diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.20868113522537562,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 4.6429,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 4.0302,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 3.6873,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 2.9784,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.9359,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.6265,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 1.3821,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.8773,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 0.8747,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.7558,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 0.606,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.543,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 0.5825,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.5753,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.6038,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.4458,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 0.4193,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.4561,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.4521,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.3784,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.105263157894737e-05,
+      "loss": 0.4241,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.4177,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.4199,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.2974,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3157894736842108e-05,
+      "loss": 0.4344,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.3558,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.3908,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.3488,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5263157894736846e-05,
+      "loss": 0.4594,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.3809,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.305,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.4816,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.4132,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.3801,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.433,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.3657,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9473684210526318e-05,
+      "loss": 0.4347,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.3113,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998353909465021e-05,
+      "loss": 0.4317,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9967078189300412e-05,
+      "loss": 0.3692,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.995061728395062e-05,
+      "loss": 0.494,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9934156378600823e-05,
+      "loss": 0.4323,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.991769547325103e-05,
+      "loss": 0.3354,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9901234567901237e-05,
+      "loss": 0.5693,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.988477366255144e-05,
+      "loss": 0.4703,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9868312757201648e-05,
+      "loss": 0.5352,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.3777,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.983539094650206e-05,
+      "loss": 0.3986,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9818930041152265e-05,
+      "loss": 0.4579,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9802469135802472e-05,
+      "loss": 0.346,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9786008230452676e-05,
+      "loss": 0.311,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9769547325102883e-05,
+      "loss": 0.4169,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 0.5365,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9736625514403294e-05,
+      "loss": 0.4696,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.97201646090535e-05,
+      "loss": 0.4226,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9703703703703704e-05,
+      "loss": 0.4278,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.968724279835391e-05,
+      "loss": 0.4218,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9670781893004115e-05,
+      "loss": 0.4335,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9654320987654322e-05,
+      "loss": 0.375,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.963786008230453e-05,
+      "loss": 0.4118,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9621399176954733e-05,
+      "loss": 0.4314,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.960493827160494e-05,
+      "loss": 0.4317,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9588477366255147e-05,
+      "loss": 0.2964,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9572016460905354e-05,
+      "loss": 0.4329,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.4416,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9539094650205764e-05,
+      "loss": 0.4064,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9522633744855968e-05,
+      "loss": 0.4792,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9506172839506175e-05,
+      "loss": 0.4869,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9489711934156382e-05,
+      "loss": 0.4992,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9473251028806586e-05,
+      "loss": 0.3628,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9456790123456793e-05,
+      "loss": 0.3652,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9440329218106996e-05,
+      "loss": 0.3615,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9423868312757203e-05,
+      "loss": 0.5364,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9407407407407407e-05,
+      "loss": 0.5206,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9390946502057614e-05,
+      "loss": 0.4478,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.937448559670782e-05,
+      "loss": 0.3366,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9358024691358025e-05,
+      "loss": 0.4752,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.934156378600823e-05,
+      "loss": 0.4703,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.932510288065844e-05,
+      "loss": 0.4441,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9308641975308646e-05,
+      "loss": 0.4702,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.929218106995885e-05,
+      "loss": 0.4248,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9275720164609056e-05,
+      "loss": 0.3839,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.3811,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9242798353909467e-05,
+      "loss": 0.3367,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9226337448559674e-05,
+      "loss": 0.3849,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9209876543209878e-05,
+      "loss": 0.4238,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9193415637860085e-05,
+      "loss": 0.3963,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9176954732510288e-05,
+      "loss": 0.4862,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9160493827160495e-05,
+      "loss": 0.4667,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.91440329218107e-05,
+      "loss": 0.4221,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9127572016460906e-05,
+      "loss": 0.3748,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 0.4173,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9094650205761317e-05,
+      "loss": 0.3978,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9078189300411524e-05,
+      "loss": 0.3904,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.906172839506173e-05,
+      "loss": 0.4062,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9045267489711938e-05,
+      "loss": 0.356,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.902880658436214e-05,
+      "loss": 0.4323,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.901234567901235e-05,
+      "loss": 0.3911,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8995884773662555e-05,
+      "loss": 0.4748,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.897942386831276e-05,
+      "loss": 0.3283,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8962962962962966e-05,
+      "loss": 0.3476,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.894650205761317e-05,
+      "loss": 0.395,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8930041152263377e-05,
+      "loss": 0.4423,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.891358024691358e-05,
+      "loss": 0.4262,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8897119341563787e-05,
+      "loss": 0.3841,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8880658436213994e-05,
+      "loss": 0.3878,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8864197530864198e-05,
+      "loss": 0.445,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8847736625514405e-05,
+      "loss": 0.3299,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.883127572016461e-05,
+      "loss": 0.3767,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8814814814814816e-05,
+      "loss": 0.2873,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8798353909465023e-05,
+      "loss": 0.4953,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.878189300411523e-05,
+      "loss": 0.4015,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 0.4362,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.874897119341564e-05,
+      "loss": 0.4397,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8732510288065847e-05,
+      "loss": 0.3356,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.871604938271605e-05,
+      "loss": 0.383,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8699588477366258e-05,
+      "loss": 0.355,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.868312757201646e-05,
+      "loss": 0.4438,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.4318,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8650205761316872e-05,
+      "loss": 0.3627,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.863374485596708e-05,
+      "loss": 0.3563,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8617283950617286e-05,
+      "loss": 0.3352,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.860082304526749e-05,
+      "loss": 0.3308,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8584362139917697e-05,
+      "loss": 0.4253,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.85679012345679e-05,
+      "loss": 0.3725,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8551440329218108e-05,
+      "loss": 0.4191,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8534979423868315e-05,
+      "loss": 0.3796,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.3464,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8502057613168725e-05,
+      "loss": 0.4045,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8485596707818932e-05,
+      "loss": 0.4845,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.846913580246914e-05,
+      "loss": 0.3837,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8452674897119343e-05,
+      "loss": 0.4572,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.843621399176955e-05,
