+      "step": 1434
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.417886178861789e-06,
+      "loss": 1.2784,
+      "step": 1435
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.414634146341464e-06,
+      "loss": 1.0875,
+      "step": 1436
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.4113821138211385e-06,
+      "loss": 1.3814,
+      "step": 1437
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.408130081300813e-06,
+      "loss": 1.0919,
+      "step": 1438
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.404878048780489e-06,
+      "loss": 1.1414,
+      "step": 1439
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.401626016260163e-06,
+      "loss": 1.2051,
+      "step": 1440
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.3983739837398376e-06,
+      "loss": 1.1694,
+      "step": 1441
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.395121951219513e-06,
+      "loss": 1.093,
+      "step": 1442
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.391869918699187e-06,
+      "loss": 1.1953,
+      "step": 1443
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.3886178861788614e-06,
+      "loss": 1.134,
+      "step": 1444
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.3853658536585375e-06,
+      "loss": 1.1052,
+      "step": 1445
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 5.382113821138212e-06,
+      "loss": 1.036,
+      "step": 1446
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.378861788617886e-06,
+      "loss": 1.1524,
+      "step": 1447
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.375609756097562e-06,
+      "loss": 1.1723,
+      "step": 1448
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.3723577235772365e-06,
+      "loss": 1.1324,
+      "step": 1449
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.369105691056911e-06,
+      "loss": 1.065,
+      "step": 1450
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.0276,
+      "step": 1451
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.36260162601626e-06,
+      "loss": 1.2683,
+      "step": 1452
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.359349593495935e-06,
+      "loss": 1.0784,
+      "step": 1453
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.356097560975611e-06,
+      "loss": 1.1129,
+      "step": 1454
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.352845528455285e-06,
+      "loss": 1.1163,
+      "step": 1455
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.349593495934959e-06,
+      "loss": 1.0712,
+      "step": 1456
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.346341463414635e-06,
+      "loss": 1.034,
+      "step": 1457
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.343089430894309e-06,
+      "loss": 1.2147,
+      "step": 1458
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.339837398373984e-06,
+      "loss": 0.9145,
+      "step": 1459
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.336585365853659e-06,
+      "loss": 1.1978,
+      "step": 1460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.0238,
+      "step": 1461
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.330081300813008e-06,
+      "loss": 1.2682,
+      "step": 1462
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.326829268292684e-06,
+      "loss": 1.0094,
+      "step": 1463
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.323577235772358e-06,
+      "loss": 1.2181,
+      "step": 1464
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.320325203252033e-06,
+      "loss": 0.9647,
+      "step": 1465
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.317073170731708e-06,
+      "loss": 1.0831,
+      "step": 1466
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.313821138211382e-06,
+      "loss": 1.378,
+      "step": 1467
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.3105691056910566e-06,
+      "loss": 1.131,
+      "step": 1468
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.307317073170733e-06,
+      "loss": 1.1164,
+      "step": 1469
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.304065040650407e-06,
+      "loss": 1.0365,
+      "step": 1470
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.300813008130081e-06,
+      "loss": 1.2663,
+      "step": 1471
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.2975609756097565e-06,
+      "loss": 1.2511,
+      "step": 1472
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.294308943089431e-06,
+      "loss": 0.9363,
+      "step": 1473
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.291056910569106e-06,
+      "loss": 0.9159,
+      "step": 1474
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.287804878048781e-06,
+      "loss": 1.3559,
+      "step": 1475
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.2845528455284555e-06,
+      "loss": 1.2279,
+      "step": 1476
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.28130081300813e-06,
+      "loss": 1.0565,
+      "step": 1477
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.278048780487806e-06,
+      "loss": 0.9825,
+      "step": 1478
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.27479674796748e-06,
+      "loss": 1.1157,
+      "step": 1479
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.2715447154471546e-06,
+      "loss": 1.3445,
+      "step": 1480
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.26829268292683e-06,
+      "loss": 1.1031,
+      "step": 1481
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.265040650406504e-06,
+      "loss": 1.2746,
+      "step": 1482
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.2617886178861784e-06,
+      "loss": 1.0763,
+      "step": 1483
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.2585365853658545e-06,
+      "loss": 1.2036,
+      "step": 1484
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.255284552845529e-06,
+      "loss": 1.1001,
+      "step": 1485
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.252032520325203e-06,
+      "loss": 1.2596,
+      "step": 1486
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.248780487804878e-06,
+      "loss": 0.9532,
+      "step": 1487
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.2455284552845535e-06,
+      "loss": 1.0957,
+      "step": 1488
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.242276422764228e-06,
+      "loss": 1.0599,
+      "step": 1489
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.239024390243903e-06,
+      "loss": 1.1065,
+      "step": 1490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.235772357723577e-06,
+      "loss": 1.1304,
+      "step": 1491
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 5.232520325203252e-06,
+      "loss": 1.2442,
+      "step": 1492
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.229268292682928e-06,
+      "loss": 1.1269,
+      "step": 1493
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.226016260162602e-06,
+      "loss": 1.0788,
+      "step": 1494
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.2227642276422764e-06,
+      "loss": 1.2402,
+      "step": 1495
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.219512195121952e-06,
+      "loss": 1.2955,
+      "step": 1496
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.216260162601626e-06,
+      "loss": 1.2202,
+      "step": 1497
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.2130081300813e-06,
+      "loss": 1.1198,
+      "step": 1498
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.209756097560976e-06,
+      "loss": 1.1342,
+      "step": 1499
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.206504065040651e-06,
+      "loss": 1.0975,
+      "step": 1500
+    },
+    {
+      "epoch": 4.83,
+      "eval_loss": 1.7174352407455444,
+      "eval_runtime": 133.5304,
+      "eval_samples_per_second": 4.538,
+      "eval_steps_per_second": 4.538,
+      "step": 1500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 1.1915,
+      "step": 1501
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.2e-06,
+      "loss": 1.0767,
+      "step": 1502
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.196747967479675e-06,
+      "loss": 1.1084,
+      "step": 1503
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1934959349593506e-06,
+      "loss": 1.2892,
+      "step": 1504
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.190243902439025e-06,
+      "loss": 1.2166,
+      "step": 1505
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.186991869918699e-06,
+      "loss": 1.0476,
+      "step": 1506
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.183739837398375e-06,
+      "loss": 1.0178,
+      "step": 1507
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.18048780487805e-06,
+      "loss": 1.1154,
+      "step": 1508
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.177235772357724e-06,
+      "loss": 1.0165,
+      "step": 1509
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.173983739837399e-06,
+      "loss": 0.9262,
+      "step": 1510
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.1707317073170735e-06,
+      "loss": 1.276,
+      "step": 1511
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.167479674796748e-06,
+      "loss": 0.9563,
+      "step": 1512
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.164227642276424e-06,
+      "loss": 1.3505,
+      "step": 1513
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.160975609756098e-06,
+      "loss": 1.0969,
+      "step": 1514
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.1577235772357725e-06,
+      "loss": 0.8611,
+      "step": 1515
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.154471544715448e-06,
+      "loss": 0.9759,
+      "step": 1516
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.151219512195123e-06,
+      "loss": 1.0851,
+      "step": 1517
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.147967479674797e-06,
+      "loss": 1.0684,
+      "step": 1518
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.144715447154472e-06,
+      "loss": 1.0799,
+      "step": 1519
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.141463414634147e-06,
+      "loss": 0.9139,
+      "step": 1520
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.138211382113821e-06,
+      "loss": 1.1816,
+      "step": 1521
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.134959349593497e-06,
+      "loss": 1.146,
+      "step": 1522
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.1317073170731715e-06,
+      "loss": 0.9601,
+      "step": 1523
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.128455284552846e-06,
+      "loss": 0.9562,
+      "step": 1524
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.125203252032521e-06,
+      "loss": 0.8877,
+      "step": 1525
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 1.1634,
+      "step": 1526
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.11869918699187e-06,
+      "loss": 1.2269,
+      "step": 1527
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.115447154471546e-06,
+      "loss": 0.9813,
+      "step": 1528
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.11219512195122e-06,
+      "loss": 0.9424,
+      "step": 1529
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.108943089430894e-06,
+      "loss": 1.116,
+      "step": 1530
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.1056910569105696e-06,
+      "loss": 1.1866,
+      "step": 1531
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.102439024390245e-06,
+      "loss": 1.1873,
+      "step": 1532
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.099186991869919e-06,
+      "loss": 1.0021,
+      "step": 1533
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.095934959349594e-06,
+      "loss": 1.2074,
+      "step": 1534
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.092682926829269e-06,
+      "loss": 0.9239,
+      "step": 1535
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.089430894308943e-06,
+      "loss": 1.141,
+      "step": 1536
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.086178861788619e-06,
+      "loss": 0.9917,
+      "step": 1537
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.082926829268293e-06,
+      "loss": 1.0532,
+      "step": 1538
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.079674796747968e-06,
+      "loss": 1.0571,
+      "step": 1539
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.076422764227643e-06,
+      "loss": 1.1129,
+      "step": 1540
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.073170731707317e-06,
+      "loss": 1.058,
+      "step": 1541
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.069918699186992e-06,
+      "loss": 1.153,
+      "step": 1542
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.287,
+      "step": 1543
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.063414634146342e-06,
+      "loss": 1.1749,
+      "step": 1544
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.060162601626016e-06,
+      "loss": 1.2977,
+      "step": 1545
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.056910569105692e-06,
+      "loss": 1.1601,
+      "step": 1546
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.053658536585367e-06,
+      "loss": 1.156,
+      "step": 1547
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.050406504065041e-06,
+      "loss": 1.1581,
+      "step": 1548
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.047154471544716e-06,
+      "loss": 1.0911,
+      "step": 1549
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.0439024390243905e-06,
+      "loss": 1.1136,
+      "step": 1550
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.040650406504065e-06,
+      "loss": 0.8821,
+      "step": 1551
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.037398373983741e-06,
+      "loss": 1.0402,
+      "step": 1552
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.034146341463415e-06,
+      "loss": 1.0701,
+      "step": 1553
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.0308943089430895e-06,
+      "loss": 1.0937,
+      "step": 1554
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.027642276422765e-06,
+      "loss": 1.0575,
+      "step": 1555
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.024390243902439e-06,
+      "loss": 0.8814,
+      "step": 1556
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.021138211382114e-06,
+      "loss": 1.0469,
+      "step": 1557
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 5.0178861788617894e-06,
+      "loss": 1.0073,
+      "step": 1558
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.014634146341464e-06,
+      "loss": 0.7996,
+      "step": 1559
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.011382113821138e-06,
+      "loss": 0.8944,
+      "step": 1560
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 5.008130081300814e-06,
+      "loss": 0.98,
+      "step": 1561
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.0048780487804885e-06,
+      "loss": 1.1446,
+      "step": 1562
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 5.001626016260163e-06,
+      "loss": 1.2242,
+      "step": 1563
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.998373983739838e-06,
+      "loss": 1.0575,
+      "step": 1564
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.995121951219512e-06,
+      "loss": 0.891,
+      "step": 1565
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.9918699186991875e-06,
+      "loss": 0.9742,
+      "step": 1566
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.988617886178862e-06,
+      "loss": 1.0757,
+      "step": 1567
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.985365853658537e-06,
+      "loss": 0.7602,
+      "step": 1568
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.982113821138212e-06,
+      "loss": 1.0343,
+      "step": 1569
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.978861788617887e-06,
+      "loss": 0.8672,
+      "step": 1570
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.975609756097562e-06,
+      "loss": 1.0293,
+      "step": 1571
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.972357723577236e-06,
+      "loss": 1.1179,
+      "step": 1572
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.9691056910569104e-06,
+      "loss": 1.1178,
+      "step": 1573
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.965853658536586e-06,
+      "loss": 1.1127,
+      "step": 1574
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.962601626016261e-06,
+      "loss": 0.893,
+      "step": 1575
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.959349593495935e-06,
+      "loss": 0.9309,
+      "step": 1576
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.95609756097561e-06,
+      "loss": 1.0755,
+      "step": 1577
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.9528455284552855e-06,
+      "loss": 1.0875,
+      "step": 1578
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.94959349593496e-06,
+      "loss": 0.933,
+      "step": 1579
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.946341463414634e-06,
+      "loss": 0.9711,
+      "step": 1580
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.943089430894309e-06,
+      "loss": 0.9712,
+      "step": 1581
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.939837398373984e-06,
+      "loss": 1.0088,
+      "step": 1582
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.936585365853659e-06,
+      "loss": 0.9043,
+      "step": 1583
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.933333333333334e-06,
+      "loss": 0.8725,
+      "step": 1584
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.9300813008130084e-06,
+      "loss": 0.914,
+      "step": 1585
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.926829268292684e-06,
+      "loss": 0.9006,
+      "step": 1586
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.923577235772358e-06,
+      "loss": 1.1017,
+      "step": 1587
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.920325203252032e-06,
+      "loss": 1.0621,
+      "step": 1588
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.9170731707317075e-06,
+      "loss": 1.0823,
+      "step": 1589
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.913821138211383e-06,
+      "loss": 1.0287,
+      "step": 1590
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.910569105691057e-06,
+      "loss": 1.1324,
+      "step": 1591
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.907317073170732e-06,
+      "loss": 1.2411,
+      "step": 1592
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.904065040650407e-06,
+      "loss": 1.0052,
+      "step": 1593
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.900813008130082e-06,
+      "loss": 0.9881,
+      "step": 1594
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.897560975609756e-06,
+      "loss": 0.9639,
+      "step": 1595
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.894308943089431e-06,
+      "loss": 0.9631,
+      "step": 1596
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.8910569105691064e-06,
+      "loss": 0.9968,
+      "step": 1597
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.887804878048781e-06,
+      "loss": 0.9476,
+      "step": 1598
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.884552845528456e-06,
+      "loss": 0.8729,
+      "step": 1599
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.881300813008131e-06,
+      "loss": 1.1358,
+      "step": 1600
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9778,
+      "step": 1601
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.87479674796748e-06,
+      "loss": 1.1723,
+      "step": 1602
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.871544715447155e-06,
+      "loss": 1.0625,
+      "step": 1603
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.868292682926829e-06,
+      "loss": 0.9825,
+      "step": 1604
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.8650406504065045e-06,
+      "loss": 1.0015,
+      "step": 1605
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.86178861788618e-06,
+      "loss": 1.0408,
+      "step": 1606
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.858536585365854e-06,
+      "loss": 1.034,
+      "step": 1607
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.855284552845529e-06,
+      "loss": 0.9738,
+      "step": 1608
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.852032520325204e-06,
+      "loss": 1.0383,
+      "step": 1609
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.848780487804878e-06,
+      "loss": 1.1666,
+      "step": 1610
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.845528455284553e-06,
+      "loss": 1.1905,
+      "step": 1611
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.842276422764228e-06,
+      "loss": 1.0977,
+      "step": 1612
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.839024390243903e-06,
+      "loss": 1.0756,
+      "step": 1613
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.835772357723578e-06,
+      "loss": 0.9964,
+      "step": 1614
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.832520325203253e-06,
+      "loss": 1.0978,
+      "step": 1615
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.829268292682927e-06,
+      "loss": 1.0525,
+      "step": 1616
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.826016260162602e-06,
+      "loss": 1.137,
+      "step": 1617
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.822764227642277e-06,
+      "loss": 0.9033,
+      "step": 1618
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.819512195121951e-06,
+      "loss": 1.0666,
+      "step": 1619
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.816260162601626e-06,
+      "loss": 1.0074,
+      "step": 1620
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.813008130081302e-06,
+      "loss": 0.9782,
+      "step": 1621
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.809756097560976e-06,
+      "loss": 0.99,
+      "step": 1622
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.806504065040651e-06,
+      "loss": 1.0449,
+      "step": 1623
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.8032520325203254e-06,
+      "loss": 0.9315,
+      "step": 1624
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.1645,
+      "step": 1625
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.796747967479675e-06,
+      "loss": 0.9909,
+      "step": 1626
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.79349593495935e-06,
+      "loss": 0.9757,
+      "step": 1627
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7902439024390245e-06,
+      "loss": 0.8856,
+      "step": 1628
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7869918699187e-06,
+      "loss": 0.9138,
+      "step": 1629
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.783739837398375e-06,
+      "loss": 0.9399,
+      "step": 1630
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.780487804878049e-06,
+      "loss": 0.8749,
+      "step": 1631
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7772357723577235e-06,
+      "loss": 1.2018,
+      "step": 1632
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.773983739837399e-06,
+      "loss": 1.1466,
+      "step": 1633
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.770731707317073e-06,
+      "loss": 0.9997,
+      "step": 1634
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.767479674796748e-06,
+      "loss": 0.8639,
+      "step": 1635
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.7642276422764234e-06,
+      "loss": 1.0171,
+      "step": 1636
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.760975609756098e-06,
+      "loss": 1.0589,
+      "step": 1637
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.757723577235773e-06,
+      "loss": 1.0185,
+      "step": 1638
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.754471544715447e-06,
+      "loss": 1.069,
+      "step": 1639
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.7512195121951225e-06,
+      "loss": 1.031,
+      "step": 1640
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.747967479674797e-06,
+      "loss": 0.9436,
+      "step": 1641
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.744715447154472e-06,
+      "loss": 0.9884,
+      "step": 1642
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.741463414634146e-06,
+      "loss": 1.0833,
+      "step": 1643
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.7382113821138215e-06,
+      "loss": 1.0898,
+      "step": 1644
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.734959349593497e-06,
+      "loss": 0.9203,
+      "step": 1645
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.731707317073171e-06,
+      "loss": 0.8002,
+      "step": 1646
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.728455284552846e-06,
+      "loss": 1.0153,
+      "step": 1647
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.725203252032521e-06,
+      "loss": 0.9054,
+      "step": 1648
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.721951219512195e-06,
+      "loss": 0.8151,
+      "step": 1649
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.71869918699187e-06,
+      "loss": 0.9417,
+      "step": 1650
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.715447154471545e-06,
+      "loss": 0.9304,
+      "step": 1651
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.71219512195122e-06,
+      "loss": 1.0276,
+      "step": 1652
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.708943089430895e-06,
+      "loss": 1.0429,
+      "step": 1653
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.70569105691057e-06,
+      "loss": 0.8541,
+      "step": 1654
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.702439024390244e-06,
+      "loss": 1.1636,
+      "step": 1655
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.699186991869919e-06,
+      "loss": 1.0707,
+      "step": 1656
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.695934959349594e-06,
+      "loss": 1.0803,
+      "step": 1657
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.692682926829268e-06,
+      "loss": 0.9922,
+      "step": 1658
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.689430894308943e-06,
+      "loss": 0.8802,
+      "step": 1659
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.686178861788619e-06,
+      "loss": 0.9824,
+      "step": 1660
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.682926829268293e-06,
+      "loss": 1.05,
+      "step": 1661
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.679674796747968e-06,
+      "loss": 1.0207,
+      "step": 1662
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.6764227642276425e-06,
+      "loss": 1.1401,
+      "step": 1663
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.673170731707318e-06,
+      "loss": 0.9641,
+      "step": 1664
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.669918699186992e-06,
+      "loss": 1.0713,
+      "step": 1665
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 0.9847,
+      "step": 1666
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.663414634146342e-06,
+      "loss": 0.9872,
+      "step": 1667
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.660162601626017e-06,
+      "loss": 1.0622,
+      "step": 1668
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.656910569105692e-06,
+      "loss": 0.981,
+      "step": 1669
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.653658536585366e-06,
+      "loss": 1.0773,
+      "step": 1670
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.6504065040650406e-06,
+      "loss": 1.0735,
+      "step": 1671
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.647154471544716e-06,
+      "loss": 1.1425,
+      "step": 1672
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.643902439024391e-06,
+      "loss": 0.7145,
+      "step": 1673
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.640650406504065e-06,
+      "loss": 0.9577,
+      "step": 1674
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.6373983739837404e-06,
+      "loss": 1.041,
+      "step": 1675
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.0476,
+      "step": 1676
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.63089430894309e-06,
+      "loss": 0.9724,
+      "step": 1677
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.627642276422764e-06,
+      "loss": 1.0636,
+      "step": 1678
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.6243902439024395e-06,
+      "loss": 1.0343,
+      "step": 1679
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.621138211382114e-06,
+      "loss": 1.0782,
+      "step": 1680
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.617886178861789e-06,
+      "loss": 0.8986,
+      "step": 1681
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.614634146341464e-06,
+      "loss": 0.9133,
+      "step": 1682
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.6113821138211385e-06,
+      "loss": 1.0716,
+      "step": 1683
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.608130081300814e-06,
+      "loss": 1.0457,
+      "step": 1684
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.604878048780488e-06,
+      "loss": 0.9866,
+      "step": 1685
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.601626016260162e-06,
+      "loss": 1.1142,
+      "step": 1686
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.598373983739838e-06,
+      "loss": 0.9573,
+      "step": 1687
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.595121951219513e-06,
+      "loss": 0.8973,
+      "step": 1688
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.591869918699187e-06,
+      "loss": 1.0993,
+      "step": 1689
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.588617886178862e-06,
+      "loss": 1.1309,
+      "step": 1690
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.5853658536585375e-06,
+      "loss": 0.9295,
+      "step": 1691
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.582113821138212e-06,
+      "loss": 0.9254,
+      "step": 1692
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.578861788617886e-06,
+      "loss": 0.7652,
+      "step": 1693
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.575609756097561e-06,
+      "loss": 1.0433,
+      "step": 1694
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.572357723577236e-06,
+      "loss": 0.9052,
+      "step": 1695
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.569105691056911e-06,
+      "loss": 1.1401,
+      "step": 1696
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.565853658536586e-06,
+      "loss": 0.9253,
+      "step": 1697
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.56260162601626e-06,
+      "loss": 1.1499,
+      "step": 1698
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.559349593495936e-06,
+      "loss": 0.9237,
+      "step": 1699
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.55609756097561e-06,
+      "loss": 1.0417,
+      "step": 1700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 1.0393,
+      "step": 1701
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.5495934959349595e-06,
+      "loss": 1.2808,
+      "step": 1702
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.546341463414635e-06,
+      "loss": 0.8069,
+      "step": 1703
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.543089430894309e-06,
+      "loss": 1.243,
+      "step": 1704
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.539837398373984e-06,
+      "loss": 1.0154,
+      "step": 1705
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.536585365853659e-06,
+      "loss": 0.9058,
+      "step": 1706
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.1662,
+      "step": 1707
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.530081300813008e-06,
+      "loss": 1.1499,
+      "step": 1708
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.526829268292683e-06,
+      "loss": 0.8535,
+      "step": 1709
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.5235772357723576e-06,
+      "loss": 1.0625,
+      "step": 1710
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.520325203252033e-06,
+      "loss": 0.8479,
+      "step": 1711
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.517073170731708e-06,
+      "loss": 0.882,
+      "step": 1712
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.513821138211382e-06,
+      "loss": 0.9613,
+      "step": 1713
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.5105691056910575e-06,
+      "loss": 0.7801,
+      "step": 1714
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.507317073170732e-06,
+      "loss": 1.0748,
+      "step": 1715
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.504065040650407e-06,
+      "loss": 1.0605,
+      "step": 1716
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.500813008130081e-06,
+      "loss": 0.9567,
+      "step": 1717
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.4975609756097565e-06,
+      "loss": 1.1219,
+      "step": 1718
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.494308943089431e-06,
+      "loss": 1.0658,
+      "step": 1719
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.491056910569106e-06,
+      "loss": 0.8569,
+      "step": 1720
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.487804878048781e-06,
+      "loss": 1.1191,
+      "step": 1721
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.4845528455284556e-06,
+      "loss": 0.8976,
+      "step": 1722
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.481300813008131e-06,
+      "loss": 1.0538,
+      "step": 1723
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.478048780487805e-06,
+      "loss": 0.9879,
+      "step": 1724
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.47479674796748e-06,
+      "loss": 1.1799,
+      "step": 1725
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.471544715447155e-06,
+      "loss": 1.091,
+      "step": 1726
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.46829268292683e-06,
+      "loss": 1.0856,
+      "step": 1727
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.465040650406505e-06,
+      "loss": 1.077,
+      "step": 1728
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.461788617886179e-06,
+      "loss": 1.0406,
+      "step": 1729
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.4585365853658545e-06,
+      "loss": 1.0031,
+      "step": 1730
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.455284552845529e-06,
+      "loss": 0.9285,
+      "step": 1731
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.452032520325203e-06,
+      "loss": 1.0186,
+      "step": 1732
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.448780487804878e-06,
+      "loss": 1.0259,
+      "step": 1733
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.4455284552845535e-06,
+      "loss": 1.1161,
+      "step": 1734
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.442276422764228e-06,
+      "loss": 1.1398,
+      "step": 1735
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.439024390243903e-06,
+      "loss": 0.8897,
+      "step": 1736
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.435772357723578e-06,
+      "loss": 0.954,
+      "step": 1737
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.432520325203253e-06,
+      "loss": 1.0067,
+      "step": 1738
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.429268292682927e-06,
+      "loss": 0.9467,
+      "step": 1739
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.426016260162602e-06,
+      "loss": 1.0833,
+      "step": 1740
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.4227642276422765e-06,
+      "loss": 1.1594,
+      "step": 1741
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.419512195121952e-06,
+      "loss": 0.9732,
+      "step": 1742
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.416260162601627e-06,
+      "loss": 1.049,
+      "step": 1743
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.413008130081301e-06,
+      "loss": 0.8979,
+      "step": 1744
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.409756097560976e-06,
+      "loss": 1.1798,
+      "step": 1745
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.406504065040651e-06,
+      "loss": 1.0215,
+      "step": 1746
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.403252032520325e-06,
+      "loss": 1.1107,
+      "step": 1747
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.4e-06,
+      "loss": 0.8588,
+      "step": 1748
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.396747967479675e-06,
+      "loss": 0.9707,
+      "step": 1749
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.39349593495935e-06,
+      "loss": 0.9518,
+      "step": 1750
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 0.9148,
+      "step": 1751
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.3869918699187e-06,
+      "loss": 1.0662,
+      "step": 1752
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.3837398373983745e-06,
+      "loss": 0.9097,
+      "step": 1753
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.380487804878049e-06,
+      "loss": 0.8217,
+      "step": 1754
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.377235772357724e-06,
+      "loss": 0.889,
+      "step": 1755
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.373983739837398e-06,
+      "loss": 1.0726,
+      "step": 1756
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.3707317073170735e-06,
+      "loss": 0.9623,
+      "step": 1757
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.367479674796749e-06,
+      "loss": 0.9076,
+      "step": 1758
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.364227642276423e-06,
+      "loss": 0.9864,
+      "step": 1759
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.360975609756098e-06,
+      "loss": 1.1972,
+      "step": 1760
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.3577235772357726e-06,
+      "loss": 1.0512,
+      "step": 1761
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.354471544715447e-06,
+      "loss": 0.9183,
+      "step": 1762
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.351219512195122e-06,
+      "loss": 1.0082,
+      "step": 1763
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.347967479674797e-06,
+      "loss": 1.1348,
+      "step": 1764
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.344715447154472e-06,
+      "loss": 0.9026,
+      "step": 1765
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.341463414634147e-06,
+      "loss": 1.0871,
+      "step": 1766
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.338211382113822e-06,
+      "loss": 0.9296,
+      "step": 1767
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.334959349593496e-06,
+      "loss": 0.9988,
+      "step": 1768
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.331707317073171e-06,
+      "loss": 0.9448,
+      "step": 1769
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.328455284552846e-06,
+      "loss": 1.0465,
+      "step": 1770
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.32520325203252e-06,
+      "loss": 1.0305,
+      "step": 1771
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.321951219512195e-06,
+      "loss": 0.9429,
+      "step": 1772
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.3186991869918706e-06,
+      "loss": 1.0488,
+      "step": 1773
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.315447154471545e-06,
+      "loss": 1.0681,
+      "step": 1774
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.31219512195122e-06,
+      "loss": 0.9315,
+      "step": 1775
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.308943089430894e-06,
+      "loss": 1.0433,
+      "step": 1776
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.30569105691057e-06,
+      "loss": 1.0257,
+      "step": 1777
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.302439024390244e-06,
+      "loss": 0.966,
+      "step": 1778
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.299186991869919e-06,
+      "loss": 0.731,
+      "step": 1779
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.2959349593495935e-06,
+      "loss": 1.0005,
+      "step": 1780
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.292682926829269e-06,
+      "loss": 0.8389,
+      "step": 1781
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.289430894308944e-06,
+      "loss": 0.8862,
+      "step": 1782
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.286178861788618e-06,
+      "loss": 1.0339,
+      "step": 1783
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.282926829268293e-06,
+      "loss": 1.1874,
+      "step": 1784
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.279674796747968e-06,
+      "loss": 0.8794,
+      "step": 1785
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.276422764227642e-06,
+      "loss": 0.9229,
+      "step": 1786
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.273170731707317e-06,
+      "loss": 1.0549,
+      "step": 1787
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.269918699186992e-06,
+      "loss": 1.1458,
+      "step": 1788
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 0.9497,
+      "step": 1789
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.263414634146342e-06,
+      "loss": 1.089,
+      "step": 1790
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.260162601626016e-06,
+      "loss": 1.0583,
+      "step": 1791
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.2569105691056915e-06,
+      "loss": 1.023,
+      "step": 1792
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.253658536585366e-06,
+      "loss": 1.0398,
+      "step": 1793
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.250406504065041e-06,
+      "loss": 0.9563,
+      "step": 1794
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.247154471544716e-06,
+      "loss": 1.0675,
+      "step": 1795
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.2439024390243905e-06,
+      "loss": 1.1054,
+      "step": 1796
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.240650406504066e-06,
+      "loss": 1.0275,
+      "step": 1797
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.23739837398374e-06,
+      "loss": 0.9938,
+      "step": 1798
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.234146341463415e-06,
+      "loss": 1.1712,
+      "step": 1799
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.2308943089430896e-06,
+      "loss": 1.0511,
+      "step": 1800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.8939,
+      "step": 1801
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.224390243902439e-06,
+      "loss": 1.1,
+      "step": 1802
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.221138211382114e-06,
+      "loss": 1.1417,
+      "step": 1803
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.2178861788617895e-06,
+      "loss": 0.9785,
+      "step": 1804
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.214634146341464e-06,
+      "loss": 0.9913,
+      "step": 1805
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.211382113821139e-06,
+      "loss": 1.0716,
+      "step": 1806
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.208130081300813e-06,
+      "loss": 1.0369,
+      "step": 1807
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.204878048780488e-06,
+      "loss": 1.078,
+      "step": 1808
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.201626016260163e-06,
+      "loss": 1.0219,
+      "step": 1809
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.198373983739838e-06,
+      "loss": 0.7799,
+      "step": 1810
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.195121951219512e-06,
+      "loss": 0.7781,
+      "step": 1811
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1918699186991876e-06,
+      "loss": 1.0363,
+      "step": 1812
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.188617886178863e-06,
+      "loss": 1.0418,
+      "step": 1813
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.185365853658537e-06,
+      "loss": 1.1128,
+      "step": 1814
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.1821138211382114e-06,
+      "loss": 1.0048,
+      "step": 1815
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.178861788617887e-06,
+      "loss": 1.0371,
+      "step": 1816
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.175609756097561e-06,
+      "loss": 0.878,
+      "step": 1817
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.172357723577236e-06,
+      "loss": 1.0204,
+      "step": 1818
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.169105691056911e-06,
+      "loss": 1.2511,
+      "step": 1819
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.165853658536586e-06,
+      "loss": 1.1646,
+      "step": 1820
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.162601626016261e-06,
+      "loss": 0.9247,
+      "step": 1821
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.159349593495935e-06,
+      "loss": 1.1487,
+      "step": 1822
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.1560975609756095e-06,
+      "loss": 0.9759,
+      "step": 1823
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.152845528455285e-06,
+      "loss": 0.9489,
+      "step": 1824
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.14959349593496e-06,
+      "loss": 0.8965,
+      "step": 1825
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 0.879,
+      "step": 1826
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.1430894308943094e-06,
+      "loss": 1.0125,
+      "step": 1827
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.139837398373985e-06,
+      "loss": 0.8963,
+      "step": 1828
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.136585365853659e-06,
+      "loss": 1.1947,
+      "step": 1829
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 0.9801,
+      "step": 1830
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.1300813008130085e-06,
+      "loss": 1.0278,
+      "step": 1831
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.126829268292683e-06,
+      "loss": 1.0233,
+      "step": 1832
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.123577235772358e-06,
+      "loss": 0.8487,
+      "step": 1833
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.120325203252033e-06,
+      "loss": 0.8598,
+      "step": 1834
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.1170731707317075e-06,
+      "loss": 1.2199,
+      "step": 1835
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.113821138211383e-06,
+      "loss": 0.9292,
+      "step": 1836
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.110569105691057e-06,
+      "loss": 1.0013,
+      "step": 1837
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.107317073170731e-06,
+      "loss": 1.2117,
+      "step": 1838
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.1040650406504066e-06,
+      "loss": 1.1563,
+      "step": 1839
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.100813008130082e-06,
+      "loss": 1.182,
+      "step": 1840
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.097560975609756e-06,
+      "loss": 1.0429,
+      "step": 1841
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.094308943089431e-06,
+      "loss": 1.1435,
+      "step": 1842
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.0910569105691065e-06,
+      "loss": 0.9678,
+      "step": 1843
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.087804878048781e-06,
+      "loss": 0.9857,
+      "step": 1844
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.084552845528455e-06,
+      "loss": 1.0402,
+      "step": 1845
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.08130081300813e-06,
+      "loss": 0.9489,
+      "step": 1846
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.078048780487805e-06,
+      "loss": 1.0016,
+      "step": 1847
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.07479674796748e-06,
+      "loss": 1.0776,
+      "step": 1848
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.071544715447155e-06,
+      "loss": 0.9279,
+      "step": 1849
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.068292682926829e-06,
+      "loss": 0.9384,
+      "step": 1850
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 0.991,
+      "step": 1851
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.061788617886179e-06,
+      "loss": 1.1924,
+      "step": 1852
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.058536585365854e-06,
+      "loss": 1.0255,