+      "loss": 0.4051,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8419753086419754e-05,
+      "loss": 0.397,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.840329218106996e-05,
+      "loss": 0.302,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8386831275720164e-05,
+      "loss": 0.3247,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 0.3571,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.835390946502058e-05,
+      "loss": 0.388,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8337448559670782e-05,
+      "loss": 0.431,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.832098765432099e-05,
+      "loss": 0.4123,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8304526748971193e-05,
+      "loss": 0.3475,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.82880658436214e-05,
+      "loss": 0.3351,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8271604938271607e-05,
+      "loss": 0.3518,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8255144032921814e-05,
+      "loss": 0.3359,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8238683127572017e-05,
+      "loss": 0.3153,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 0.3294,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.820576131687243e-05,
+      "loss": 0.3986,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8189300411522635e-05,
+      "loss": 0.3623,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8172839506172842e-05,
+      "loss": 0.3611,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8156378600823046e-05,
+      "loss": 0.3021,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8139917695473253e-05,
+      "loss": 0.4334,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.812345679012346e-05,
+      "loss": 0.4103,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8106995884773663e-05,
+      "loss": 0.3672,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.809053497942387e-05,
+      "loss": 0.4632,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.3634,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.805761316872428e-05,
+      "loss": 0.4697,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8041152263374485e-05,
+      "loss": 0.465,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 0.2864,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.80082304526749e-05,
+      "loss": 0.4012,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7991769547325106e-05,
+      "loss": 0.3549,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.797530864197531e-05,
+      "loss": 0.3591,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7958847736625516e-05,
+      "loss": 0.3735,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7942386831275723e-05,
+      "loss": 0.4396,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7925925925925927e-05,
+      "loss": 0.3875,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7909465020576134e-05,
+      "loss": 0.3084,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7893004115226338e-05,
+      "loss": 0.3325,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7876543209876545e-05,
+      "loss": 0.3039,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.786008230452675e-05,
+      "loss": 0.3332,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7843621399176955e-05,
+      "loss": 0.3569,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7827160493827162e-05,
+      "loss": 0.3892,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7810699588477366e-05,
+      "loss": 0.4047,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7794238683127573e-05,
+      "loss": 0.3696,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.3527,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7761316872427984e-05,
+      "loss": 0.2831,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.774485596707819e-05,
+      "loss": 0.3992,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7728395061728398e-05,
+      "loss": 0.3194,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.77119341563786e-05,
+      "loss": 0.3828,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.769547325102881e-05,
+      "loss": 0.3952,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7679012345679015e-05,
+      "loss": 0.3509,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.766255144032922e-05,
+      "loss": 0.3846,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7646090534979426e-05,
+      "loss": 0.3908,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7629629629629633e-05,
+      "loss": 0.3552,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7613168724279837e-05,
+      "loss": 0.4473,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7596707818930044e-05,
+      "loss": 0.3479,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7580246913580247e-05,
+      "loss": 0.4019,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7563786008230454e-05,
+      "loss": 0.3951,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7547325102880658e-05,
+      "loss": 0.4387,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7530864197530865e-05,
+      "loss": 0.4799,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7514403292181072e-05,
+      "loss": 0.3532,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7497942386831276e-05,
+      "loss": 0.3819,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.3474,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.746502057613169e-05,
+      "loss": 0.2974,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7448559670781893e-05,
+      "loss": 0.4489,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.74320987654321e-05,
+      "loss": 0.3348,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7415637860082307e-05,
+      "loss": 0.4396,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.739917695473251e-05,
+      "loss": 0.374,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7382716049382718e-05,
+      "loss": 0.3216,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7366255144032925e-05,
+      "loss": 0.3076,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.734979423868313e-05,
+      "loss": 0.3216,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.3385,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.731687242798354e-05,
+      "loss": 0.4436,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7300411522633746e-05,
+      "loss": 0.4418,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 0.3815,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7267489711934157e-05,
+      "loss": 0.4089,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7251028806584364e-05,
+      "loss": 0.3483,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7234567901234568e-05,
+      "loss": 0.3427,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7218106995884775e-05,
+      "loss": 0.3978,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.720164609053498e-05,
+      "loss": 0.3463,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7185185185185185e-05,
+      "loss": 0.3599,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7168724279835392e-05,
+      "loss": 0.3075,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.71522633744856e-05,
+      "loss": 0.3821,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7135802469135806e-05,
+      "loss": 0.4101,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.711934156378601e-05,
+      "loss": 0.2658,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7102880658436217e-05,
+      "loss": 0.4748,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.708641975308642e-05,
+      "loss": 0.3602,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7069958847736628e-05,
+      "loss": 0.3989,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.705349794238683e-05,
+      "loss": 0.2967,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.3614,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7020576131687242e-05,
+      "loss": 0.3685,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.700411522633745e-05,
+      "loss": 0.42,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6987654320987656e-05,
+      "loss": 0.4236,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.697119341563786e-05,
+      "loss": 0.3404,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6954732510288067e-05,
+      "loss": 0.4477,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6938271604938274e-05,
+      "loss": 0.5353,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.692181069958848e-05,
+      "loss": 0.3449,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6905349794238684e-05,
+      "loss": 0.3362,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.278,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.68724279835391e-05,
+      "loss": 0.4077,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6855967078189302e-05,
+      "loss": 0.3874,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.683950617283951e-05,
+      "loss": 0.2847,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6823045267489713e-05,
+      "loss": 0.4704,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.680658436213992e-05,
+      "loss": 0.3949,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 0.492,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.677366255144033e-05,
+      "loss": 0.3415,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6757201646090537e-05,
+      "loss": 0.4086,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.674074074074074e-05,
+      "loss": 0.3879,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6724279835390948e-05,
+      "loss": 0.4109,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.670781893004115e-05,
+      "loss": 0.3051,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.669135802469136e-05,
+      "loss": 0.3553,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6674897119341566e-05,
+      "loss": 0.3551,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6658436213991773e-05,
+      "loss": 0.36,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6641975308641976e-05,
+      "loss": 0.3106,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6625514403292183e-05,
+      "loss": 0.3763,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.660905349794239e-05,
+      "loss": 0.4088,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.3246,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.65761316872428e-05,
+      "loss": 0.3343,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6559670781893005e-05,
+      "loss": 0.3859,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.654320987654321e-05,
+      "loss": 0.3355,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6526748971193415e-05,
+      "loss": 0.3924,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6510288065843622e-05,
+      "loss": 0.2973,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.649382716049383e-05,
+      "loss": 0.3494,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6477366255144033e-05,
+      "loss": 0.2753,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.646090534979424e-05,
+      "loss": 0.4038,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 0.3741,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.642798353909465e-05,
+      "loss": 0.3988,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6411522633744858e-05,
+      "loss": 0.3988,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6395061728395065e-05,
+      "loss": 0.4406,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6378600823045268e-05,
+      "loss": 0.396,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6362139917695475e-05,
+      "loss": 0.4425,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6345679012345682e-05,
+      "loss": 0.3474,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6329218106995886e-05,
+      "loss": 0.3294,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6312757201646093e-05,
+      "loss": 0.3623,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.4464,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6279835390946504e-05,
+      "loss": 0.3757,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.626337448559671e-05,
+      "loss": 0.447,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6246913580246914e-05,
+      "loss": 0.342,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.623045267489712e-05,
+      "loss": 0.3863,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6213991769547325e-05,
+      "loss": 0.2998,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6197530864197532e-05,
+      "loss": 0.3208,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6181069958847736e-05,
+      "loss": 0.3717,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6164609053497943e-05,
+      "loss": 0.4215,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.614814814814815e-05,
+      "loss": 0.3579,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6131687242798357e-05,
+      "loss": 0.3921,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.611522633744856e-05,
+      "loss": 0.4227,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6098765432098767e-05,
+      "loss": 0.3351,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6082304526748974e-05,
+      "loss": 0.4863,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6065843621399178e-05,
+      "loss": 0.3907,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6049382716049385e-05,
+      "loss": 0.3153,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.603292181069959e-05,
+      "loss": 0.3209,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6016460905349796e-05,
+      "loss": 0.3294,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.4566,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5983539094650206e-05,
+      "loss": 0.3075,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5967078189300413e-05,
+      "loss": 0.3337,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5950617283950617e-05,
+      "loss": 0.324,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5934156378600824e-05,
+      "loss": 0.3143,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5917695473251028e-05,
+      "loss": 0.3538,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5901234567901235e-05,
+      "loss": 0.3179,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.588477366255144e-05,
+      "loss": 0.3839,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.586831275720165e-05,
+      "loss": 0.39,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5851851851851852e-05,
+      "loss": 0.3121,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.583539094650206e-05,
+      "loss": 0.3825,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5818930041152266e-05,
+      "loss": 0.4305,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.580246913580247e-05,
+      "loss": 0.4055,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5786008230452677e-05,
+      "loss": 0.5461,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5769547325102884e-05,
+      "loss": 0.3162,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5753086419753088e-05,
+      "loss": 0.33,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5736625514403295e-05,
+      "loss": 0.3268,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5720164609053498e-05,
+      "loss": 0.3049,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5703703703703705e-05,
+      "loss": 0.4057,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.568724279835391e-05,
+      "loss": 0.2935,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5670781893004116e-05,
+      "loss": 0.3972,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.565432098765432e-05,
+      "loss": 0.3419,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5637860082304527e-05,
+      "loss": 0.4633,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5621399176954734e-05,
+      "loss": 0.3043,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.560493827160494e-05,
+      "loss": 0.2642,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5588477366255144e-05,
+      "loss": 0.4312,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.557201646090535e-05,
+      "loss": 0.2683,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.4104,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5539094650205762e-05,
+      "loss": 0.3223,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.552263374485597e-05,
+      "loss": 0.4913,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5506172839506176e-05,
+      "loss": 0.3874,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.548971193415638e-05,
+      "loss": 0.3457,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5473251028806587e-05,
+      "loss": 0.3643,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.545679012345679e-05,
+      "loss": 0.3625,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5440329218106997e-05,
+      "loss": 0.4309,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.54238683127572e-05,
+      "loss": 0.3758,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.3136,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5390946502057615e-05,
+      "loss": 0.3217,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.537448559670782e-05,
+      "loss": 0.3894,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5358024691358026e-05,
+      "loss": 0.4406,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5341563786008233e-05,
+      "loss": 0.3795,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5325102880658436e-05,
+      "loss": 0.3859,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5308641975308643e-05,
+      "loss": 0.3593,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.529218106995885e-05,
+      "loss": 0.3847,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5275720164609057e-05,
+      "loss": 0.3565,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.525925925925926e-05,
+      "loss": 0.3368,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5242798353909466e-05,
+      "loss": 0.4562,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5226337448559672e-05,
+      "loss": 0.3737,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5209876543209879e-05,
+      "loss": 0.2895,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5193415637860082e-05,
+      "loss": 0.4131,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.517695473251029e-05,
+      "loss": 0.373,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5160493827160495e-05,
+      "loss": 0.4048,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.51440329218107e-05,
+      "loss": 0.3356,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5127572016460907e-05,
+      "loss": 0.2875,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.3596,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.509465020576132e-05,
+      "loss": 0.3437,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5078189300411523e-05,
+      "loss": 0.3026,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.506172839506173e-05,
+      "loss": 0.4497,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5045267489711934e-05,
+      "loss": 0.3365,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.502880658436214e-05,
+      "loss": 0.3269,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5012345679012348e-05,
+      "loss": 0.3567,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4995884773662553e-05,
+      "loss": 0.2776,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4979423868312758e-05,
+      "loss": 0.4037,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4962962962962964e-05,
+      "loss": 0.4409,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.494650205761317e-05,
+      "loss": 0.3843,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4930041152263374e-05,
+      "loss": 0.41,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4913580246913581e-05,
+      "loss": 0.3771,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4897119341563788e-05,
+      "loss": 0.3469,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4880658436213992e-05,
+      "loss": 0.3783,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4864197530864199e-05,
+      "loss": 0.4191,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4847736625514404e-05,
+      "loss": 0.2527,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4831275720164611e-05,
+      "loss": 0.3348,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.4326,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4798353909465022e-05,
+      "loss": 0.3648,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4781893004115229e-05,
+      "loss": 0.4717,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4765432098765433e-05,
+      "loss": 0.3187,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.474897119341564e-05,
+      "loss": 0.3988,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4732510288065845e-05,
+      "loss": 0.4229,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.471604938271605e-05,
+      "loss": 0.3688,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4699588477366256e-05,
+      "loss": 0.2868,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4683127572016463e-05,
+      "loss": 0.3285,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.2391,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4650205761316873e-05,
+      "loss": 0.3412,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.463374485596708e-05,
+      "loss": 0.314,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4617283950617284e-05,
+      "loss": 0.32,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4600823045267491e-05,
+      "loss": 0.3806,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4584362139917696e-05,
+      "loss": 0.4183,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4567901234567903e-05,
+      "loss": 0.2853,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4551440329218107e-05,
+      "loss": 0.3904,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4534979423868314e-05,
+      "loss": 0.4316,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4518518518518521e-05,
+      "loss": 0.383,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4502057613168725e-05,
+      "loss": 0.3435,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4485596707818932e-05,
+      "loss": 0.4828,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4469135802469137e-05,
+      "loss": 0.4114,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4452674897119342e-05,
+      "loss": 0.359,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4436213991769548e-05,