+      "step": 1853
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.0552845528455284e-06,
+      "loss": 1.1402,
+      "step": 1854
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.052032520325204e-06,
+      "loss": 0.8838,
+      "step": 1855
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.048780487804879e-06,
+      "loss": 1.2136,
+      "step": 1856
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.045528455284553e-06,
+      "loss": 1.0435,
+      "step": 1857
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.042276422764228e-06,
+      "loss": 0.917,
+      "step": 1858
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.039024390243903e-06,
+      "loss": 0.932,
+      "step": 1859
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.035772357723578e-06,
+      "loss": 1.134,
+      "step": 1860
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.032520325203252e-06,
+      "loss": 1.0444,
+      "step": 1861
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.029268292682927e-06,
+      "loss": 1.0635,
+      "step": 1862
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.026016260162602e-06,
+      "loss": 0.9388,
+      "step": 1863
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.022764227642277e-06,
+      "loss": 0.9281,
+      "step": 1864
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.019512195121952e-06,
+      "loss": 0.8939,
+      "step": 1865
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.0162601626016264e-06,
+      "loss": 0.8657,
+      "step": 1866
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.013008130081301e-06,
+      "loss": 0.9531,
+      "step": 1867
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.009756097560976e-06,
+      "loss": 0.7936,
+      "step": 1868
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.00650406504065e-06,
+      "loss": 0.9092,
+      "step": 1869
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.0032520325203255e-06,
+      "loss": 0.9876,
+      "step": 1870
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6736,
+      "step": 1871
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.996747967479675e-06,
+      "loss": 1.0525,
+      "step": 1872
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.99349593495935e-06,
+      "loss": 1.0019,
+      "step": 1873
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.9902439024390245e-06,
+      "loss": 1.1189,
+      "step": 1874
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.9869918699187e-06,
+      "loss": 0.8656,
+      "step": 1875
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.983739837398374e-06,
+      "loss": 0.9236,
+      "step": 1876
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.980487804878049e-06,
+      "loss": 0.935,
+      "step": 1877
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.977235772357724e-06,
+      "loss": 0.9972,
+      "step": 1878
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.973983739837399e-06,
+      "loss": 0.8999,
+      "step": 1879
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.970731707317074e-06,
+      "loss": 0.9624,
+      "step": 1880
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.967479674796748e-06,
+      "loss": 1.0351,
+      "step": 1881
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.9642276422764235e-06,
+      "loss": 1.0313,
+      "step": 1882
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.960975609756098e-06,
+      "loss": 0.9764,
+      "step": 1883
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.957723577235772e-06,
+      "loss": 0.8969,
+      "step": 1884
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.954471544715447e-06,
+      "loss": 0.9615,
+      "step": 1885
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.9512195121951225e-06,
+      "loss": 0.8641,
+      "step": 1886
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.947967479674797e-06,
+      "loss": 0.8014,
+      "step": 1887
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.944715447154472e-06,
+      "loss": 1.0481,
+      "step": 1888
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.941463414634147e-06,
+      "loss": 0.9914,
+      "step": 1889
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.9382113821138216e-06,
+      "loss": 0.998,
+      "step": 1890
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.934959349593496e-06,
+      "loss": 0.9976,
+      "step": 1891
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.931707317073171e-06,
+      "loss": 0.8197,
+      "step": 1892
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.9284552845528454e-06,
+      "loss": 0.9526,
+      "step": 1893
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.925203252032521e-06,
+      "loss": 0.8514,
+      "step": 1894
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.921951219512196e-06,
+      "loss": 0.8903,
+      "step": 1895
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.91869918699187e-06,
+      "loss": 0.8912,
+      "step": 1896
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.915447154471545e-06,
+      "loss": 0.9688,
+      "step": 1897
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.91219512195122e-06,
+      "loss": 1.0551,
+      "step": 1898
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.908943089430894e-06,
+      "loss": 0.9609,
+      "step": 1899
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.905691056910569e-06,
+      "loss": 0.9888,
+      "step": 1900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.035,
+      "step": 1901
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.899186991869919e-06,
+      "loss": 0.9564,
+      "step": 1902
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.895934959349594e-06,
+      "loss": 0.9267,
+      "step": 1903
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.892682926829269e-06,
+      "loss": 0.8916,
+      "step": 1904
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.8894308943089434e-06,
+      "loss": 0.8148,
+      "step": 1905
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.886178861788618e-06,
+      "loss": 0.9783,
+      "step": 1906
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.882926829268293e-06,
+      "loss": 0.7659,
+      "step": 1907
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.879674796747967e-06,
+      "loss": 1.0616,
+      "step": 1908
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.8764227642276425e-06,
+      "loss": 0.9694,
+      "step": 1909
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.873170731707318e-06,
+      "loss": 0.851,
+      "step": 1910
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.869918699186992e-06,
+      "loss": 0.8441,
+      "step": 1911
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 0.776,
+      "step": 1912
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.8634146341463415e-06,
+      "loss": 0.8359,
+      "step": 1913
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.860162601626017e-06,
+      "loss": 0.7946,
+      "step": 1914
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.856910569105691e-06,
+      "loss": 1.1071,
+      "step": 1915
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.853658536585366e-06,
+      "loss": 0.9086,
+      "step": 1916
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.8504065040650414e-06,
+      "loss": 1.1,
+      "step": 1917
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.847154471544716e-06,
+      "loss": 0.8813,
+      "step": 1918
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.843902439024391e-06,
+      "loss": 0.847,
+      "step": 1919
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.840650406504065e-06,
+      "loss": 0.9163,
+      "step": 1920
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.83739837398374e-06,
+      "loss": 0.7906,
+      "step": 1921
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.834146341463415e-06,
+      "loss": 0.8899,
+      "step": 1922
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.83089430894309e-06,
+      "loss": 0.9328,
+      "step": 1923
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.827642276422764e-06,
+      "loss": 0.9875,
+      "step": 1924
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.8243902439024395e-06,
+      "loss": 1.0501,
+      "step": 1925
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.821138211382115e-06,
+      "loss": 0.98,
+      "step": 1926
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.817886178861789e-06,
+      "loss": 0.8235,
+      "step": 1927
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.814634146341464e-06,
+      "loss": 1.1655,
+      "step": 1928
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.8113821138211386e-06,
+      "loss": 0.8592,
+      "step": 1929
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.8081300813008133e-06,
+      "loss": 0.7341,
+      "step": 1930
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.804878048780488e-06,
+      "loss": 0.9262,
+      "step": 1931
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.8016260162601633e-06,
+      "loss": 0.883,
+      "step": 1932
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.7983739837398376e-06,
+      "loss": 0.8786,
+      "step": 1933
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.7951219512195124e-06,
+      "loss": 0.9604,
+      "step": 1934
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.7918699186991876e-06,
+      "loss": 0.7019,
+      "step": 1935
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.788617886178862e-06,
+      "loss": 0.8081,
+      "step": 1936
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.785365853658537e-06,
+      "loss": 0.6996,
+      "step": 1937
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.782113821138212e-06,
+      "loss": 1.0248,
+      "step": 1938
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.778861788617886e-06,
+      "loss": 0.8878,
+      "step": 1939
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.7756097560975614e-06,
+      "loss": 0.9386,
+      "step": 1940
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.772357723577236e-06,
+      "loss": 0.9676,
+      "step": 1941
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.769105691056911e-06,
+      "loss": 1.0035,
+      "step": 1942
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.7658536585365857e-06,
+      "loss": 0.793,
+      "step": 1943
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.762601626016261e-06,
+      "loss": 1.1203,
+      "step": 1944
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.759349593495935e-06,
+      "loss": 0.8483,
+      "step": 1945
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.75609756097561e-06,
+      "loss": 0.8529,
+      "step": 1946
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.752845528455285e-06,
+      "loss": 0.8823,
+      "step": 1947
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.7495934959349595e-06,
+      "loss": 1.0097,
+      "step": 1948
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.7463414634146343e-06,
+      "loss": 0.9337,
+      "step": 1949
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.7430894308943094e-06,
+      "loss": 0.9409,
+      "step": 1950
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.7398373983739838e-06,
+      "loss": 0.8907,
+      "step": 1951
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.736585365853659e-06,
+      "loss": 1.0337,
+      "step": 1952
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.8336,
+      "step": 1953
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.730081300813008e-06,
+      "loss": 0.9601,
+      "step": 1954
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.7268292682926833e-06,
+      "loss": 1.0177,
+      "step": 1955
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.723577235772358e-06,
+      "loss": 0.8697,
+      "step": 1956
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.7203252032520328e-06,
+      "loss": 0.7707,
+      "step": 1957
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.7170731707317075e-06,
+      "loss": 0.7886,
+      "step": 1958
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.7138211382113827e-06,
+      "loss": 0.6953,
+      "step": 1959
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.710569105691057e-06,
+      "loss": 0.9908,
+      "step": 1960
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.707317073170732e-06,
+      "loss": 1.064,
+      "step": 1961
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.704065040650407e-06,
+      "loss": 0.909,
+      "step": 1962
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.7008130081300814e-06,
+      "loss": 0.8856,
+      "step": 1963
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.6975609756097565e-06,
+      "loss": 0.9831,
+      "step": 1964
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.6943089430894313e-06,
+      "loss": 0.7093,
+      "step": 1965
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.6910569105691056e-06,
+      "loss": 0.8911,
+      "step": 1966
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.687804878048781e-06,
+      "loss": 0.8979,
+      "step": 1967
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.6845528455284556e-06,
+      "loss": 0.8008,
+      "step": 1968
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.68130081300813e-06,
+      "loss": 0.8355,
+      "step": 1969
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.678048780487805e-06,
+      "loss": 0.9001,
+      "step": 1970
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.6747967479674803e-06,
+      "loss": 0.8143,
+      "step": 1971
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.6715447154471546e-06,
+      "loss": 0.9599,
+      "step": 1972
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.6682926829268294e-06,
+      "loss": 1.0918,
+      "step": 1973
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.6650406504065046e-06,
+      "loss": 0.8885,
+      "step": 1974
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.661788617886179e-06,
+      "loss": 0.9461,
+      "step": 1975
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 0.9291,
+      "step": 1976
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.655284552845529e-06,
+      "loss": 0.9345,
+      "step": 1977
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.6520325203252032e-06,
+      "loss": 0.7848,
+      "step": 1978
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.6487804878048784e-06,
+      "loss": 0.9142,
+      "step": 1979
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.645528455284553e-06,
+      "loss": 0.8617,
+      "step": 1980
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.6422764227642283e-06,
+      "loss": 0.89,
+      "step": 1981
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.6390243902439027e-06,
+      "loss": 1.0352,
+      "step": 1982
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.6357723577235775e-06,
+      "loss": 0.7762,
+      "step": 1983
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.6325203252032526e-06,
+      "loss": 0.841,
+      "step": 1984
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.629268292682927e-06,
+      "loss": 1.2,
+      "step": 1985
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.626016260162602e-06,
+      "loss": 1.0145,
+      "step": 1986
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.622764227642277e-06,
+      "loss": 0.9831,
+      "step": 1987
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.6195121951219513e-06,
+      "loss": 0.9194,
+      "step": 1988
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.6162601626016264e-06,
+      "loss": 0.824,
+      "step": 1989
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.613008130081301e-06,
+      "loss": 1.1369,
+      "step": 1990
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.609756097560976e-06,
+      "loss": 0.9234,
+      "step": 1991
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.6065040650406507e-06,
+      "loss": 1.0569,
+      "step": 1992
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.603252032520326e-06,
+      "loss": 1.0965,
+      "step": 1993
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.8484,
+      "step": 1994
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.596747967479675e-06,
+      "loss": 0.8975,
+      "step": 1995
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.59349593495935e-06,
+      "loss": 0.9968,
+      "step": 1996
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.5902439024390245e-06,
+      "loss": 0.9823,
+      "step": 1997
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.5869918699186993e-06,
+      "loss": 0.9228,
+      "step": 1998
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.5837398373983745e-06,
+      "loss": 1.0516,
+      "step": 1999
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.580487804878049e-06,
+      "loss": 1.0103,
+      "step": 2000
+    },
+    {
+      "epoch": 6.44,
+      "eval_loss": 1.8504787683486938,
+      "eval_runtime": 136.1332,
+      "eval_samples_per_second": 4.452,
+      "eval_steps_per_second": 4.452,
+      "step": 2000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.9511,
+      "step": 2001
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.5739837398373988e-06,
+      "loss": 0.9275,
+      "step": 2002
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.570731707317073e-06,
+      "loss": 0.8369,
+      "step": 2003
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.5674796747967483e-06,
+      "loss": 0.9894,
+      "step": 2004
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.564227642276423e-06,
+      "loss": 0.9381,
+      "step": 2005
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.560975609756098e-06,
+      "loss": 1.0664,
+      "step": 2006
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.5577235772357726e-06,
+      "loss": 0.9109,
+      "step": 2007
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.5544715447154478e-06,
+      "loss": 0.8327,
+      "step": 2008
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.551219512195122e-06,
+      "loss": 1.0944,
+      "step": 2009
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.547967479674797e-06,
+      "loss": 0.8797,
+      "step": 2010
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.544715447154472e-06,
+      "loss": 1.0077,
+      "step": 2011
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.5414634146341464e-06,
+      "loss": 1.0638,
+      "step": 2012
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.5382113821138216e-06,
+      "loss": 0.9856,
+      "step": 2013
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.5349593495934964e-06,
+      "loss": 0.8515,
+      "step": 2014
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.5317073170731707e-06,
+      "loss": 0.8652,
+      "step": 2015
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.528455284552846e-06,
+      "loss": 1.0582,
+      "step": 2016
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.5252032520325206e-06,
+      "loss": 0.9842,
+      "step": 2017
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.5219512195121954e-06,
+      "loss": 0.9431,
+      "step": 2018
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.51869918699187e-06,
+      "loss": 0.9628,
+      "step": 2019
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.5154471544715454e-06,
+      "loss": 1.0198,
+      "step": 2020
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.5121951219512197e-06,
+      "loss": 0.889,
+      "step": 2021
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.5089430894308945e-06,
+      "loss": 0.8974,
+      "step": 2022
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.5056910569105696e-06,
+      "loss": 1.1061,
+      "step": 2023
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.502439024390244e-06,
+      "loss": 0.8784,
+      "step": 2024
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.4991869918699187e-06,
+      "loss": 0.8681,
+      "step": 2025
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.495934959349594e-06,
+      "loss": 1.0441,
+      "step": 2026
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.4926829268292683e-06,
+      "loss": 0.9156,
+      "step": 2027
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.4894308943089435e-06,
+      "loss": 0.9789,
+      "step": 2028
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.4861788617886182e-06,
+      "loss": 0.9505,
+      "step": 2029
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.4829268292682926e-06,
+      "loss": 0.8413,
+      "step": 2030
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.4796747967479677e-06,
+      "loss": 0.8392,
+      "step": 2031
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.4764227642276425e-06,
+      "loss": 1.018,
+      "step": 2032
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.4731707317073173e-06,
+      "loss": 0.8463,
+      "step": 2033
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.469918699186992e-06,
+      "loss": 0.9816,
+      "step": 2034
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.1267,
+      "step": 2035
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.4634146341463416e-06,
+      "loss": 1.2423,
+      "step": 2036
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.4601626016260163e-06,
+      "loss": 1.0274,
+      "step": 2037
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.4569105691056915e-06,
+      "loss": 1.056,
+      "step": 2038
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.453658536585366e-06,
+      "loss": 0.7852,
+      "step": 2039
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.450406504065041e-06,
+      "loss": 0.7867,
+      "step": 2040
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.447154471544716e-06,
+      "loss": 0.8754,
+      "step": 2041
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.44390243902439e-06,
+      "loss": 0.861,
+      "step": 2042
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.4406504065040653e-06,
+      "loss": 1.0673,
+      "step": 2043
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.43739837398374e-06,
+      "loss": 0.8535,
+      "step": 2044
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.4341463414634153e-06,
+      "loss": 0.876,
+      "step": 2045
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.4308943089430896e-06,
+      "loss": 0.6934,
+      "step": 2046
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.427642276422765e-06,
+      "loss": 0.7953,
+      "step": 2047
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.4243902439024396e-06,
+      "loss": 1.0014,
+      "step": 2048
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.421138211382114e-06,
+      "loss": 0.9898,
+      "step": 2049
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.417886178861789e-06,
+      "loss": 0.9181,
+      "step": 2050
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.9409,
+      "step": 2051
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.411382113821138e-06,
+      "loss": 0.9135,
+      "step": 2052
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.4081300813008134e-06,
+      "loss": 0.9503,
+      "step": 2053
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.404878048780488e-06,
+      "loss": 0.9767,
+      "step": 2054
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.401626016260163e-06,
+      "loss": 0.8891,
+      "step": 2055
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.3983739837398377e-06,
+      "loss": 0.8357,
+      "step": 2056
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.395121951219513e-06,
+      "loss": 0.9624,
+      "step": 2057
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.391869918699187e-06,
+      "loss": 1.0297,
+      "step": 2058
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.388617886178862e-06,
+      "loss": 0.8822,
+      "step": 2059
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.385365853658537e-06,
+      "loss": 0.9762,
+      "step": 2060
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.3821138211382115e-06,
+      "loss": 0.7808,
+      "step": 2061