+      "loss": 0.4045,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4419753086419755e-05,
+      "loss": 0.4491,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4403292181069962e-05,
+      "loss": 0.3601,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4386831275720165e-05,
+      "loss": 0.3834,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4370370370370372e-05,
+      "loss": 0.3441,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4353909465020578e-05,
+      "loss": 0.251,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4337448559670783e-05,
+      "loss": 0.2712,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4320987654320988e-05,
+      "loss": 0.3346,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4304526748971195e-05,
+      "loss": 0.273,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4288065843621399e-05,
+      "loss": 0.344,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4271604938271606e-05,
+      "loss": 0.3425,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4255144032921813e-05,
+      "loss": 0.3792,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4238683127572017e-05,
+      "loss": 0.433,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.3377,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4205761316872429e-05,
+      "loss": 0.3571,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4189300411522636e-05,
+      "loss": 0.3764,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.417283950617284e-05,
+      "loss": 0.3217,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4156378600823047e-05,
+      "loss": 0.3562,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4139917695473254e-05,
+      "loss": 0.3375,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4123456790123457e-05,
+      "loss": 0.3139,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4106995884773664e-05,
+      "loss": 0.3621,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.409053497942387e-05,
+      "loss": 0.2542,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.4264,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.405761316872428e-05,
+      "loss": 0.4221,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4041152263374487e-05,
+      "loss": 0.3437,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4024691358024694e-05,
+      "loss": 0.3532,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4008230452674898e-05,
+      "loss": 0.3088,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3991769547325105e-05,
+      "loss": 0.3715,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3975308641975309e-05,
+      "loss": 0.3912,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3958847736625516e-05,
+      "loss": 0.3602,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3942386831275721e-05,
+      "loss": 0.3664,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3925925925925928e-05,
+      "loss": 0.2906,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3909465020576133e-05,
+      "loss": 0.2448,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3893004115226339e-05,
+      "loss": 0.33,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3876543209876546e-05,
+      "loss": 0.3771,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.386008230452675e-05,
+      "loss": 0.2936,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3843621399176956e-05,
+      "loss": 0.2752,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3827160493827162e-05,
+      "loss": 0.3374,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3810699588477367e-05,
+      "loss": 0.2791,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3794238683127572e-05,
+      "loss": 0.2697,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 0.3949,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3761316872427986e-05,
+      "loss": 0.3241,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.374485596707819e-05,
+      "loss": 0.3209,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3728395061728397e-05,
+      "loss": 0.3522,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.37119341563786e-05,
+      "loss": 0.3303,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3695473251028808e-05,
+      "loss": 0.2844,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3679012345679013e-05,
+      "loss": 0.3065,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.366255144032922e-05,
+      "loss": 0.3287,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3646090534979425e-05,
+      "loss": 0.3386,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.3215,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3613168724279838e-05,
+      "loss": 0.3374,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3596707818930041e-05,
+      "loss": 0.294,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3580246913580248e-05,
+      "loss": 0.251,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3563786008230454e-05,
+      "loss": 0.3254,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3547325102880659e-05,
+      "loss": 0.3768,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3530864197530866e-05,
+      "loss": 0.265,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3514403292181071e-05,
+      "loss": 0.2708,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3497942386831278e-05,
+      "loss": 0.457,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3481481481481482e-05,
+      "loss": 0.3608,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3465020576131689e-05,
+      "loss": 0.2924,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3448559670781893e-05,
+      "loss": 0.3382,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.34320987654321e-05,
+      "loss": 0.3109,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3415637860082307e-05,
+      "loss": 0.3654,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3399176954732512e-05,
+      "loss": 0.2796,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3382716049382717e-05,
+      "loss": 0.4257,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3366255144032923e-05,
+      "loss": 0.3077,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.334979423868313e-05,
+      "loss": 0.3237,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4051,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.331687242798354e-05,
+      "loss": 0.3901,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3300411522633746e-05,
+      "loss": 0.3818,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3283950617283951e-05,
+      "loss": 0.3582,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3267489711934158e-05,
+      "loss": 0.3504,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3251028806584363e-05,
+      "loss": 0.2986,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.323456790123457e-05,
+      "loss": 0.2762,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3218106995884774e-05,
+      "loss": 0.2953,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3201646090534981e-05,
+      "loss": 0.2667,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3185185185185185e-05,
+      "loss": 0.3169,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3168724279835392e-05,
+      "loss": 0.3389,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3152263374485599e-05,
+      "loss": 0.3123,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.3135802469135804e-05,
+      "loss": 0.3135,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.311934156378601e-05,
+      "loss": 0.4142,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3102880658436215e-05,
+      "loss": 0.2955,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3086419753086422e-05,
+      "loss": 0.2994,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3069958847736625e-05,
+      "loss": 0.3906,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3053497942386832e-05,
+      "loss": 0.2669,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.303703703703704e-05,
+      "loss": 0.274,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3020576131687243e-05,
+      "loss": 0.3334,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.300411522633745e-05,
+      "loss": 0.3907,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2987654320987655e-05,
+      "loss": 0.2525,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2971193415637862e-05,
+      "loss": 0.2961,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2954732510288066e-05,
+      "loss": 0.3694,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2938271604938273e-05,
+      "loss": 0.3302,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2921810699588477e-05,
+      "loss": 0.3697,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2905349794238684e-05,
+      "loss": 0.3838,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 0.3076,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2872427983539096e-05,
+      "loss": 0.3233,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2855967078189301e-05,
+      "loss": 0.3489,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2839506172839507e-05,
+      "loss": 0.3197,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2823045267489714e-05,
+      "loss": 0.2732,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2806584362139917e-05,
+      "loss": 0.3938,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2790123456790124e-05,
+      "loss": 0.323,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2773662551440331e-05,
+      "loss": 0.3817,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2757201646090535e-05,
+      "loss": 0.3459,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.3012,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2724279835390947e-05,
+      "loss": 0.3353,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2707818930041154e-05,
+      "loss": 0.3912,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2691358024691358e-05,
+      "loss": 0.4021,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2674897119341565e-05,
+      "loss": 0.2983,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2658436213991772e-05,
+      "loss": 0.4075,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2641975308641976e-05,
+      "loss": 0.3025,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2625514403292183e-05,
+      "loss": 0.3814,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2609053497942388e-05,
+      "loss": 0.2446,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.3114,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2576131687242799e-05,
+      "loss": 0.3111,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2559670781893006e-05,
+      "loss": 0.2996,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2543209876543213e-05,
+      "loss": 0.3598,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2526748971193416e-05,
+      "loss": 0.3145,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2510288065843623e-05,
+      "loss": 0.3098,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2493827160493827e-05,
+      "loss": 0.3384,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2477366255144034e-05,
+      "loss": 0.3427,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.246090534979424e-05,
+      "loss": 0.3062,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.2949,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.242798353909465e-05,