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.3788617886178866e-06,
+      "loss": 0.9069,
+      "step": 2062
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.3756097560975614e-06,
+      "loss": 1.121,
+      "step": 2063
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.3723577235772358e-06,
+      "loss": 1.0017,
+      "step": 2064
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.369105691056911e-06,
+      "loss": 1.0012,
+      "step": 2065
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.3658536585365857e-06,
+      "loss": 0.8257,
+      "step": 2066
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.3626016260162605e-06,
+      "loss": 0.747,
+      "step": 2067
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3593495934959352e-06,
+      "loss": 1.0732,
+      "step": 2068
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3560975609756104e-06,
+      "loss": 1.03,
+      "step": 2069
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.3528455284552847e-06,
+      "loss": 0.9261,
+      "step": 2070
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3495934959349595e-06,
+      "loss": 0.9237,
+      "step": 2071
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.3463414634146347e-06,
+      "loss": 0.9019,
+      "step": 2072
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.343089430894309e-06,
+      "loss": 0.8419,
+      "step": 2073
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.3398373983739842e-06,
+      "loss": 0.7351,
+      "step": 2074
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.336585365853659e-06,
+      "loss": 0.7765,
+      "step": 2075
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.8724,
+      "step": 2076
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3300813008130085e-06,
+      "loss": 0.9896,
+      "step": 2077
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3268292682926833e-06,
+      "loss": 1.0152,
+      "step": 2078
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.3235772357723576e-06,
+      "loss": 0.9131,
+      "step": 2079
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.320325203252033e-06,
+      "loss": 0.8974,
+      "step": 2080
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.3170731707317076e-06,
+      "loss": 0.993,
+      "step": 2081
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.3138211382113823e-06,
+      "loss": 0.9583,
+      "step": 2082
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.310569105691057e-06,
+      "loss": 1.0108,
+      "step": 2083
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.3073170731707323e-06,
+      "loss": 0.8679,
+      "step": 2084
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.3040650406504066e-06,
+      "loss": 0.8467,
+      "step": 2085
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.3008130081300814e-06,
+      "loss": 0.8583,
+      "step": 2086
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.2975609756097566e-06,
+      "loss": 1.0166,
+      "step": 2087
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.294308943089431e-06,
+      "loss": 1.0442,
+      "step": 2088
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.291056910569106e-06,
+      "loss": 0.8316,
+      "step": 2089
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.287804878048781e-06,
+      "loss": 0.9543,
+      "step": 2090
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.284552845528455e-06,
+      "loss": 0.8314,
+      "step": 2091
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.2813008130081304e-06,
+      "loss": 0.8931,
+      "step": 2092
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.278048780487805e-06,
+      "loss": 0.8846,
+      "step": 2093
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.27479674796748e-06,
+      "loss": 0.852,
+      "step": 2094
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.2715447154471547e-06,
+      "loss": 0.9743,
+      "step": 2095
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.26829268292683e-06,
+      "loss": 0.8912,
+      "step": 2096
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.265040650406504e-06,
+      "loss": 0.7174,
+      "step": 2097
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.261788617886179e-06,
+      "loss": 1.0874,
+      "step": 2098
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.258536585365854e-06,
+      "loss": 0.941,
+      "step": 2099
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.2552845528455285e-06,
+      "loss": 0.7706,
+      "step": 2100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.9006,
+      "step": 2101
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.2487804878048784e-06,
+      "loss": 0.8029,
+      "step": 2102
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.2455284552845528e-06,
+      "loss": 0.9445,
+      "step": 2103
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.242276422764228e-06,
+      "loss": 0.8377,
+      "step": 2104
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.2390243902439027e-06,
+      "loss": 0.977,
+      "step": 2105
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.235772357723577e-06,
+      "loss": 0.9464,
+      "step": 2106
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.2325203252032522e-06,
+      "loss": 0.8617,
+      "step": 2107
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.229268292682927e-06,
+      "loss": 0.9309,
+      "step": 2108
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.226016260162602e-06,
+      "loss": 0.962,
+      "step": 2109
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.2227642276422765e-06,
+      "loss": 0.9524,
+      "step": 2110
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.2195121951219517e-06,
+      "loss": 1.1001,
+      "step": 2111
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.2162601626016265e-06,
+      "loss": 0.9124,
+      "step": 2112
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.213008130081301e-06,
+      "loss": 1.0752,
+      "step": 2113
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.209756097560976e-06,
+      "loss": 0.9587,
+      "step": 2114
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.2065040650406508e-06,
+      "loss": 0.8241,
+      "step": 2115
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.2032520325203255e-06,
+      "loss": 0.9492,
+      "step": 2116
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.834,
+      "step": 2117
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.1967479674796755e-06,
+      "loss": 0.9151,
+      "step": 2118
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.19349593495935e-06,
+      "loss": 0.8899,
+      "step": 2119
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.1902439024390246e-06,
+      "loss": 1.1593,
+      "step": 2120
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.1869918699186998e-06,
+      "loss": 0.8879,
+      "step": 2121
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.183739837398374e-06,
+      "loss": 1.049,
+      "step": 2122
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.1804878048780493e-06,
+      "loss": 0.7427,
+      "step": 2123
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.177235772357724e-06,
+      "loss": 1.1888,
+      "step": 2124
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.1739837398373984e-06,
+      "loss": 0.9002,
+      "step": 2125
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 0.9241,
+      "step": 2126
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.1674796747967483e-06,
+      "loss": 0.9225,
+      "step": 2127
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.1642276422764227e-06,
+      "loss": 0.8637,
+      "step": 2128
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.160975609756098e-06,
+      "loss": 0.8239,
+      "step": 2129
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.157723577235773e-06,
+      "loss": 0.8676,
+      "step": 2130
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.1544715447154474e-06,
+      "loss": 0.9477,
+      "step": 2131
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.151219512195122e-06,
+      "loss": 0.8552,
+      "step": 2132
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.1479674796747973e-06,
+      "loss": 1.0986,
+      "step": 2133
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.1447154471544717e-06,
+      "loss": 0.9442,
+      "step": 2134
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.1414634146341464e-06,
+      "loss": 0.8476,
+      "step": 2135
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.1382113821138216e-06,
+      "loss": 0.9852,
+      "step": 2136
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.134959349593496e-06,
+      "loss": 0.8168,
+      "step": 2137
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.131707317073171e-06,
+      "loss": 0.905,
+      "step": 2138
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.128455284552846e-06,
+      "loss": 0.8858,
+      "step": 2139
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.1252032520325202e-06,
+      "loss": 0.8909,
+      "step": 2140
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.1219512195121954e-06,
+      "loss": 0.9782,
+      "step": 2141
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.11869918699187e-06,
+      "loss": 0.9283,
+      "step": 2142
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.115447154471545e-06,
+      "loss": 0.7736,
+      "step": 2143
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.1121951219512197e-06,
+      "loss": 1.0487,
+      "step": 2144
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.108943089430895e-06,
+      "loss": 1.0196,
+      "step": 2145
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.1056910569105692e-06,
+      "loss": 0.9465,
+      "step": 2146
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.102439024390244e-06,
+      "loss": 0.9735,
+      "step": 2147
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.099186991869919e-06,
+      "loss": 0.9101,
+      "step": 2148
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.0959349593495935e-06,
+      "loss": 0.9809,
+      "step": 2149
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.0926829268292687e-06,
+      "loss": 0.7576,
+      "step": 2150
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.0894308943089435e-06,
+      "loss": 0.8896,
+      "step": 2151
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.086178861788618e-06,
+      "loss": 1.0485,
+      "step": 2152
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.082926829268293e-06,
+      "loss": 0.8838,
+      "step": 2153
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.0796747967479678e-06,
+      "loss": 0.9711,
+      "step": 2154
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.076422764227642e-06,
+      "loss": 0.7102,
+      "step": 2155
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.0731707317073173e-06,
+      "loss": 0.9466,
+      "step": 2156
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.069918699186992e-06,
+      "loss": 0.9156,
+      "step": 2157
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.066666666666667e-06,
+      "loss": 0.8512,
+      "step": 2158
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.0634146341463416e-06,
+      "loss": 0.6947,
+      "step": 2159
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.0601626016260168e-06,
+      "loss": 1.0756,
+      "step": 2160
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.056910569105691e-06,
+      "loss": 0.9478,
+      "step": 2161
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.053658536585366e-06,
+      "loss": 0.942,
+      "step": 2162
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.050406504065041e-06,
+      "loss": 0.9737,
+      "step": 2163
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0471544715447154e-06,
+      "loss": 0.8993,
+      "step": 2164
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0439024390243906e-06,
+      "loss": 0.9142,
+      "step": 2165
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0406504065040653e-06,
+      "loss": 1.1502,
+      "step": 2166
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0373983739837397e-06,
+      "loss": 0.8182,
+      "step": 2167
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.034146341463415e-06,
+      "loss": 0.8502,
+      "step": 2168
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.0308943089430896e-06,
+      "loss": 0.9923,
+      "step": 2169
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.0276422764227644e-06,
+      "loss": 0.7758,
+      "step": 2170
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.024390243902439e-06,
+      "loss": 0.8192,
+      "step": 2171
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.0211382113821143e-06,
+      "loss": 0.9336,
+      "step": 2172
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.017886178861789e-06,
+      "loss": 0.8563,
+      "step": 2173
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.0146341463414634e-06,
+      "loss": 0.8811,
+      "step": 2174
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.0113821138211386e-06,
+      "loss": 0.8534,
+      "step": 2175
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.0081300813008134e-06,
+      "loss": 0.8641,
+      "step": 2176
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.004878048780488e-06,
+      "loss": 0.8024,
+      "step": 2177
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.001626016260163e-06,
+      "loss": 0.9066,
+      "step": 2178
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.998373983739838e-06,
+      "loss": 0.8098,
+      "step": 2179
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.9951219512195124e-06,
+      "loss": 0.8276,
+      "step": 2180
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.991869918699187e-06,
+      "loss": 0.8935,
+      "step": 2181
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.9886178861788624e-06,
+      "loss": 0.8951,
+      "step": 2182
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.9853658536585367e-06,
+      "loss": 0.751,
+      "step": 2183
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.9821138211382115e-06,
+      "loss": 0.7723,
+      "step": 2184
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.9788617886178867e-06,
+      "loss": 0.8516,
+      "step": 2185
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.975609756097561e-06,
+      "loss": 0.8107,
+      "step": 2186
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.972357723577236e-06,
+      "loss": 0.9058,
+      "step": 2187
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.969105691056911e-06,
+      "loss": 0.8259,
+      "step": 2188
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.9658536585365853e-06,
+      "loss": 0.7877,
+      "step": 2189
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.9626016260162605e-06,
+      "loss": 0.8705,
+      "step": 2190
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.9593495934959352e-06,
+      "loss": 0.8533,
+      "step": 2191
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.95609756097561e-06,
+      "loss": 0.7963,
+      "step": 2192
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.9528455284552848e-06,
+      "loss": 0.8521,
+      "step": 2193
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.94959349593496e-06,
+      "loss": 0.8532,
+      "step": 2194
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.9463414634146343e-06,
+      "loss": 0.7695,
+      "step": 2195
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.943089430894309e-06,
+      "loss": 0.7606,
+      "step": 2196
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.9398373983739842e-06,
+      "loss": 0.7434,
+      "step": 2197
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.9365853658536586e-06,
+      "loss": 0.8957,
+      "step": 2198
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 0.7534,
+      "step": 2199
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.9300813008130085e-06,
+      "loss": 0.8752,
+      "step": 2200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.8731,
+      "step": 2201
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.923577235772358e-06,
+      "loss": 0.9121,
+      "step": 2202
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.920325203252033e-06,
+      "loss": 0.9779,
+      "step": 2203
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.9170731707317076e-06,
+      "loss": 0.7358,
+      "step": 2204
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.9138211382113823e-06,
+      "loss": 0.855,
+      "step": 2205
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.9105691056910575e-06,
+      "loss": 0.7898,
+      "step": 2206
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.907317073170732e-06,
+      "loss": 0.8852,
+      "step": 2207
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.9040650406504066e-06,
+      "loss": 0.9402,
+      "step": 2208
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.900813008130082e-06,
+      "loss": 0.7951,
+      "step": 2209
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.897560975609756e-06,
+      "loss": 0.9656,
+      "step": 2210
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.894308943089431e-06,
+      "loss": 0.8592,
+      "step": 2211
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.891056910569106e-06,
+      "loss": 0.7763,
+      "step": 2212
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.8878048780487804e-06,
+      "loss": 0.9697,
+      "step": 2213
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.8845528455284556e-06,
+      "loss": 0.7119,
+      "step": 2214
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.8813008130081304e-06,
+      "loss": 0.9611,
+      "step": 2215
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.8780487804878047e-06,
+      "loss": 0.9157,
+      "step": 2216
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.87479674796748e-06,
+      "loss": 0.82,
+      "step": 2217
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.8715447154471547e-06,
+      "loss": 0.7656,
+      "step": 2218
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.8682926829268294e-06,
+      "loss": 0.8231,
+      "step": 2219
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.865040650406504e-06,
+      "loss": 0.8933,
+      "step": 2220
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.8617886178861794e-06,
+      "loss": 0.8709,
+      "step": 2221
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.8585365853658537e-06,
+      "loss": 0.7575,
+      "step": 2222
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.8552845528455285e-06,
+      "loss": 0.9114,
+      "step": 2223
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.8520325203252037e-06,
+      "loss": 0.8967,
+      "step": 2224
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.848780487804878e-06,
+      "loss": 0.7845,
+      "step": 2225
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.845528455284553e-06,
+      "loss": 0.9071,
+      "step": 2226
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.842276422764228e-06,
+      "loss": 0.7849,
+      "step": 2227
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.8390243902439023e-06,
+      "loss": 0.9459,
+      "step": 2228
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.8357723577235775e-06,
+      "loss": 0.7767,
+      "step": 2229
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.8325203252032522e-06,
+      "loss": 0.8472,
+      "step": 2230
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.8292682926829266e-06,
+      "loss": 0.8758,
+      "step": 2231
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.8260162601626018e-06,
+      "loss": 0.8213,
+      "step": 2232
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.822764227642277e-06,
+      "loss": 0.7413,
+      "step": 2233
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.8195121951219513e-06,
+      "loss": 0.8647,
+      "step": 2234
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.816260162601626e-06,
+      "loss": 0.7746,
+      "step": 2235
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.8130081300813012e-06,
+      "loss": 0.9917,
+      "step": 2236
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.809756097560976e-06,
+      "loss": 0.8488,
+      "step": 2237
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.8065040650406503e-06,
+      "loss": 1.0255,
+      "step": 2238
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.8032520325203255e-06,
+      "loss": 0.97,
+      "step": 2239
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.7203,
+      "step": 2240
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.796747967479675e-06,
+      "loss": 0.9577,
+      "step": 2241
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.79349593495935e-06,
+      "loss": 0.8545,
+      "step": 2242
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.790243902439025e-06,
+      "loss": 0.7536,
+      "step": 2243
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.7869918699186993e-06,
+      "loss": 0.7657,
+      "step": 2244
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.783739837398374e-06,
+      "loss": 0.7966,
+      "step": 2245
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.7804878048780493e-06,
+      "loss": 0.9121,
+      "step": 2246
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.7772357723577236e-06,
+      "loss": 0.8049,
+      "step": 2247
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.773983739837399e-06,
+      "loss": 0.8254,
+      "step": 2248
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.7707317073170736e-06,
+      "loss": 0.8531,
+      "step": 2249
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.767479674796748e-06,
+      "loss": 0.6721,
+      "step": 2250
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.764227642276423e-06,
+      "loss": 0.9438,
+      "step": 2251
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.760975609756098e-06,
+      "loss": 0.9089,
+      "step": 2252
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.7577235772357726e-06,
+      "loss": 0.9019,
+      "step": 2253
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.7544715447154474e-06,
+      "loss": 0.961,
+      "step": 2254
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.7512195121951226e-06,
+      "loss": 0.9087,
+      "step": 2255
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.747967479674797e-06,
+      "loss": 0.8563,
+      "step": 2256
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.7447154471544717e-06,
+      "loss": 0.9439,
+      "step": 2257
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.741463414634147e-06,
+      "loss": 0.8596,
+      "step": 2258
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.738211382113821e-06,
+      "loss": 0.8567,
+      "step": 2259
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.7349593495934964e-06,
+      "loss": 0.9573,
+      "step": 2260
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.731707317073171e-06,
+      "loss": 0.7164,
+      "step": 2261
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.7284552845528455e-06,
+      "loss": 0.8552,
+      "step": 2262
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.7252032520325207e-06,
+      "loss": 0.8396,
+      "step": 2263
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.7219512195121954e-06,
+      "loss": 1.0325,
+      "step": 2264
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.7186991869918698e-06,
+      "loss": 0.8955,
+      "step": 2265
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.715447154471545e-06,
+      "loss": 0.9125,
+      "step": 2266
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.7121951219512197e-06,
+      "loss": 1.0099,
+      "step": 2267
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.7089430894308945e-06,
+      "loss": 0.8683,
+      "step": 2268
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.7056910569105693e-06,
+      "loss": 0.8666,
+      "step": 2269
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.7024390243902444e-06,
+      "loss": 0.8622,
+      "step": 2270
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.6991869918699188e-06,
+      "loss": 0.7621,
+      "step": 2271
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.6959349593495935e-06,
+      "loss": 0.821,
+      "step": 2272
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.6926829268292687e-06,
+      "loss": 0.9121,
+      "step": 2273
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.689430894308943e-06,
+      "loss": 0.7924,
+      "step": 2274
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.6861788617886183e-06,
+      "loss": 0.7761,
+      "step": 2275
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 0.703,
+      "step": 2276
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.6796747967479674e-06,
+      "loss": 0.8808,
+      "step": 2277
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.6764227642276425e-06,
+      "loss": 0.654,
+      "step": 2278
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.6731707317073173e-06,
+      "loss": 0.9136,
+      "step": 2279
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.669918699186992e-06,
+      "loss": 0.9102,
+      "step": 2280
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.7751,
+      "step": 2281
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.663414634146342e-06,
+      "loss": 0.8172,
+      "step": 2282
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.6601626016260164e-06,
+      "loss": 0.9678,
+      "step": 2283
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.656910569105691e-06,
+      "loss": 0.8708,
+      "step": 2284
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.6536585365853663e-06,
+      "loss": 0.8476,
+      "step": 2285
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.6504065040650406e-06,
+      "loss": 0.7699,
+      "step": 2286
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.6471544715447154e-06,
+      "loss": 0.8181,
+      "step": 2287
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.6439024390243906e-06,
+      "loss": 0.8092,
+      "step": 2288
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.640650406504065e-06,
+      "loss": 0.7167,
+      "step": 2289
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.63739837398374e-06,
+      "loss": 0.8349,
+      "step": 2290
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.634146341463415e-06,
+      "loss": 0.929,
+      "step": 2291
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.6308943089430892e-06,
+      "loss": 0.7713,
+      "step": 2292
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.6276422764227644e-06,
+      "loss": 0.8627,
+      "step": 2293
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.624390243902439e-06,
+      "loss": 0.8584,
+      "step": 2294
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.621138211382114e-06,
+      "loss": 0.7761,
+      "step": 2295
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.6178861788617887e-06,
+      "loss": 0.9668,
+      "step": 2296
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.614634146341464e-06,
+      "loss": 0.7447,
+      "step": 2297
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.6113821138211382e-06,
+      "loss": 0.9702,
+      "step": 2298
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.608130081300813e-06,
+      "loss": 0.9934,
+      "step": 2299
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.604878048780488e-06,
+      "loss": 0.8562,
+      "step": 2300
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.772,
+      "step": 2301
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5983739837398377e-06,
+      "loss": 0.8474,
+      "step": 2302
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5951219512195124e-06,
+      "loss": 0.817,
+      "step": 2303
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5918699186991876e-06,
+      "loss": 0.8827,
+      "step": 2304
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.588617886178862e-06,
+      "loss": 0.7928,
+      "step": 2305
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5853658536585367e-06,
+      "loss": 0.7097,
+      "step": 2306
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.582113821138212e-06,
+      "loss": 0.9019,
+      "step": 2307
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5788617886178863e-06,
+      "loss": 0.7652,
+      "step": 2308
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5756097560975614e-06,
+      "loss": 0.7483,
+      "step": 2309
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.572357723577236e-06,
+      "loss": 0.8069,
+      "step": 2310
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5691056910569105e-06,
+      "loss": 0.8829,
+      "step": 2311
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5658536585365857e-06,
+      "loss": 0.9629,
+      "step": 2312
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5626016260162605e-06,
+      "loss": 0.9027,
+      "step": 2313
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.559349593495935e-06,
+      "loss": 0.776,
+      "step": 2314
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.55609756097561e-06,
+      "loss": 0.9625,
+      "step": 2315
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5528455284552848e-06,
+      "loss": 0.8059,
+      "step": 2316
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5495934959349595e-06,
+      "loss": 0.871,
+      "step": 2317
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5463414634146343e-06,
+      "loss": 0.7561,
+      "step": 2318
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5430894308943095e-06,
+      "loss": 0.8356,
+      "step": 2319
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.539837398373984e-06,
+      "loss": 0.8328,
+      "step": 2320
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5365853658536586e-06,
+      "loss": 0.9405,
+      "step": 2321
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5333333333333338e-06,
+      "loss": 0.8312,
+      "step": 2322
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.530081300813008e-06,
+      "loss": 0.7929,
+      "step": 2323
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5268292682926833e-06,
+      "loss": 0.9431,
+      "step": 2324
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.523577235772358e-06,
+      "loss": 0.8986,
+      "step": 2325
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5203252032520324e-06,
+      "loss": 0.8828,
+      "step": 2326
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5170731707317076e-06,
+      "loss": 0.7367,
+      "step": 2327
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5138211382113824e-06,
+      "loss": 0.7658,
+      "step": 2328
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.510569105691057e-06,
+      "loss": 0.8784,
+      "step": 2329
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.507317073170732e-06,
+      "loss": 1.0107,
+      "step": 2330
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.504065040650407e-06,
+      "loss": 0.8809,
+      "step": 2331
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.5008130081300814e-06,
+      "loss": 0.8402,
+      "step": 2332
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.497560975609756e-06,
+      "loss": 0.8338,
+      "step": 2333
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.494308943089431e-06,
+      "loss": 0.9476,
+      "step": 2334
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.491056910569106e-06,
+      "loss": 0.8664,
+      "step": 2335
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.487804878048781e-06,
+      "loss": 0.9418,
+      "step": 2336
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4845528455284552e-06,
+      "loss": 0.8065,
+      "step": 2337
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4813008130081304e-06,
+      "loss": 0.8279,
+      "step": 2338
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.478048780487805e-06,
+      "loss": 0.99,
+      "step": 2339
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.47479674796748e-06,
+      "loss": 0.9441,
+      "step": 2340
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4715447154471547e-06,
+      "loss": 0.842,
+      "step": 2341
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4682926829268295e-06,
+      "loss": 0.7666,
+      "step": 2342
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4650406504065042e-06,
+      "loss": 0.8928,
+      "step": 2343
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.461788617886179e-06,
+      "loss": 0.7312,
+      "step": 2344
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4585365853658537e-06,
+      "loss": 0.943,
+      "step": 2345
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4552845528455285e-06,
+      "loss": 0.8144,
+      "step": 2346
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4520325203252037e-06,
+      "loss": 0.97,
+      "step": 2347
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.448780487804878e-06,
+      "loss": 0.787,
+      "step": 2348
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4455284552845532e-06,
+      "loss": 0.9174,
+      "step": 2349
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.442276422764228e-06,
+      "loss": 0.7924,
+      "step": 2350
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.0329,
+      "step": 2351
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4357723577235775e-06,
+      "loss": 0.8766,
+      "step": 2352
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4325203252032523e-06,
+      "loss": 0.9223,
+      "step": 2353
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.429268292682927e-06,
+      "loss": 0.7521,
+      "step": 2354
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.426016260162602e-06,
+      "loss": 0.9403,
+      "step": 2355
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4227642276422766e-06,
+      "loss": 0.7921,
+      "step": 2356
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4195121951219513e-06,
+      "loss": 0.7651,
+      "step": 2357
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4162601626016265e-06,
+      "loss": 0.7628,
+      "step": 2358
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.413008130081301e-06,
+      "loss": 0.8316,
+      "step": 2359
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4097560975609756e-06,
+      "loss": 0.7476,
+      "step": 2360
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.406504065040651e-06,
+      "loss": 0.7875,
+      "step": 2361
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4032520325203256e-06,
+      "loss": 0.9822,
+      "step": 2362
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.0787,
+      "step": 2363
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.396747967479675e-06,
+      "loss": 0.8608,
+      "step": 2364
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.39349593495935e-06,
+      "loss": 0.934,
+      "step": 2365
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.3902439024390246e-06,
+      "loss": 0.8149,
+      "step": 2366
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.3869918699186994e-06,
+      "loss": 0.9057,
+      "step": 2367
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.383739837398374e-06,
+      "loss": 0.796,
+      "step": 2368
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.380487804878049e-06,
+      "loss": 0.7725,
+      "step": 2369
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.3772357723577237e-06,
+      "loss": 0.7891,
+      "step": 2370
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.3739837398373984e-06,
+      "loss": 0.8239,
+      "step": 2371
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.370731707317073e-06,
+      "loss": 0.9413,
+      "step": 2372
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.3674796747967484e-06,
+      "loss": 0.7191,
+      "step": 2373
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.364227642276423e-06,
+      "loss": 0.8987,
+      "step": 2374
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.3609756097560975e-06,
+      "loss": 0.8322,
+      "step": 2375
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.3577235772357727e-06,
+      "loss": 0.878,
+      "step": 2376
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.3544715447154474e-06,
+      "loss": 0.873,
+      "step": 2377
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.351219512195122e-06,
+      "loss": 0.83,
+      "step": 2378
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.347967479674797e-06,
+      "loss": 0.8514,
+      "step": 2379
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.3447154471544717e-06,
+      "loss": 0.8253,
+      "step": 2380
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.3414634146341465e-06,
+      "loss": 0.7751,
+      "step": 2381
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.3382113821138212e-06,
+      "loss": 0.9781,
+      "step": 2382
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.334959349593496e-06,
+      "loss": 0.8251,
+      "step": 2383
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.331707317073171e-06,
+      "loss": 0.9438,
+      "step": 2384
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.328455284552846e-06,
+      "loss": 0.7915,
+      "step": 2385
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.3252032520325203e-06,
+      "loss": 1.0171,
+      "step": 2386
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.3219512195121955e-06,
+      "loss": 0.872,
+      "step": 2387
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.3186991869918702e-06,
+      "loss": 0.815,
+      "step": 2388
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.315447154471545e-06,
+      "loss": 0.7383,
+      "step": 2389
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.3121951219512197e-06,
+      "loss": 0.9586,
+      "step": 2390
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.3089430894308945e-06,
+      "loss": 0.8285,
+      "step": 2391
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.3056910569105693e-06,
+      "loss": 0.9637,
+      "step": 2392
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.302439024390244e-06,
+      "loss": 0.9438,
+      "step": 2393
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.299186991869919e-06,
+      "loss": 0.9946,
+      "step": 2394
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.2959349593495936e-06,
+      "loss": 1.0081,
+      "step": 2395
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.2926829268292687e-06,
+      "loss": 0.8554,
+      "step": 2396
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.289430894308943e-06,
+      "loss": 0.9605,
+      "step": 2397
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.286178861788618e-06,
+      "loss": 0.7358,
+      "step": 2398
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.282926829268293e-06,
+      "loss": 1.0177,
+      "step": 2399
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.279674796747968e-06,
+      "loss": 0.8862,
+      "step": 2400
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.8721,
+      "step": 2401
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.2731707317073173e-06,
+      "loss": 0.848,
+      "step": 2402
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.269918699186992e-06,
+      "loss": 0.8828,
+      "step": 2403
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 0.7963,
+      "step": 2404
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.2634146341463416e-06,
+      "loss": 0.9201,
+      "step": 2405
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.2601626016260164e-06,
+      "loss": 0.9501,
+      "step": 2406
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.256910569105691e-06,
+      "loss": 0.9589,
+      "step": 2407
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.253658536585366e-06,
+      "loss": 0.6695,
+      "step": 2408
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.2504065040650407e-06,
+      "loss": 0.766,
+      "step": 2409
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.2471544715447154e-06,
+      "loss": 0.8731,
+      "step": 2410
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.2439024390243906e-06,
+      "loss": 0.8522,
+      "step": 2411
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.2406504065040654e-06,
+      "loss": 0.8862,
+      "step": 2412
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.23739837398374e-06,
+      "loss": 0.8205,
+      "step": 2413
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.234146341463415e-06,
+      "loss": 0.8431,
+      "step": 2414
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.2308943089430897e-06,
+      "loss": 0.884,
+      "step": 2415
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.2276422764227644e-06,
+      "loss": 0.7732,
+      "step": 2416
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.224390243902439e-06,
+      "loss": 0.9772,
+      "step": 2417
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.221138211382114e-06,
+      "loss": 0.9689,
+      "step": 2418
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.217886178861789e-06,
+      "loss": 0.746,
+      "step": 2419
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.2146341463414635e-06,
+      "loss": 0.8014,
+      "step": 2420
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.2113821138211382e-06,
+      "loss": 0.8857,
+      "step": 2421
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.2081300813008134e-06,
+      "loss": 0.9308,
+      "step": 2422
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.204878048780488e-06,
+      "loss": 0.7734,
+      "step": 2423
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.2016260162601625e-06,
+      "loss": 0.794,
+      "step": 2424
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.1983739837398377e-06,
+      "loss": 0.8437,
+      "step": 2425
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 0.9291,
+      "step": 2426
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.1918699186991872e-06,
+      "loss": 0.8318,
+      "step": 2427
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.188617886178862e-06,
+      "loss": 0.7526,
+      "step": 2428
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.1853658536585368e-06,
+      "loss": 0.8722,
+      "step": 2429
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.1821138211382115e-06,
+      "loss": 0.853,
+      "step": 2430
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.1788617886178863e-06,
+      "loss": 0.9531,
+      "step": 2431
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.175609756097561e-06,
+      "loss": 0.9004,
+      "step": 2432
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.172357723577236e-06,
+      "loss": 0.8212,
+      "step": 2433
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.169105691056911e-06,
+      "loss": 0.9251,
+      "step": 2434
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.1658536585365853e-06,
+      "loss": 0.8167,
+      "step": 2435
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.16260162601626e-06,
+      "loss": 0.8732,
+      "step": 2436
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.1593495934959353e-06,
+      "loss": 0.9728,
+      "step": 2437
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.15609756097561e-06,
+      "loss": 0.716,
+      "step": 2438
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.152845528455285e-06,
+      "loss": 0.8149,
+      "step": 2439
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.1495934959349596e-06,
+      "loss": 0.8749,
+      "step": 2440
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.1463414634146343e-06,
+      "loss": 0.7472,
+      "step": 2441
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.143089430894309e-06,
+      "loss": 0.9653,
+      "step": 2442
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.139837398373984e-06,
+      "loss": 0.8085,
+      "step": 2443
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.1365853658536586e-06,
+      "loss": 0.8921,
+      "step": 2444
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 0.7861,
+      "step": 2445
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.130081300813008e-06,
+      "loss": 1.0619,
+      "step": 2446
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.126829268292683e-06,
+      "loss": 0.8891,
+      "step": 2447
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.123577235772358e-06,
+      "loss": 0.7818,
+      "step": 2448
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.120325203252033e-06,
+      "loss": 0.8955,
+      "step": 2449
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.1170731707317076e-06,
+      "loss": 0.844,
+      "step": 2450
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.1138211382113824e-06,
+      "loss": 0.772,
+      "step": 2451
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.110569105691057e-06,
+      "loss": 0.8978,
+      "step": 2452
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.107317073170732e-06,
+      "loss": 1.0224,
+      "step": 2453
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.1040650406504067e-06,
+      "loss": 0.7475,
+      "step": 2454
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.1008130081300814e-06,
+      "loss": 0.7998,
+      "step": 2455
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.097560975609756e-06,
+      "loss": 0.8307,
+      "step": 2456
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.0943089430894314e-06,
+      "loss": 0.8995,
+      "step": 2457
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.0910569105691057e-06,
+      "loss": 0.9728,
+      "step": 2458
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.0878048780487805e-06,
+      "loss": 0.8196,
+      "step": 2459
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.0845528455284557e-06,
+      "loss": 0.8251,
+      "step": 2460
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.0813008130081304e-06,
+      "loss": 0.8712,
+      "step": 2461
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.0780487804878048e-06,
+      "loss": 0.6191,
+      "step": 2462
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.07479674796748e-06,
+      "loss": 0.8669,
+      "step": 2463
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.0715447154471547e-06,
+      "loss": 0.8346,
+      "step": 2464
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.0682926829268295e-06,
+      "loss": 0.7507,
+      "step": 2465
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.0650406504065042e-06,
+      "loss": 0.8224,
+      "step": 2466
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.061788617886179e-06,
+      "loss": 0.8622,
+      "step": 2467
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.0585365853658538e-06,
+      "loss": 0.9387,
+      "step": 2468
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.0552845528455285e-06,
+      "loss": 0.9829,
+      "step": 2469
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.0520325203252033e-06,
+      "loss": 0.9258,
+      "step": 2470
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.048780487804878e-06,
+      "loss": 0.9407,
+      "step": 2471
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.0455284552845532e-06,
+      "loss": 0.8397,
+      "step": 2472
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.0422764227642276e-06,
+      "loss": 0.7914,
+      "step": 2473
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.0390243902439023e-06,
+      "loss": 0.8109,
+      "step": 2474
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.0357723577235775e-06,
+      "loss": 0.8599,
+      "step": 2475
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.0325203252032523e-06,
+      "loss": 0.8731,
+      "step": 2476
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.029268292682927e-06,
+      "loss": 0.8209,
+      "step": 2477
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.026016260162602e-06,
+      "loss": 0.7348,
+      "step": 2478
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0227642276422766e-06,
+      "loss": 0.7656,
+      "step": 2479
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0195121951219513e-06,
+      "loss": 0.8014,
+      "step": 2480