+      "loss": 0.3231,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2411522633744857e-05,
+      "loss": 0.255,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2395061728395064e-05,
+      "loss": 0.3093,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2378600823045268e-05,
+      "loss": 0.2908,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2362139917695475e-05,
+      "loss": 0.3488,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.234567901234568e-05,
+      "loss": 0.3822,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2329218106995885e-05,
+      "loss": 0.269,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.231275720164609e-05,
+      "loss": 0.3207,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2296296296296298e-05,
+      "loss": 0.2923,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2279835390946505e-05,
+      "loss": 0.3483,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2263374485596708e-05,
+      "loss": 0.309,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2246913580246915e-05,
+      "loss": 0.4025,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2230452674897119e-05,
+      "loss": 0.3327,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2213991769547326e-05,
+      "loss": 0.3128,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2197530864197531e-05,
+      "loss": 0.2384,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2181069958847738e-05,
+      "loss": 0.2168,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2164609053497944e-05,
+      "loss": 0.2493,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.3698,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2131687242798356e-05,
+      "loss": 0.3376,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.211522633744856e-05,
+      "loss": 0.3142,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2098765432098767e-05,
+      "loss": 0.2496,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2082304526748972e-05,
+      "loss": 0.3737,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2065843621399177e-05,
+      "loss": 0.3364,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2049382716049384e-05,
+      "loss": 0.3678,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.203292181069959e-05,
+      "loss": 0.3217,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2016460905349797e-05,
+      "loss": 0.2747,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2e-05,
+      "loss": 0.3137,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1983539094650207e-05,
+      "loss": 0.2929,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1967078189300411e-05,
+      "loss": 0.3775,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1950617283950618e-05,
+      "loss": 0.338,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1934156378600823e-05,
+      "loss": 0.2964,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.191769547325103e-05,
+      "loss": 0.3562,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1901234567901236e-05,
+      "loss": 0.3784,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1884773662551441e-05,
+      "loss": 0.2934,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1868312757201648e-05,
+      "loss": 0.3429,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.3057,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1835390946502059e-05,
+      "loss": 0.3349,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1818930041152264e-05,
+      "loss": 0.2922,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.180246913580247e-05,
+      "loss": 0.2859,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1786008230452676e-05,
+      "loss": 0.2864,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1769547325102882e-05,
+      "loss": 0.2615,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1753086419753089e-05,
+      "loss": 0.3219,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1736625514403292e-05,
+      "loss": 0.362,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.17201646090535e-05,
+      "loss": 0.29,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1703703703703703e-05,
+      "loss": 0.2841,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.168724279835391e-05,
+      "loss": 0.2978,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1670781893004117e-05,
+      "loss": 0.3601,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1654320987654322e-05,
+      "loss": 0.275,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1637860082304528e-05,
+      "loss": 0.2899,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1621399176954733e-05,
+      "loss": 0.27,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.160493827160494e-05,
+      "loss": 0.3738,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1588477366255144e-05,
+      "loss": 0.2776,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.157201646090535e-05,
+      "loss": 0.3055,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.3643,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1539094650205761e-05,
+      "loss": 0.287,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1522633744855968e-05,
+      "loss": 0.33,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1506172839506174e-05,
+      "loss": 0.2763,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.148971193415638e-05,
+      "loss": 0.3173,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1473251028806584e-05,
+      "loss": 0.3029,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1456790123456791e-05,
+      "loss": 0.2915,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1440329218106997e-05,
+      "loss": 0.3493,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1423868312757202e-05,
+      "loss": 0.2869,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1407407407407409e-05,
+      "loss": 0.3187,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1390946502057614e-05,
+      "loss": 0.3904,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.137448559670782e-05,
+      "loss": 0.344,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1358024691358025e-05,
+      "loss": 0.2748,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1341563786008232e-05,
+      "loss": 0.3612,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1325102880658436e-05,
+      "loss": 0.2807,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1308641975308643e-05,
+      "loss": 0.2302,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.129218106995885e-05,
+      "loss": 0.307,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1275720164609055e-05,
+      "loss": 0.3066,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.125925925925926e-05,
+      "loss": 0.3703,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1242798353909466e-05,
+      "loss": 0.3456,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1226337448559673e-05,
+      "loss": 0.2843,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1209876543209876e-05,
+      "loss": 0.3352,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1193415637860083e-05,
+      "loss": 0.3548,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.117695473251029e-05,
+      "loss": 0.3126,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1160493827160494e-05,
+      "loss": 0.3779,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1144032921810701e-05,
+      "loss": 0.3095,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1127572016460906e-05,
+      "loss": 0.2439,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.3911,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1094650205761317e-05,
+      "loss": 0.3073,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1078189300411524e-05,
+      "loss": 0.3066,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1061728395061728e-05,
+      "loss": 0.2656,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1045267489711935e-05,
+      "loss": 0.3713,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1028806584362142e-05,
+      "loss": 0.3448,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1012345679012347e-05,
+      "loss": 0.2975,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0995884773662552e-05,
+      "loss": 0.3386,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0979423868312758e-05,
+      "loss": 0.4122,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0962962962962965e-05,
+      "loss": 0.2241,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0946502057613168e-05,
+      "loss": 0.2832,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0930041152263375e-05,
+      "loss": 0.3319,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0913580246913582e-05,
+      "loss": 0.3703,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0897119341563786e-05,
+      "loss": 0.2612,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0880658436213993e-05,
+      "loss": 0.2687,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0864197530864198e-05,
+      "loss": 0.2904,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0847736625514405e-05,
+      "loss": 0.2279,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0831275720164609e-05,
+      "loss": 0.3174,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0814814814814816e-05,
+      "loss": 0.2877,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0798353909465023e-05,
+      "loss": 0.2822,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0781893004115227e-05,
+      "loss": 0.3628,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0765432098765434e-05,
+      "loss": 0.2726,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0748971193415639e-05,
+      "loss": 0.2888,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0732510288065844e-05,
+      "loss": 0.3151,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.071604938271605e-05,
+      "loss": 0.2477,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0699588477366257e-05,
+      "loss": 0.2657,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0683127572016464e-05,
+      "loss": 0.3014,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.3043,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0650205761316874e-05,
+      "loss": 0.2718,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0633744855967078e-05,
+      "loss": 0.2174,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0617283950617285e-05,
+      "loss": 0.3945,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.060082304526749e-05,
+      "loss": 0.2529,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0584362139917697e-05,
+      "loss": 0.2657,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0567901234567901e-05,
+      "loss": 0.3098,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0551440329218108e-05,
+      "loss": 0.2947,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0534979423868315e-05,
+      "loss": 0.2904,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0518518518518519e-05,
+      "loss": 0.3572,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0502057613168726e-05,
+      "loss": 0.3669,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0485596707818931e-05,
+      "loss": 0.24,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0469135802469136e-05,
+      "loss": 0.2714,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0452674897119342e-05,
+      "loss": 0.2905,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0436213991769549e-05,
+      "loss": 0.2968,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0419753086419756e-05,
+      "loss": 0.2899,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.040329218106996e-05,
+      "loss": 0.2417,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0386831275720166e-05,
+      "loss": 0.3465,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.287,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0353909465020577e-05,
+      "loss": 0.3227,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0337448559670782e-05,
+      "loss": 0.3196,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.032098765432099e-05,
+      "loss": 0.2929,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0304526748971195e-05,
+      "loss": 0.2584,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.02880658436214e-05,
+      "loss": 0.3444,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0271604938271607e-05,
+      "loss": 0.2931,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.025514403292181e-05,
+      "loss": 0.2424,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0238683127572018e-05,
+      "loss": 0.3464,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 0.2856,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0205761316872428e-05,
+      "loss": 0.2449,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0189300411522634e-05,
+      "loss": 0.362,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.017283950617284e-05,
+      "loss": 0.2216,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0156378600823048e-05,
+      "loss": 0.2976,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0139917695473251e-05,
+      "loss": 0.2885,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0123456790123458e-05,
+      "loss": 0.3294,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0106995884773662e-05,
+      "loss": 0.2955,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0090534979423869e-05,
+      "loss": 0.201,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0074074074074074e-05,
+      "loss": 0.2836,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0057613168724281e-05,
+      "loss": 0.399,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0041152263374487e-05,
+      "loss": 0.2547,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0024691358024692e-05,
+      "loss": 0.3039,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.0008230452674899e-05,
+      "loss": 0.3337,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.991769547325104e-06,
+      "loss": 0.2599,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97530864197531e-06,
+      "loss": 0.3843,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958847736625515e-06,
+      "loss": 0.2523,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.94238683127572e-06,
+      "loss": 0.2996,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.925925925925927e-06,
+      "loss": 0.283,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.909465020576133e-06,
+      "loss": 0.321,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.893004115226338e-06,
+      "loss": 0.2275,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 0.3076,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.86008230452675e-06,
+      "loss": 0.2312,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843621399176956e-06,
+      "loss": 0.2874,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827160493827161e-06,
+      "loss": 0.2963,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810699588477366e-06,
+      "loss": 0.3082,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.794238683127573e-06,
+      "loss": 0.2803,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.2541,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.761316872427984e-06,
+      "loss": 0.2783,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.744855967078191e-06,
+      "loss": 0.2458,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.728395061728396e-06,
+      "loss": 0.222,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711934156378602e-06,
+      "loss": 0.3674,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695473251028807e-06,
+      "loss": 0.3086,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679012345679012e-06,
+      "loss": 0.309,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66255144032922e-06,
+      "loss": 0.2571,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.646090534979425e-06,
+      "loss": 0.3018,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.2583,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.613168724279837e-06,
+      "loss": 0.2976,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.596707818930042e-06,
+      "loss": 0.3245,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.580246913580248e-06,
+      "loss": 0.3578,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563786008230453e-06,
+      "loss": 0.3159,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547325102880658e-06,
+      "loss": 0.2892,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.530864197530865e-06,
+      "loss": 0.3599,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.51440329218107e-06,
+      "loss": 0.3214,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.497942386831278e-06,
+      "loss": 0.2542,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 0.3568,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.465020576131688e-06,
+      "loss": 0.2912,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.448559670781894e-06,
+      "loss": 0.2713,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.432098765432099e-06,
+      "loss": 0.2419,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.415637860082304e-06,
+      "loss": 0.322,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.399176954732511e-06,
+      "loss": 0.294,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 0.3007,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.366255144032924e-06,
+      "loss": 0.3304,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.349794238683129e-06,
+      "loss": 0.3429,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.3873,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.31687242798354e-06,
+      "loss": 0.2795,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.300411522633745e-06,
+      "loss": 0.3081,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.28395061728395e-06,
+      "loss": 0.3892,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.267489711934157e-06,
+      "loss": 0.3009,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.251028806584363e-06,
+      "loss": 0.4026,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.23456790123457e-06,
+      "loss": 0.502,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.218106995884775e-06,
+      "loss": 0.3237,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.20164609053498e-06,
+      "loss": 0.3562,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.3351,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.168724279835391e-06,
+      "loss": 0.3277,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.152263374485596e-06,
+      "loss": 0.4425,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.135802469135803e-06,
+      "loss": 0.3134,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.119341563786009e-06,
+      "loss": 0.3619,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.102880658436216e-06,
+      "loss": 0.1908,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.086419753086421e-06,
+      "loss": 0.3832,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.069958847736626e-06,
+      "loss": 0.3215,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.053497942386832e-06,
+      "loss": 0.3136,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.037037037037037e-06,
+      "loss": 0.2723,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.020576131687242e-06,
+      "loss": 0.2879,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.00411522633745e-06,
+      "loss": 0.2987,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.987654320987655e-06,
+      "loss": 0.2797,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.971193415637862e-06,
+      "loss": 0.2468,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.954732510288067e-06,
+      "loss": 0.3714,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.938271604938272e-06,
+      "loss": 0.3067,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.921810699588478e-06,
+      "loss": 0.2903,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.905349794238683e-06,
+      "loss": 0.3745,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.2607,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.872427983539095e-06,
+      "loss": 0.3692,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.8559670781893e-06,
+      "loss": 0.304,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.839506172839508e-06,
+      "loss": 0.3689,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.823045267489713e-06,
+      "loss": 0.3057,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.806584362139918e-06,
+      "loss": 0.2546,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.790123456790124e-06,
+      "loss": 0.3264,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.773662551440329e-06,
+      "loss": 0.2931,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.757201646090536e-06,
+      "loss": 0.3151,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.740740740740741e-06,
+      "loss": 0.3108,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.724279835390947e-06,
+      "loss": 0.3067,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.707818930041154e-06,
+      "loss": 0.2739,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.691358024691359e-06,
+      "loss": 0.2993,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.674897119341564e-06,
+      "loss": 0.2497,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.65843621399177e-06,
+      "loss": 0.323,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 0.2899,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.625514403292182e-06,
+      "loss": 0.2228,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.609053497942387e-06,
+      "loss": 0.2838,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.592592592592593e-06,
+      "loss": 0.315,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.5761316872428e-06,
+      "loss": 0.3024,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.559670781893005e-06,
+      "loss": 0.2145,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.54320987654321e-06,
+      "loss": 0.2835,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.526748971193416e-06,
+      "loss": 0.3081,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.510288065843621e-06,
+      "loss": 0.2786,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.493827160493828e-06,
+      "loss": 0.3021,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.477366255144033e-06,
+      "loss": 0.2951,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.46090534979424e-06,