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.016260162601626e-06,
+      "loss": 0.8323,
+      "step": 2481
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.013008130081301e-06,
+      "loss": 0.789,
+      "step": 2482
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.009756097560976e-06,
+      "loss": 0.6998,
+      "step": 2483
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.0065040650406504e-06,
+      "loss": 0.8238,
+      "step": 2484
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.003252032520325e-06,
+      "loss": 0.832,
+      "step": 2485
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7618,
+      "step": 2486
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.996747967479675e-06,
+      "loss": 0.8485,
+      "step": 2487
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.99349593495935e-06,
+      "loss": 0.7481,
+      "step": 2488
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.9902439024390246e-06,
+      "loss": 0.792,
+      "step": 2489
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.9869918699186994e-06,
+      "loss": 0.7453,
+      "step": 2490
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.983739837398374e-06,
+      "loss": 0.8376,
+      "step": 2491
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.980487804878049e-06,
+      "loss": 0.5767,
+      "step": 2492
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.9772357723577237e-06,
+      "loss": 0.7342,
+      "step": 2493
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.9739837398373984e-06,
+      "loss": 0.7111,
+      "step": 2494
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.9707317073170736e-06,
+      "loss": 0.6576,
+      "step": 2495
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.967479674796748e-06,
+      "loss": 0.6067,
+      "step": 2496
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.9642276422764227e-06,
+      "loss": 0.8442,
+      "step": 2497
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.960975609756098e-06,
+      "loss": 0.5893,
+      "step": 2498
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.9577235772357727e-06,
+      "loss": 0.7885,
+      "step": 2499
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.954471544715447e-06,
+      "loss": 0.9262,
+      "step": 2500
+    },
+    {
+      "epoch": 8.05,
+      "eval_loss": 1.9857499599456787,
+      "eval_runtime": 135.9973,
+      "eval_samples_per_second": 4.456,
+      "eval_steps_per_second": 4.456,
+      "step": 2500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 0.8382,
+      "step": 2501
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.947967479674797e-06,
+      "loss": 0.8426,
+      "step": 2502
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9447154471544717e-06,
+      "loss": 0.7915,
+      "step": 2503
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9414634146341465e-06,
+      "loss": 0.7574,
+      "step": 2504
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9382113821138212e-06,
+      "loss": 0.7837,
+      "step": 2505
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.934959349593496e-06,
+      "loss": 0.6643,
+      "step": 2506
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.9317073170731708e-06,
+      "loss": 0.7335,
+      "step": 2507
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.9284552845528455e-06,
+      "loss": 0.771,
+      "step": 2508
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.9252032520325207e-06,
+      "loss": 0.757,
+      "step": 2509
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.9219512195121955e-06,
+      "loss": 0.7616,
+      "step": 2510
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.91869918699187e-06,
+      "loss": 0.9229,
+      "step": 2511
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.915447154471545e-06,
+      "loss": 0.8214,
+      "step": 2512
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.9121951219512198e-06,
+      "loss": 0.8518,
+      "step": 2513
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.9089430894308945e-06,
+      "loss": 0.6969,
+      "step": 2514
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.9056910569105693e-06,
+      "loss": 0.7273,
+      "step": 2515
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.902439024390244e-06,
+      "loss": 0.7005,
+      "step": 2516
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.8991869918699188e-06,
+      "loss": 0.7219,
+      "step": 2517
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.8959349593495938e-06,
+      "loss": 0.7595,
+      "step": 2518
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.8926829268292686e-06,
+      "loss": 0.8309,
+      "step": 2519
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.889430894308943e-06,
+      "loss": 0.914,
+      "step": 2520
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.886178861788618e-06,
+      "loss": 0.8434,
+      "step": 2521
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8829268292682928e-06,
+      "loss": 0.79,
+      "step": 2522
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8796747967479676e-06,
+      "loss": 0.9106,
+      "step": 2523
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.8764227642276426e-06,
+      "loss": 0.7494,
+      "step": 2524
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.8731707317073171e-06,
+      "loss": 0.823,
+      "step": 2525
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.8699186991869919e-06,
+      "loss": 0.8547,
+      "step": 2526
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 0.8139,
+      "step": 2527
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.8634146341463416e-06,
+      "loss": 0.8484,
+      "step": 2528
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.8601626016260164e-06,
+      "loss": 0.7859,
+      "step": 2529
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.8569105691056914e-06,
+      "loss": 0.7593,
+      "step": 2530
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.853658536585366e-06,
+      "loss": 0.7534,
+      "step": 2531
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.8504065040650407e-06,
+      "loss": 0.8198,
+      "step": 2532
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.8471544715447157e-06,
+      "loss": 0.8299,
+      "step": 2533
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.8439024390243904e-06,
+      "loss": 0.841,
+      "step": 2534
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.840650406504065e-06,
+      "loss": 0.894,
+      "step": 2535
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.8373983739837402e-06,
+      "loss": 0.8208,
+      "step": 2536
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.8341463414634147e-06,
+      "loss": 0.6134,
+      "step": 2537
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.8308943089430895e-06,
+      "loss": 0.7202,
+      "step": 2538
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.8276422764227644e-06,
+      "loss": 0.7472,
+      "step": 2539
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.8243902439024392e-06,
+      "loss": 0.7892,
+      "step": 2540
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.8211382113821142e-06,
+      "loss": 0.7168,
+      "step": 2541
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.8178861788617887e-06,
+      "loss": 0.9145,
+      "step": 2542
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.8146341463414635e-06,
+      "loss": 0.6541,
+      "step": 2543
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.8113821138211385e-06,
+      "loss": 1.0083,
+      "step": 2544
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.8081300813008132e-06,
+      "loss": 0.7809,
+      "step": 2545
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.804878048780488e-06,
+      "loss": 0.726,
+      "step": 2546
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.801626016260163e-06,
+      "loss": 0.8,
+      "step": 2547
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.7983739837398375e-06,
+      "loss": 0.8325,
+      "step": 2548
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.7951219512195123e-06,
+      "loss": 0.7973,
+      "step": 2549
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.7918699186991872e-06,
+      "loss": 0.8428,
+      "step": 2550
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.788617886178862e-06,
+      "loss": 0.7955,
+      "step": 2551
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.7853658536585366e-06,
+      "loss": 0.7905,
+      "step": 2552
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.7821138211382115e-06,
+      "loss": 0.8911,
+      "step": 2553
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.7788617886178863e-06,
+      "loss": 0.8791,
+      "step": 2554
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.775609756097561e-06,
+      "loss": 0.7528,
+      "step": 2555
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.772357723577236e-06,
+      "loss": 0.9365,
+      "step": 2556
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.7691056910569108e-06,
+      "loss": 0.8988,
+      "step": 2557
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.7658536585365853e-06,
+      "loss": 0.8189,
+      "step": 2558
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.7626016260162603e-06,
+      "loss": 0.8614,
+      "step": 2559
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.759349593495935e-06,
+      "loss": 0.8022,
+      "step": 2560
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.7560975609756098e-06,
+      "loss": 0.8298,
+      "step": 2561
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.7528455284552848e-06,
+      "loss": 0.841,
+      "step": 2562
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.7495934959349594e-06,
+      "loss": 0.863,
+      "step": 2563
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.7463414634146341e-06,
+      "loss": 0.6562,
+      "step": 2564
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.7430894308943091e-06,
+      "loss": 0.8378,
+      "step": 2565
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.7398373983739839e-06,
+      "loss": 0.5639,
+      "step": 2566
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.7365853658536586e-06,
+      "loss": 0.6666,
+      "step": 2567
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.7333333333333336e-06,
+      "loss": 0.896,
+      "step": 2568
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.7300813008130082e-06,
+      "loss": 0.8718,
+      "step": 2569
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.726829268292683e-06,
+      "loss": 0.7406,
+      "step": 2570
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.723577235772358e-06,
+      "loss": 0.8727,
+      "step": 2571
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.7203252032520327e-06,
+      "loss": 0.7111,
+      "step": 2572
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.7170731707317076e-06,
+      "loss": 0.7121,
+      "step": 2573
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.7138211382113824e-06,
+      "loss": 0.72,
+      "step": 2574
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.710569105691057e-06,
+      "loss": 0.8466,
+      "step": 2575
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 0.7748,
+      "step": 2576
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.7040650406504067e-06,
+      "loss": 0.8114,
+      "step": 2577
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.7008130081300814e-06,
+      "loss": 0.7182,
+      "step": 2578
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.6975609756097564e-06,
+      "loss": 0.691,
+      "step": 2579
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.694308943089431e-06,
+      "loss": 0.6144,
+      "step": 2580
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.6910569105691057e-06,
+      "loss": 0.7782,
+      "step": 2581
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.6878048780487807e-06,
+      "loss": 0.9333,
+      "step": 2582
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.6845528455284555e-06,
+      "loss": 0.7379,
+      "step": 2583
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.6813008130081302e-06,
+      "loss": 0.8715,
+      "step": 2584
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.6780487804878052e-06,
+      "loss": 0.8561,
+      "step": 2585
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.6747967479674798e-06,
+      "loss": 0.8806,
+      "step": 2586
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.6715447154471545e-06,
+      "loss": 0.6913,
+      "step": 2587
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.6682926829268295e-06,
+      "loss": 0.7961,
+      "step": 2588
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.6650406504065043e-06,
+      "loss": 0.7317,
+      "step": 2589
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.6617886178861788e-06,
+      "loss": 0.7641,
+      "step": 2590
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.6585365853658538e-06,
+      "loss": 0.7064,
+      "step": 2591
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.6552845528455285e-06,
+      "loss": 0.7138,
+      "step": 2592
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.6520325203252033e-06,
+      "loss": 0.9043,
+      "step": 2593
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.6487804878048783e-06,
+      "loss": 0.7792,
+      "step": 2594
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.645528455284553e-06,
+      "loss": 0.7451,
+      "step": 2595
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.6422764227642276e-06,
+      "loss": 0.7532,
+      "step": 2596
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.6390243902439026e-06,
+      "loss": 0.8676,
+      "step": 2597
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.6357723577235773e-06,
+      "loss": 0.8371,
+      "step": 2598
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.632520325203252e-06,
+      "loss": 0.8396,
+      "step": 2599
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.629268292682927e-06,
+      "loss": 0.6775,
+      "step": 2600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.9844,
+      "step": 2601
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.6227642276422764e-06,
+      "loss": 0.9088,
+      "step": 2602
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.6195121951219514e-06,
+      "loss": 0.8084,
+      "step": 2603
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.6162601626016261e-06,
+      "loss": 0.64,
+      "step": 2604
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.613008130081301e-06,
+      "loss": 0.8108,
+      "step": 2605
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.6097560975609759e-06,
+      "loss": 0.7847,
+      "step": 2606
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.6065040650406504e-06,
+      "loss": 0.8091,
+      "step": 2607
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.6032520325203254e-06,
+      "loss": 0.8447,
+      "step": 2608
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.8159,
+      "step": 2609
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.596747967479675e-06,
+      "loss": 0.9322,
+      "step": 2610
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.5934959349593499e-06,
+      "loss": 0.7182,
+      "step": 2611
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.5902439024390246e-06,
+      "loss": 0.8803,
+      "step": 2612
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.5869918699186992e-06,
+      "loss": 0.8079,
+      "step": 2613
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.5837398373983742e-06,
+      "loss": 0.8583,
+      "step": 2614
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.580487804878049e-06,
+      "loss": 0.71,
+      "step": 2615
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.5772357723577237e-06,
+      "loss": 0.6742,
+      "step": 2616
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.5739837398373987e-06,
+      "loss": 0.79,
+      "step": 2617
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.5707317073170732e-06,
+      "loss": 0.7912,
+      "step": 2618
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.567479674796748e-06,
+      "loss": 0.9503,
+      "step": 2619
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.564227642276423e-06,
+      "loss": 0.7591,
+      "step": 2620
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.5609756097560977e-06,
+      "loss": 0.8561,
+      "step": 2621
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.5577235772357725e-06,
+      "loss": 0.6734,
+      "step": 2622
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.5544715447154474e-06,
+      "loss": 0.7319,
+      "step": 2623
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.551219512195122e-06,
+      "loss": 0.816,
+      "step": 2624
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.5479674796747968e-06,
+      "loss": 0.825,
+      "step": 2625
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.5447154471544717e-06,
+      "loss": 0.8167,
+      "step": 2626
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.5414634146341465e-06,
+      "loss": 0.6726,
+      "step": 2627
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.538211382113821e-06,
+      "loss": 0.8204,
+      "step": 2628
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.534959349593496e-06,
+      "loss": 0.8421,
+      "step": 2629
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.5317073170731708e-06,
+      "loss": 0.7228,
+      "step": 2630
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.5284552845528455e-06,
+      "loss": 0.7864,
+      "step": 2631
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.5252032520325205e-06,
+      "loss": 0.8534,
+      "step": 2632
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.5219512195121953e-06,
+      "loss": 0.804,
+      "step": 2633
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.5186991869918698e-06,
+      "loss": 0.8841,
+      "step": 2634
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.5154471544715448e-06,
+      "loss": 0.7959,
+      "step": 2635
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.5121951219512196e-06,
+      "loss": 0.8906,
+      "step": 2636
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.5089430894308945e-06,
+      "loss": 0.7454,
+      "step": 2637
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.5056910569105693e-06,
+      "loss": 0.8813,
+      "step": 2638
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.502439024390244e-06,
+      "loss": 0.7613,
+      "step": 2639
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.499186991869919e-06,
+      "loss": 0.8562,
+      "step": 2640
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4959349593495936e-06,
+      "loss": 0.9844,
+      "step": 2641
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.4926829268292684e-06,
+      "loss": 0.7293,
+      "step": 2642
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4894308943089433e-06,
+      "loss": 0.7452,
+      "step": 2643
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.486178861788618e-06,
+      "loss": 0.7536,
+      "step": 2644
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.4829268292682926e-06,
+      "loss": 0.9007,
+      "step": 2645
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4796747967479676e-06,
+      "loss": 0.7968,
+      "step": 2646
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4764227642276424e-06,
+      "loss": 0.7665,
+      "step": 2647
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.4731707317073171e-06,
+      "loss": 0.5906,
+      "step": 2648
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4699186991869921e-06,
+      "loss": 0.7636,
+      "step": 2649
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4666666666666669e-06,
+      "loss": 0.8901,
+      "step": 2650
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 0.7748,
+      "step": 2651
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4601626016260164e-06,
+      "loss": 0.8218,
+      "step": 2652
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.4569105691056912e-06,
+      "loss": 0.7197,
+      "step": 2653
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.453658536585366e-06,
+      "loss": 0.72,
+      "step": 2654
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.450406504065041e-06,
+      "loss": 0.716,
+      "step": 2655
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4471544715447155e-06,
+      "loss": 0.7285,
+      "step": 2656
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.4439024390243902e-06,
+      "loss": 0.8443,
+      "step": 2657
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4406504065040652e-06,
+      "loss": 0.7434,
+      "step": 2658
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.43739837398374e-06,
+      "loss": 0.7237,
+      "step": 2659
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4341463414634147e-06,
+      "loss": 0.8555,
+      "step": 2660
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.4308943089430897e-06,
+      "loss": 0.8884,
+      "step": 2661
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.4276422764227642e-06,
+      "loss": 0.9173,
+      "step": 2662
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.424390243902439e-06,
+      "loss": 0.8327,
+      "step": 2663
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.421138211382114e-06,
+      "loss": 0.8434,
+      "step": 2664
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4178861788617887e-06,
+      "loss": 0.7792,
+      "step": 2665
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4146341463414633e-06,
+      "loss": 0.8767,
+      "step": 2666
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.4113821138211385e-06,
+      "loss": 0.9598,
+      "step": 2667
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.408130081300813e-06,
+      "loss": 0.9087,
+      "step": 2668
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.404878048780488e-06,
+      "loss": 0.7902,
+      "step": 2669
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.4016260162601628e-06,
+      "loss": 0.8338,
+      "step": 2670
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.3983739837398375e-06,
+      "loss": 0.6965,
+      "step": 2671
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.3951219512195125e-06,
+      "loss": 0.8239,
+      "step": 2672
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.391869918699187e-06,
+      "loss": 0.7067,
+      "step": 2673
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.3886178861788618e-06,
+      "loss": 0.8679,
+      "step": 2674
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.3853658536585368e-06,
+      "loss": 0.9516,
+      "step": 2675
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.3821138211382116e-06,
+      "loss": 0.6944,
+      "step": 2676
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.3788617886178863e-06,
+      "loss": 0.8955,
+      "step": 2677
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.3756097560975613e-06,
+      "loss": 0.7021,
+      "step": 2678
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.3723577235772358e-06,
+      "loss": 0.7746,
+      "step": 2679
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.3691056910569106e-06,
+      "loss": 0.7376,
+      "step": 2680
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.3658536585365856e-06,
+      "loss": 0.7502,
+      "step": 2681
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.3626016260162603e-06,
+      "loss": 0.681,
+      "step": 2682
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3593495934959349e-06,
+      "loss": 0.8126,
+      "step": 2683
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3560975609756099e-06,
+      "loss": 0.8904,
+      "step": 2684
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.3528455284552846e-06,
+      "loss": 0.7834,
+      "step": 2685
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.3495934959349594e-06,
+      "loss": 0.8819,
+      "step": 2686
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.3463414634146344e-06,
+      "loss": 0.6271,
+      "step": 2687