+      "loss": 0.2849,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 0.2949,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.427983539094651e-06,
+      "loss": 0.2967,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.411522633744856e-06,
+      "loss": 0.2529,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 0.254,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.378600823045269e-06,
+      "loss": 0.2876,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.362139917695474e-06,
+      "loss": 0.3655,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.34567901234568e-06,
+      "loss": 0.3487,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.329218106995886e-06,
+      "loss": 0.2936,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.312757201646092e-06,
+      "loss": 0.2515,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.296296296296297e-06,
+      "loss": 0.2598,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.279835390946502e-06,
+      "loss": 0.2962,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.263374485596708e-06,
+      "loss": 0.2462,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.246913580246915e-06,
+      "loss": 0.278,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.23045267489712e-06,
+      "loss": 0.2959,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.213991769547325e-06,
+      "loss": 0.29,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.197530864197532e-06,
+      "loss": 0.228,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.181069958847738e-06,
+      "loss": 0.3202,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.164609053497943e-06,
+      "loss": 0.2715,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.3005,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.131687242798355e-06,
+      "loss": 0.1865,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.11522633744856e-06,
+      "loss": 0.278,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.098765432098766e-06,
+      "loss": 0.2898,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.082304526748971e-06,
+      "loss": 0.2866,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.065843621399178e-06,
+      "loss": 0.2683,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.049382716049384e-06,
+      "loss": 0.2315,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.032921810699589e-06,
+      "loss": 0.1776,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.016460905349794e-06,
+      "loss": 0.3306,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2477,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.983539094650207e-06,
+      "loss": 0.3478,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.967078189300412e-06,
+      "loss": 0.3863,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.950617283950617e-06,
+      "loss": 0.2145,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.934156378600824e-06,
+      "loss": 0.3197,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.91769547325103e-06,
+      "loss": 0.2809,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.901234567901235e-06,
+      "loss": 0.2612,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.884773662551442e-06,
+      "loss": 0.3075,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.868312757201647e-06,
+      "loss": 0.2036,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.851851851851853e-06,
+      "loss": 0.283,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.835390946502058e-06,
+      "loss": 0.2586,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.818930041152263e-06,
+      "loss": 0.2758,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.80246913580247e-06,
+      "loss": 0.3118,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.786008230452676e-06,
+      "loss": 0.3288,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.769547325102881e-06,
+      "loss": 0.2522,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.753086419753088e-06,
+      "loss": 0.3221,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.736625514403293e-06,
+      "loss": 0.2602,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.720164609053499e-06,
+      "loss": 0.2712,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 0.2974,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.68724279835391e-06,
+      "loss": 0.2452,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.670781893004116e-06,
+      "loss": 0.3302,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.654320987654322e-06,
+      "loss": 0.2983,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.637860082304529e-06,
+      "loss": 0.3151,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.621399176954733e-06,
+      "loss": 0.323,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.604938271604939e-06,
+      "loss": 0.3028,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.588477366255145e-06,
+      "loss": 0.3187,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.57201646090535e-06,
+      "loss": 0.2486,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 0.337,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.5390946502057615e-06,
+      "loss": 0.2513,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.522633744855967e-06,
+      "loss": 0.262,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.506172839506174e-06,
+      "loss": 0.2975,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.489711934156379e-06,
+      "loss": 0.2592,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.473251028806585e-06,
+      "loss": 0.3215,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.456790123456791e-06,
+      "loss": 0.3414,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.440329218106996e-06,
+      "loss": 0.2395,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.423868312757202e-06,
+      "loss": 0.2723,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.3141,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.3909465020576145e-06,
+      "loss": 0.2892,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.37448559670782e-06,
+      "loss": 0.2333,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.358024691358025e-06,
+      "loss": 0.2966,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.341563786008231e-06,
+      "loss": 0.3315,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.325102880658437e-06,
+      "loss": 0.2713,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.308641975308642e-06,
+      "loss": 0.2644,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.292181069958848e-06,
+      "loss": 0.2847,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.2757201646090535e-06,
+      "loss": 0.2851,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.2592592592592605e-06,
+      "loss": 0.3342,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.242798353909466e-06,
+      "loss": 0.2807,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.226337448559671e-06,
+      "loss": 0.3165,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.209876543209877e-06,
+      "loss": 0.3117,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.193415637860083e-06,
+      "loss": 0.2652,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.176954732510289e-06,
+      "loss": 0.3127,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.160493827160494e-06,
+      "loss": 0.2617,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.1440329218106995e-06,
+      "loss": 0.3543,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.1275720164609065e-06,
+      "loss": 0.2408,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 0.2642,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.094650205761318e-06,
+      "loss": 0.2714,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.078189300411523e-06,
+      "loss": 0.2758,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.061728395061729e-06,
+      "loss": 0.2915,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.045267489711935e-06,
+      "loss": 0.2442,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.02880658436214e-06,
+      "loss": 0.3132,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.012345679012347e-06,
+      "loss": 0.3328,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.9958847736625525e-06,
+      "loss": 0.3184,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.979423868312758e-06,
+      "loss": 0.3663,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.962962962962964e-06,
+      "loss": 0.2012,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.946502057613169e-06,
+      "loss": 0.3212,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.930041152263375e-06,
+      "loss": 0.2874,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.913580246913581e-06,
+      "loss": 0.234,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.897119341563786e-06,
+      "loss": 0.2659,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.880658436213993e-06,
+      "loss": 0.3152,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.8641975308641985e-06,
+      "loss": 0.2853,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.847736625514404e-06,
+      "loss": 0.2406,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.83127572016461e-06,
+      "loss": 0.2398,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 0.2246,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.798353909465021e-06,
+      "loss": 0.2758,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.781893004115227e-06,
+      "loss": 0.3276,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.765432098765433e-06,
+      "loss": 0.3139,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.748971193415639e-06,
+      "loss": 0.2619,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.7325102880658445e-06,
+      "loss": 0.2793,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.71604938271605e-06,
+      "loss": 0.2857,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.699588477366256e-06,
+      "loss": 0.2805,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.683127572016461e-06,
+      "loss": 0.2349,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3187,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.650205761316873e-06,
+      "loss": 0.2935,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.633744855967079e-06,
+      "loss": 0.2587,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.617283950617285e-06,
+      "loss": 0.2865,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.6008230452674905e-06,
+      "loss": 0.2914,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.584362139917696e-06,
+      "loss": 0.1614,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.567901234567902e-06,
+      "loss": 0.3026,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.551440329218107e-06,
+      "loss": 0.2654,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.534979423868313e-06,
+      "loss": 0.2702,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.287,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.502057613168725e-06,
+      "loss": 0.3049,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.485596707818931e-06,
+      "loss": 0.2941,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.4691358024691365e-06,
+      "loss": 0.2761,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.452674897119342e-06,