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.3430894308943091e-06,
+      "loss": 0.8307,
+      "step": 2688
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.3398373983739837e-06,
+      "loss": 0.6898,
+      "step": 2689
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.3365853658536587e-06,
+      "loss": 0.8786,
+      "step": 2690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.8688,
+      "step": 2691
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.3300813008130082e-06,
+      "loss": 0.8375,
+      "step": 2692
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.3268292682926832e-06,
+      "loss": 0.8658,
+      "step": 2693
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.3235772357723577e-06,
+      "loss": 0.8042,
+      "step": 2694
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.3203252032520325e-06,
+      "loss": 0.7777,
+      "step": 2695
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.3170731707317074e-06,
+      "loss": 0.8576,
+      "step": 2696
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.3138211382113822e-06,
+      "loss": 0.8916,
+      "step": 2697
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.310569105691057e-06,
+      "loss": 0.8744,
+      "step": 2698
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.307317073170732e-06,
+      "loss": 0.8025,
+      "step": 2699
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.3040650406504065e-06,
+      "loss": 0.8599,
+      "step": 2700
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 0.6618,
+      "step": 2701
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.2975609756097562e-06,
+      "loss": 0.8232,
+      "step": 2702
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.294308943089431e-06,
+      "loss": 0.7964,
+      "step": 2703
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.291056910569106e-06,
+      "loss": 0.802,
+      "step": 2704
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2878048780487807e-06,
+      "loss": 0.9186,
+      "step": 2705
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2845528455284553e-06,
+      "loss": 0.6659,
+      "step": 2706
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.2813008130081302e-06,
+      "loss": 0.8122,
+      "step": 2707
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.278048780487805e-06,
+      "loss": 0.9259,
+      "step": 2708
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2747967479674798e-06,
+      "loss": 0.8316,
+      "step": 2709
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.2715447154471547e-06,
+      "loss": 0.9127,
+      "step": 2710
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.2682926829268293e-06,
+      "loss": 0.7934,
+      "step": 2711
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.265040650406504e-06,
+      "loss": 0.6651,
+      "step": 2712
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.261788617886179e-06,
+      "loss": 0.8287,
+      "step": 2713
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2585365853658538e-06,
+      "loss": 0.8649,
+      "step": 2714
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2552845528455286e-06,
+      "loss": 0.7376,
+      "step": 2715
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.2520325203252035e-06,
+      "loss": 0.7184,
+      "step": 2716
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.248780487804878e-06,
+      "loss": 0.9284,
+      "step": 2717
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.245528455284553e-06,
+      "loss": 0.8174,
+      "step": 2718
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2422764227642276e-06,
+      "loss": 0.9374,
+      "step": 2719
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.2390243902439026e-06,
+      "loss": 0.6986,
+      "step": 2720
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2357723577235773e-06,
+      "loss": 0.8361,
+      "step": 2721
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2325203252032521e-06,
+      "loss": 0.8859,
+      "step": 2722
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.2292682926829269e-06,
+      "loss": 0.8261,
+      "step": 2723
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2260162601626018e-06,
+      "loss": 0.7992,
+      "step": 2724
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2227642276422766e-06,
+      "loss": 0.8041,
+      "step": 2725
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.7586,
+      "step": 2726
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2162601626016261e-06,
+      "loss": 0.8112,
+      "step": 2727
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.213008130081301e-06,
+      "loss": 0.7406,
+      "step": 2728
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.2097560975609757e-06,
+      "loss": 0.7698,
+      "step": 2729
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2065040650406504e-06,
+      "loss": 0.8667,
+      "step": 2730
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2032520325203254e-06,
+      "loss": 0.7903,
+      "step": 2731
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.7718,
+      "step": 2732
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.196747967479675e-06,
+      "loss": 0.7601,
+      "step": 2733
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.1934959349593497e-06,
+      "loss": 0.7891,
+      "step": 2734
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.1902439024390244e-06,
+      "loss": 0.8235,
+      "step": 2735
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1869918699186992e-06,
+      "loss": 0.9065,
+      "step": 2736
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1837398373983742e-06,
+      "loss": 0.9013,
+      "step": 2737
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1804878048780487e-06,
+      "loss": 0.608,
+      "step": 2738
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.1772357723577237e-06,
+      "loss": 0.9225,
+      "step": 2739
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.1739837398373985e-06,
+      "loss": 0.8239,
+      "step": 2740
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.1707317073170732e-06,
+      "loss": 0.6603,
+      "step": 2741
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.167479674796748e-06,
+      "loss": 0.8185,
+      "step": 2742
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.164227642276423e-06,
+      "loss": 0.7435,
+      "step": 2743
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.1609756097560977e-06,
+      "loss": 0.8175,
+      "step": 2744
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.1577235772357725e-06,
+      "loss": 0.8369,
+      "step": 2745
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.1544715447154473e-06,
+      "loss": 0.7378,
+      "step": 2746
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.151219512195122e-06,
+      "loss": 0.798,
+      "step": 2747
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.1479674796747968e-06,
+      "loss": 0.6445,
+      "step": 2748
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.1447154471544715e-06,
+      "loss": 0.7721,
+      "step": 2749
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.1414634146341465e-06,
+      "loss": 0.8791,
+      "step": 2750
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.1382113821138213e-06,
+      "loss": 0.8356,
+      "step": 2751
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.134959349593496e-06,
+      "loss": 0.8165,
+      "step": 2752
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.1317073170731708e-06,
+      "loss": 0.8712,
+      "step": 2753
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.1284552845528456e-06,
+      "loss": 0.7448,
+      "step": 2754
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.1252032520325203e-06,
+      "loss": 0.7827,
+      "step": 2755
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.1219512195121953e-06,
+      "loss": 0.8214,
+      "step": 2756
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.11869918699187e-06,
+      "loss": 0.7961,
+      "step": 2757
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.1154471544715448e-06,
+      "loss": 0.6767,
+      "step": 2758
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.1121951219512196e-06,
+      "loss": 0.8714,
+      "step": 2759
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.1089430894308946e-06,
+      "loss": 0.9948,
+      "step": 2760
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.1056910569105691e-06,
+      "loss": 0.8805,
+      "step": 2761
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.102439024390244e-06,
+      "loss": 0.8816,
+      "step": 2762
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.0991869918699189e-06,
+      "loss": 0.8618,
+      "step": 2763
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.0959349593495936e-06,
+      "loss": 0.9003,
+      "step": 2764
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.0926829268292684e-06,
+      "loss": 0.8824,
+      "step": 2765
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.0894308943089431e-06,
+      "loss": 0.8338,
+      "step": 2766
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.086178861788618e-06,
+      "loss": 0.7827,
+      "step": 2767
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.0829268292682927e-06,
+      "loss": 0.6774,
+      "step": 2768
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.0796747967479676e-06,
+      "loss": 0.7873,
+      "step": 2769
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.0764227642276424e-06,
+      "loss": 0.733,
+      "step": 2770
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.0731707317073172e-06,
+      "loss": 0.7833,
+      "step": 2771
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.069918699186992e-06,
+      "loss": 0.6919,
+      "step": 2772
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 0.7902,
+      "step": 2773
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.0634146341463415e-06,
+      "loss": 0.7375,
+      "step": 2774
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.0601626016260164e-06,
+      "loss": 0.7773,
+      "step": 2775
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.0569105691056912e-06,
+      "loss": 0.7862,
+      "step": 2776
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.053658536585366e-06,
+      "loss": 0.7939,
+      "step": 2777
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.0504065040650407e-06,
+      "loss": 0.5942,
+      "step": 2778
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.0471544715447157e-06,
+      "loss": 0.835,
+      "step": 2779
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.0439024390243902e-06,
+      "loss": 0.7686,
+      "step": 2780
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.0406504065040652e-06,
+      "loss": 0.7924,
+      "step": 2781
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.03739837398374e-06,
+      "loss": 0.6881,
+      "step": 2782
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.0341463414634147e-06,
+      "loss": 0.8832,
+      "step": 2783
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.0308943089430895e-06,
+      "loss": 0.854,
+      "step": 2784
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.0276422764227643e-06,
+      "loss": 0.8666,
+      "step": 2785
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.024390243902439e-06,
+      "loss": 0.6883,
+      "step": 2786
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.0211382113821138e-06,
+      "loss": 0.7278,
+      "step": 2787
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.0178861788617888e-06,
+      "loss": 0.8931,
+      "step": 2788
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.0146341463414635e-06,
+      "loss": 0.7565,
+      "step": 2789
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.0113821138211383e-06,
+      "loss": 0.8478,
+      "step": 2790
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.008130081300813e-06,
+      "loss": 0.8019,
+      "step": 2791
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.004878048780488e-06,
+      "loss": 0.9003,
+      "step": 2792
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.0016260162601626e-06,
+      "loss": 0.8849,
+      "step": 2793
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 9.983739837398375e-07,
+      "loss": 0.7809,
+      "step": 2794
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 9.951219512195123e-07,
+      "loss": 0.7016,
+      "step": 2795
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 9.91869918699187e-07,
+      "loss": 0.8892,
+      "step": 2796
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 9.886178861788618e-07,
+      "loss": 0.8792,
+      "step": 2797
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 9.853658536585368e-07,
+      "loss": 0.6934,
+      "step": 2798
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 9.821138211382114e-07,
+      "loss": 0.698,
+      "step": 2799
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 9.788617886178863e-07,
+      "loss": 0.6995,
+      "step": 2800
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 0.5806,
+      "step": 2801
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 9.723577235772359e-07,
+      "loss": 0.8546,
+      "step": 2802
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 9.691056910569106e-07,
+      "loss": 0.8838,
+      "step": 2803
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.658536585365854e-07,
+      "loss": 0.7883,
+      "step": 2804
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.626016260162604e-07,
+      "loss": 0.9095,
+      "step": 2805
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.59349593495935e-07,
+      "loss": 0.8556,
+      "step": 2806
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 9.560975609756099e-07,
+      "loss": 0.6872,
+      "step": 2807
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 9.528455284552846e-07,
+      "loss": 0.7161,
+      "step": 2808
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 9.495934959349594e-07,
+      "loss": 0.7334,
+      "step": 2809
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 9.463414634146343e-07,
+      "loss": 0.7829,
+      "step": 2810
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 9.43089430894309e-07,
+      "loss": 0.7584,
+      "step": 2811
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 9.398373983739838e-07,
+      "loss": 0.771,
+      "step": 2812
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 9.365853658536586e-07,
+      "loss": 0.7526,
+      "step": 2813
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 9.333333333333334e-07,
+      "loss": 0.759,
+      "step": 2814
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 9.300813008130082e-07,
+      "loss": 0.7428,
+      "step": 2815
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 9.26829268292683e-07,
+      "loss": 0.8359,
+      "step": 2816
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 9.235772357723578e-07,
+      "loss": 0.786,
+      "step": 2817
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 9.203252032520325e-07,
+      "loss": 0.8223,
+      "step": 2818
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 9.170731707317074e-07,
+      "loss": 0.8749,
+      "step": 2819
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.138211382113822e-07,
+      "loss": 0.6722,
+      "step": 2820
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.105691056910571e-07,
+      "loss": 0.8349,
+      "step": 2821
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 9.073170731707317e-07,
+      "loss": 0.7477,
+      "step": 2822
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 9.040650406504066e-07,
+      "loss": 0.8174,
+      "step": 2823
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 9.008130081300815e-07,
+      "loss": 0.705,
+      "step": 2824
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 8.975609756097561e-07,
+      "loss": 0.8076,
+      "step": 2825
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 8.94308943089431e-07,
+      "loss": 0.8209,
+      "step": 2826
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 8.910569105691058e-07,
+      "loss": 0.77,
+      "step": 2827
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 8.878048780487805e-07,
+      "loss": 0.7002,
+      "step": 2828
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 8.845528455284554e-07,
+      "loss": 0.8069,
+      "step": 2829
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 8.813008130081302e-07,
+      "loss": 0.7097,
+      "step": 2830
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 8.780487804878049e-07,
+      "loss": 0.6812,
+      "step": 2831
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.747967479674797e-07,
+      "loss": 0.8141,
+      "step": 2832
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.715447154471546e-07,
+      "loss": 0.746,
+      "step": 2833
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.682926829268293e-07,
+      "loss": 0.7899,
+      "step": 2834
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 8.650406504065041e-07,
+      "loss": 0.7012,
+      "step": 2835
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 8.61788617886179e-07,
+      "loss": 0.7718,
+      "step": 2836
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 8.585365853658538e-07,
+      "loss": 0.7284,
+      "step": 2837
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 8.552845528455285e-07,
+      "loss": 0.7957,
+      "step": 2838
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 8.520325203252033e-07,
+      "loss": 0.791,
+      "step": 2839
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 8.487804878048782e-07,
+      "loss": 0.7572,
+      "step": 2840
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 8.455284552845529e-07,
+      "loss": 0.6616,
+      "step": 2841
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 8.422764227642277e-07,
+      "loss": 0.8164,
+      "step": 2842
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 8.390243902439026e-07,
+      "loss": 0.6825,
+      "step": 2843
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 8.357723577235773e-07,
+      "loss": 0.7645,
+      "step": 2844
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 8.325203252032521e-07,
+      "loss": 0.7774,
+      "step": 2845
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 8.292682926829269e-07,
+      "loss": 0.9046,
+      "step": 2846
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 8.260162601626017e-07,
+      "loss": 0.7857,
+      "step": 2847
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.227642276422765e-07,
+      "loss": 0.7671,
+      "step": 2848
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.195121951219513e-07,
+      "loss": 0.7654,
+      "step": 2849
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 8.16260162601626e-07,
+      "loss": 0.7467,
+      "step": 2850
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 8.130081300813009e-07,
+      "loss": 0.8796,
+      "step": 2851
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 8.097560975609757e-07,
+      "loss": 0.7121,
+      "step": 2852
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 8.065040650406505e-07,
+      "loss": 0.7337,
+      "step": 2853
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 8.032520325203252e-07,
+      "loss": 0.6085,
+      "step": 2854
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.7272,
+      "step": 2855
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 7.967479674796749e-07,
+      "loss": 0.873,
+      "step": 2856
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 7.934959349593496e-07,
+      "loss": 0.7689,
+      "step": 2857
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 7.902439024390245e-07,
+      "loss": 0.6676,
+      "step": 2858
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 7.869918699186993e-07,
+      "loss": 0.773,
+      "step": 2859
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.83739837398374e-07,
+      "loss": 0.6695,
+      "step": 2860
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.804878048780489e-07,
+      "loss": 0.6256,
+      "step": 2861
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.772357723577237e-07,
+      "loss": 0.7223,
+      "step": 2862
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.739837398373984e-07,
+      "loss": 0.8559,
+      "step": 2863
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.707317073170732e-07,
+      "loss": 0.7335,
+      "step": 2864
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.67479674796748e-07,
+      "loss": 0.8556,
+      "step": 2865
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.642276422764228e-07,
+      "loss": 0.9024,
+      "step": 2866
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.609756097560976e-07,
+      "loss": 0.698,
+      "step": 2867
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.577235772357724e-07,
+      "loss": 0.7298,
+      "step": 2868
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.544715447154473e-07,
+      "loss": 0.6582,
+      "step": 2869
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.51219512195122e-07,
+      "loss": 0.8144,
+      "step": 2870
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.479674796747968e-07,
+      "loss": 0.7109,
+      "step": 2871
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.447154471544717e-07,
+      "loss": 0.7478,
+      "step": 2872
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.414634146341463e-07,
+      "loss": 0.7506,
+      "step": 2873
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.382113821138212e-07,
+      "loss": 0.8577,
+      "step": 2874
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 7.349593495934961e-07,
+      "loss": 0.7285,
+      "step": 2875
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 0.6615,
+      "step": 2876
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.284552845528456e-07,
+      "loss": 0.6183,
+      "step": 2877
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.252032520325205e-07,
+      "loss": 0.7459,
+      "step": 2878
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.219512195121951e-07,
+      "loss": 0.7322,
+      "step": 2879
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.1869918699187e-07,
+      "loss": 0.8003,
+      "step": 2880
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.154471544715448e-07,
+      "loss": 0.779,
+      "step": 2881
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.121951219512195e-07,
+      "loss": 0.6606,
+      "step": 2882
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.089430894308944e-07,
+      "loss": 0.8238,
+      "step": 2883
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.056910569105692e-07,
+      "loss": 0.8082,
+      "step": 2884
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 7.02439024390244e-07,
+      "loss": 0.6694,
+      "step": 2885
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 6.991869918699188e-07,
+      "loss": 0.7601,
+      "step": 2886
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 6.959349593495935e-07,
+      "loss": 0.7453,
+      "step": 2887
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.926829268292684e-07,
+      "loss": 0.6648,
+      "step": 2888
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.894308943089432e-07,
+      "loss": 0.7912,
+      "step": 2889
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.861788617886179e-07,
+      "loss": 0.7504,
+      "step": 2890
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.829268292682928e-07,
+      "loss": 0.7907,
+      "step": 2891
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.796747967479674e-07,
+      "loss": 0.8977,
+      "step": 2892
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.764227642276423e-07,
+      "loss": 0.623,
+      "step": 2893
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.731707317073172e-07,
+      "loss": 0.6719,
+      "step": 2894
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.699186991869918e-07,
+      "loss": 0.7733,
+      "step": 2895
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 0.7796,
+      "step": 2896
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.634146341463416e-07,
+      "loss": 0.8725,
+      "step": 2897
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.601626016260162e-07,
+      "loss": 0.8382,
+      "step": 2898
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.569105691056911e-07,
+      "loss": 0.613,
+      "step": 2899
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.53658536585366e-07,
+      "loss": 0.6839,
+      "step": 2900
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 0.6895,
+      "step": 2901
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.471544715447155e-07,
+      "loss": 0.8922,
+      "step": 2902
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.439024390243904e-07,
+      "loss": 0.7634,
+      "step": 2903
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.406504065040651e-07,
+      "loss": 0.8561,
+      "step": 2904
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.373983739837399e-07,
+      "loss": 0.7892,
+      "step": 2905
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.341463414634146e-07,
+      "loss": 0.7606,
+      "step": 2906
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.308943089430895e-07,
+      "loss": 0.8116,
+      "step": 2907
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.276422764227643e-07,
+      "loss": 0.6538,
+      "step": 2908
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.24390243902439e-07,
+      "loss": 0.6972,
+      "step": 2909
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.211382113821138e-07,
+      "loss": 0.8771,
+      "step": 2910
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.178861788617887e-07,
+      "loss": 0.7378,
+      "step": 2911
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.146341463414634e-07,
+      "loss": 0.8349,
+      "step": 2912
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.113821138211383e-07,
+      "loss": 0.8339,
+      "step": 2913
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.081300813008131e-07,
+      "loss": 0.7222,
+      "step": 2914
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.048780487804878e-07,
+      "loss": 0.6694,
+      "step": 2915
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.016260162601627e-07,
+      "loss": 0.6351,
+      "step": 2916
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.983739837398375e-07,
+      "loss": 0.7884,
+      "step": 2917
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.951219512195122e-07,
+      "loss": 0.7696,
+      "step": 2918
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.918699186991871e-07,
+      "loss": 0.8542,
+      "step": 2919
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.886178861788619e-07,
+      "loss": 0.7312,
+      "step": 2920
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.853658536585366e-07,
+      "loss": 0.7508,
+      "step": 2921
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.821138211382115e-07,
+      "loss": 0.7333,
+      "step": 2922
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.788617886178862e-07,
+      "loss": 0.724,
+      "step": 2923
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.75609756097561e-07,
+      "loss": 0.8194,
+      "step": 2924
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.723577235772358e-07,
+      "loss": 0.8316,
+      "step": 2925
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.691056910569106e-07,
+      "loss": 0.8204,
+      "step": 2926
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.658536585365854e-07,
+      "loss": 0.7637,
+      "step": 2927
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.626016260162602e-07,
+      "loss": 0.7841,
+      "step": 2928
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.59349593495935e-07,
+      "loss": 0.7749,
+      "step": 2929
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.560975609756098e-07,
+      "loss": 0.6429,
+      "step": 2930
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.528455284552846e-07,
+      "loss": 0.7723,
+      "step": 2931
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.495934959349594e-07,
+      "loss": 0.8123,
+      "step": 2932
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.463414634146342e-07,
+      "loss": 0.8228,
+      "step": 2933
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.43089430894309e-07,
+      "loss": 0.8006,
+      "step": 2934
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.398373983739838e-07,
+      "loss": 0.7785,
+      "step": 2935
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.365853658536586e-07,
+      "loss": 0.6263,
+      "step": 2936
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 0.8584,
+      "step": 2937
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.300813008130082e-07,
+      "loss": 0.7026,
+      "step": 2938
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.26829268292683e-07,
+      "loss": 0.8374,
+      "step": 2939
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.235772357723578e-07,
+      "loss": 0.6544,
+      "step": 2940
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.203252032520326e-07,
+      "loss": 0.7953,
+      "step": 2941
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.170731707317074e-07,
+      "loss": 0.9212,
+      "step": 2942
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.138211382113821e-07,
+      "loss": 0.7962,
+      "step": 2943
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.105691056910569e-07,
+      "loss": 0.8592,
+      "step": 2944
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.073170731707318e-07,
+      "loss": 0.823,
+      "step": 2945
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.040650406504065e-07,
+      "loss": 0.8196,
+      "step": 2946
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.008130081300813e-07,
+      "loss": 0.6645,
+      "step": 2947
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.975609756097562e-07,
+      "loss": 0.6687,
+      "step": 2948
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.943089430894309e-07,
+      "loss": 0.6434,
+      "step": 2949
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.910569105691057e-07,
+      "loss": 0.6705,
+      "step": 2950
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 0.7679,
+      "step": 2951
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.845528455284553e-07,
+      "loss": 0.6922,
+      "step": 2952
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.813008130081302e-07,
+      "loss": 0.7319,
+      "step": 2953
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.780487804878049e-07,
+      "loss": 0.7051,
+      "step": 2954
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.747967479674797e-07,
+      "loss": 0.7376,
+      "step": 2955
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.715447154471545e-07,
+      "loss": 0.7677,
+      "step": 2956
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.682926829268293e-07,
+      "loss": 0.6615,
+      "step": 2957
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.650406504065041e-07,
+      "loss": 0.7815,
+      "step": 2958
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.617886178861789e-07,
+      "loss": 0.7607,
+      "step": 2959
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.585365853658537e-07,
+      "loss": 0.9516,
+      "step": 2960
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.5528455284552854e-07,
+      "loss": 0.8445,
+      "step": 2961
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.520325203252033e-07,
+      "loss": 0.7309,
+      "step": 2962
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.4878048780487807e-07,
+      "loss": 0.6946,
+      "step": 2963
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.455284552845529e-07,
+      "loss": 0.7381,
+      "step": 2964
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.422764227642277e-07,
+      "loss": 0.8204,
+      "step": 2965
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.3902439024390246e-07,
+      "loss": 0.7802,
+      "step": 2966
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.357723577235773e-07,
+      "loss": 0.785,
+      "step": 2967
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.3252032520325204e-07,
+      "loss": 0.8116,
+      "step": 2968
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.292682926829269e-07,
+      "loss": 0.8145,
+      "step": 2969
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.2601626016260167e-07,
+      "loss": 0.8644,
+      "step": 2970
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.2276422764227643e-07,
+      "loss": 0.728,
+      "step": 2971
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.195121951219513e-07,
+      "loss": 0.6945,
+      "step": 2972
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.1626016260162606e-07,
+      "loss": 0.8824,
+      "step": 2973
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.130081300813008e-07,
+      "loss": 0.7172,
+      "step": 2974
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.0975609756097564e-07,
+      "loss": 0.7872,
+      "step": 2975
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.0650406504065046e-07,
+      "loss": 0.7322,
+      "step": 2976
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.0325203252032527e-07,
+      "loss": 0.7376,
+      "step": 2977
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.7925,
+      "step": 2978
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.967479674796748e-07,
+      "loss": 0.7431,
+      "step": 2979
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.9349593495934967e-07,
+      "loss": 0.7109,
+      "step": 2980
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 3.9024390243902443e-07,
+      "loss": 0.7647,
+      "step": 2981
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.869918699186992e-07,
+      "loss": 0.6241,
+      "step": 2982
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.83739837398374e-07,
+      "loss": 0.8348,
+      "step": 2983
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 3.804878048780488e-07,
+      "loss": 0.581,
+      "step": 2984
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.7723577235772364e-07,
+      "loss": 0.7858,
+      "step": 2985
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.739837398373984e-07,
+      "loss": 0.6889,
+      "step": 2986
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.7073170731707316e-07,
+      "loss": 0.7606,
+      "step": 2987
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.6747967479674803e-07,
+      "loss": 0.8214,
+      "step": 2988
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.642276422764228e-07,
+      "loss": 0.8374,
+      "step": 2989
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 3.6097560975609755e-07,
+      "loss": 0.9178,
+      "step": 2990
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.577235772357724e-07,
+      "loss": 0.6012,
+      "step": 2991
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.544715447154472e-07,
+      "loss": 0.6967,
+      "step": 2992
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 3.51219512195122e-07,
+      "loss": 0.7282,
+      "step": 2993
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.4796747967479676e-07,
+      "loss": 0.6737,
+      "step": 2994
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.447154471544716e-07,
+      "loss": 0.698,
+      "step": 2995
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.414634146341464e-07,
+      "loss": 0.5722,
+      "step": 2996
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.3821138211382116e-07,
+      "loss": 0.7234,
+      "step": 2997
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.349593495934959e-07,
+      "loss": 0.833,
+      "step": 2998
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 3.317073170731708e-07,
+      "loss": 0.774,
+      "step": 2999
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.2845528455284555e-07,
+      "loss": 0.7809,
+      "step": 3000
+    },
+    {
+      "epoch": 9.66,
+      "eval_loss": 2.016146659851074,
+      "eval_runtime": 137.7505,
+      "eval_samples_per_second": 4.399,
+      "eval_steps_per_second": 4.399,
+      "step": 3000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 0.8494,
+      "step": 3001
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 3.219512195121952e-07,
+      "loss": 0.8069,
+      "step": 3002
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.1869918699186994e-07,
+      "loss": 0.7794,
+      "step": 3003
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.1544715447154476e-07,
+      "loss": 0.673,
+      "step": 3004
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 3.121951219512195e-07,
+      "loss": 0.8277,
+      "step": 3005
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.0894308943089434e-07,
+      "loss": 0.7831,
+      "step": 3006
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.0569105691056915e-07,
+      "loss": 0.7977,
+      "step": 3007
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.024390243902439e-07,
+      "loss": 0.7016,
+      "step": 3008
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.9918699186991873e-07,
+      "loss": 0.7539,
+      "step": 3009
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.9593495934959355e-07,
+      "loss": 0.7822,
+      "step": 3010
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.926829268292683e-07,
+      "loss": 0.827,
+      "step": 3011
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.894308943089431e-07,
+      "loss": 0.8253,
+      "step": 3012
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.861788617886179e-07,
+      "loss": 0.7078,
+      "step": 3013
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.829268292682927e-07,
+      "loss": 0.8537,
+      "step": 3014
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.796747967479675e-07,
+      "loss": 0.8266,
+      "step": 3015
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.764227642276423e-07,
+      "loss": 0.7195,
+      "step": 3016
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.731707317073171e-07,
+      "loss": 0.7805,
+      "step": 3017
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.699186991869919e-07,
+      "loss": 0.7942,
+      "step": 3018
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 0.629,
+      "step": 3019
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.634146341463415e-07,
+      "loss": 0.8553,
+      "step": 3020
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.601626016260163e-07,
+      "loss": 0.6356,
+      "step": 3021
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.5691056910569107e-07,
+      "loss": 0.607,
+      "step": 3022
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.536585365853659e-07,
+      "loss": 0.5912,
+      "step": 3023
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.5040650406504064e-07,
+      "loss": 0.918,
+      "step": 3024
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.4715447154471546e-07,
+      "loss": 0.9479,
+      "step": 3025
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 0.7014,
+      "step": 3026
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.406504065040651e-07,
+      "loss": 0.7413,
+      "step": 3027
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.3739837398373985e-07,
+      "loss": 0.9223,
+      "step": 3028
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.3414634146341464e-07,
+      "loss": 0.7046,
+      "step": 3029
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.3089430894308946e-07,
+      "loss": 0.6951,
+      "step": 3030
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.2764227642276427e-07,
+      "loss": 0.729,
+      "step": 3031
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.2439024390243903e-07,
+      "loss": 0.7885,
+      "step": 3032
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.2113821138211385e-07,
+      "loss": 0.6811,
+      "step": 3033
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.1788617886178864e-07,
+      "loss": 0.8302,
+      "step": 3034
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.1463414634146345e-07,
+      "loss": 0.7208,
+      "step": 3035
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.1138211382113822e-07,
+      "loss": 0.8199,
+      "step": 3036
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.0813008130081303e-07,
+      "loss": 0.8576,
+      "step": 3037
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.0487804878048782e-07,
+      "loss": 0.7669,
+      "step": 3038
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.0162601626016264e-07,
+      "loss": 0.7237,
+      "step": 3039
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.983739837398374e-07,
+      "loss": 0.8623,
+      "step": 3040
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.9512195121951221e-07,
+      "loss": 0.7529,
+      "step": 3041
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.91869918699187e-07,
+      "loss": 0.8164,
+      "step": 3042
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.8861788617886182e-07,
+      "loss": 0.9109,
+      "step": 3043
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.8536585365853658e-07,
+      "loss": 0.7097,
+      "step": 3044
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.821138211382114e-07,
+      "loss": 0.8272,
+      "step": 3045
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.788617886178862e-07,
+      "loss": 0.6894,
+      "step": 3046
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.75609756097561e-07,
+      "loss": 0.8166,
+      "step": 3047
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.723577235772358e-07,
+      "loss": 0.7592,
+      "step": 3048
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.6910569105691058e-07,
+      "loss": 0.8048,
+      "step": 3049
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.658536585365854e-07,
+      "loss": 0.7394,
+      "step": 3050
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.6260162601626018e-07,
+      "loss": 0.686,
+      "step": 3051
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.5934959349593497e-07,
+      "loss": 0.705,
+      "step": 3052
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.5609756097560976e-07,
+      "loss": 0.769,
+      "step": 3053
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.5284552845528458e-07,
+      "loss": 0.7356,
+      "step": 3054
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.4959349593495936e-07,
+      "loss": 0.61,
+      "step": 3055
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.4634146341463415e-07,
+      "loss": 0.6832,
+      "step": 3056
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.4308943089430894e-07,
+      "loss": 0.7786,
+      "step": 3057
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3983739837398376e-07,
+      "loss": 0.7605,
+      "step": 3058
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3658536585365855e-07,
+      "loss": 0.8417,
+      "step": 3059
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3333333333333336e-07,
+      "loss": 0.6628,
+      "step": 3060
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.3008130081300815e-07,
+      "loss": 0.764,
+      "step": 3061
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.2682926829268294e-07,
+      "loss": 0.8534,
+      "step": 3062
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.2357723577235773e-07,
+      "loss": 0.8033,
+      "step": 3063
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.2032520325203254e-07,
+      "loss": 0.7617,
+      "step": 3064
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1707317073170732e-07,
+      "loss": 0.815,
+      "step": 3065
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1382113821138214e-07,
+      "loss": 0.7431,
+      "step": 3066
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.1056910569105692e-07,
+      "loss": 0.7154,
+      "step": 3067
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.0731707317073173e-07,
+      "loss": 0.7488,
+      "step": 3068
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.0406504065040652e-07,
+      "loss": 0.8053,
+      "step": 3069
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.0081300813008132e-07,
+      "loss": 0.7395,
+      "step": 3070
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 9.756097560975611e-08,
+      "loss": 0.6877,
+      "step": 3071
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 9.430894308943091e-08,
+      "loss": 0.7657,
+      "step": 3072
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 9.10569105691057e-08,
+      "loss": 0.7156,
+      "step": 3073
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 8.78048780487805e-08,
+      "loss": 0.8613,
+      "step": 3074
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 8.455284552845529e-08,
+      "loss": 0.664,
+      "step": 3075
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 8.130081300813009e-08,
+      "loss": 0.8291,
+      "step": 3076
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 7.804878048780488e-08,
+      "loss": 0.6977,
+      "step": 3077
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 7.479674796747968e-08,
+      "loss": 0.824,
+      "step": 3078
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 7.154471544715447e-08,
+      "loss": 0.7538,
+      "step": 3079
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.829268292682927e-08,
+      "loss": 0.7658,
+      "step": 3080
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.504065040650408e-08,
+      "loss": 0.675,
+      "step": 3081
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.178861788617886e-08,
+      "loss": 0.7113,
+      "step": 3082
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.853658536585366e-08,
+      "loss": 0.8497,
+      "step": 3083
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.528455284552846e-08,
+      "loss": 0.7055,
+      "step": 3084
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.203252032520326e-08,
+      "loss": 0.8109,
+      "step": 3085
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.8780487804878054e-08,
+      "loss": 0.8048,
+      "step": 3086
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.552845528455285e-08,
+      "loss": 0.7784,
+      "step": 3087
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.2276422764227645e-08,
+      "loss": 0.7212,
+      "step": 3088
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.902439024390244e-08,
+      "loss": 0.7696,
+      "step": 3089
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.5772357723577236e-08,
+      "loss": 0.8425,
+      "step": 3090
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.252032520325204e-08,
+      "loss": 0.8527,
+      "step": 3091
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.926829268292683e-08,
+      "loss": 0.6625,
+      "step": 3092
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.601626016260163e-08,
+      "loss": 0.5809,
+      "step": 3093
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.2764227642276425e-08,
+      "loss": 0.6577,
+      "step": 3094
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.951219512195122e-08,
+      "loss": 0.6845,
+      "step": 3095
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.626016260162602e-08,
+      "loss": 0.7351,
+      "step": 3096
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.3008130081300814e-08,
+      "loss": 0.6486,
+      "step": 3097
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.75609756097561e-09,
+      "loss": 0.8103,
+      "step": 3098
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 6.504065040650407e-09,
+      "loss": 0.6976,
+      "step": 3099
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.2520325203252036e-09,
+      "loss": 0.8007,
+      "step": 3100
+    },
+    {
+      "epoch": 9.98,
+      "step": 3100,
+      "total_flos": 6.305195734597632e+17,
+      "train_loss": 1.1158069036083837,
+      "train_runtime": 12559.8672,
+      "train_samples_per_second": 3.958,
+      "train_steps_per_second": 0.247
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3100,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 6.305195734597632e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}