+      "loss": 0.2768,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.436213991769548e-06,
+      "loss": 0.3103,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.419753086419753e-06,
+      "loss": 0.2107,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.403292181069959e-06,
+      "loss": 0.2555,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.386831275720166e-06,
+      "loss": 0.2427,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.370370370370371e-06,
+      "loss": 0.3058,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.353909465020577e-06,
+      "loss": 0.2445,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.3374485596707825e-06,
+      "loss": 0.2414,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.320987654320988e-06,
+      "loss": 0.2323,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.304526748971194e-06,
+      "loss": 0.2689,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.288065843621399e-06,
+      "loss": 0.2233,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.271604938271606e-06,
+      "loss": 0.2722,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.255144032921812e-06,
+      "loss": 0.3739,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.238683127572017e-06,
+      "loss": 0.2424,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.2393,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.2057613168724285e-06,
+      "loss": 0.2319,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.189300411522634e-06,
+      "loss": 0.3131,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.17283950617284e-06,
+      "loss": 0.4243,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.156378600823045e-06,
+      "loss": 0.2794,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.139917695473252e-06,
+      "loss": 0.239,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.123456790123458e-06,
+      "loss": 0.3139,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.106995884773663e-06,
+      "loss": 0.2871,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.090534979423869e-06,
+      "loss": 0.3136,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.0740740740740745e-06,
+      "loss": 0.3055,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.05761316872428e-06,
+      "loss": 0.2432,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.041152263374486e-06,
+      "loss": 0.2589,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.024691358024692e-06,
+      "loss": 0.2632,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.008230452674898e-06,
+      "loss": 0.273,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.991769547325104e-06,
+      "loss": 0.3014,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.975308641975309e-06,
+      "loss": 0.2923,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.958847736625515e-06,
+      "loss": 0.2398,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9423868312757205e-06,
+      "loss": 0.2763,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.2845,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.909465020576132e-06,
+      "loss": 0.3448,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.893004115226338e-06,
+      "loss": 0.2793,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.876543209876544e-06,
+      "loss": 0.2154,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.86008230452675e-06,
+      "loss": 0.3244,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.843621399176955e-06,
+      "loss": 0.215,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.827160493827161e-06,
+      "loss": 0.2354,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.8106995884773665e-06,
+      "loss": 0.2485,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.794238683127572e-06,
+      "loss": 0.2589,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 0.3278,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.761316872427984e-06,
+      "loss": 0.2773,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.74485596707819e-06,
+      "loss": 0.2243,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.728395061728396e-06,
+      "loss": 0.3358,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.711934156378601e-06,
+      "loss": 0.2753,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.695473251028807e-06,
+      "loss": 0.293,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.6790123456790125e-06,
+      "loss": 0.2191,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.662551440329218e-06,
+      "loss": 0.2578,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.646090534979425e-06,
+      "loss": 0.2269,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 0.3249,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.613168724279836e-06,
+      "loss": 0.1927,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.596707818930042e-06,
+      "loss": 0.2249,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.580246913580247e-06,
+      "loss": 0.2791,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.563786008230453e-06,
+      "loss": 0.2616,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.5473251028806585e-06,
+      "loss": 0.2086,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.530864197530864e-06,
+      "loss": 0.2904,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.514403292181071e-06,
+      "loss": 0.2982,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.497942386831276e-06,
+      "loss": 0.3262,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.481481481481482e-06,
+      "loss": 0.2928,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.465020576131688e-06,
+      "loss": 0.2536,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.448559670781893e-06,
+      "loss": 0.2453,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.432098765432099e-06,
+      "loss": 0.3128,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.4156378600823045e-06,
+      "loss": 0.3244,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.3991769547325115e-06,
+      "loss": 0.201,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.382716049382717e-06,
+      "loss": 0.2821,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.366255144032922e-06,
+      "loss": 0.3215,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.349794238683128e-06,
+      "loss": 0.2601,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.2612,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.316872427983539e-06,
+      "loss": 0.3272,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.300411522633745e-06,
+      "loss": 0.2983,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.2839506172839505e-06,
+      "loss": 0.245,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.2674897119341575e-06,
+      "loss": 0.2423,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.251028806584363e-06,
+      "loss": 0.2654,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.234567901234568e-06,
+      "loss": 0.327,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.218106995884774e-06,
+      "loss": 0.4289,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.20164609053498e-06,
+      "loss": 0.2147,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.2734,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.168724279835391e-06,
+      "loss": 0.3855,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.152263374485597e-06,
+      "loss": 0.2498,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.1358024691358035e-06,
+      "loss": 0.2854,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.119341563786009e-06,
+      "loss": 0.2627,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.102880658436214e-06,
+      "loss": 0.3228,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.08641975308642e-06,
+      "loss": 0.2606,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.069958847736626e-06,
+      "loss": 0.2349,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.053497942386831e-06,
+      "loss": 0.2368,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.037037037037037e-06,
+      "loss": 0.1961,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.020576131687243e-06,
+      "loss": 0.3466,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.0041152263374495e-06,
+      "loss": 0.2575,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.987654320987655e-06,
+      "loss": 0.2525,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.97119341563786e-06,
+      "loss": 0.2662,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.954732510288066e-06,
+      "loss": 0.2363,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 0.3499,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.921810699588478e-06,
+      "loss": 0.3025,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905349794238683e-06,
+      "loss": 0.316,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 0.2435,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8724279835390955e-06,
+      "loss": 0.3171,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.855967078189301e-06,
+      "loss": 0.2926,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.839506172839506e-06,
+      "loss": 0.2855,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.823045267489712e-06,
+      "loss": 0.2092,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8065843621399185e-06,
+      "loss": 0.294,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.790123456790124e-06,
+      "loss": 0.2709,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.773662551440329e-06,
+      "loss": 0.2577,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.757201646090535e-06,
+      "loss": 0.2638,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 0.2688,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.724279835390947e-06,
+      "loss": 0.2346,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.707818930041152e-06,
+      "loss": 0.359,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 0.2997,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6748971193415645e-06,
+      "loss": 0.3003,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.65843621399177e-06,
+      "loss": 0.291,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.641975308641975e-06,
+      "loss": 0.2967,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.625514403292181e-06,
+      "loss": 0.2519,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6090534979423875e-06,
+      "loss": 0.3122,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.592592592592593e-06,
+      "loss": 0.2775,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.576131687242798e-06,
+      "loss": 0.2488,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.559670781893004e-06,
+      "loss": 0.2845,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5432098765432105e-06,
+      "loss": 0.3085,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526748971193416e-06,
+      "loss": 0.3742,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510288065843621e-06,
+      "loss": 0.2972,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.493827160493827e-06,
+      "loss": 0.2501,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4773662551440335e-06,
+      "loss": 0.2652,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.460905349794239e-06,
+      "loss": 0.2411,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.3045,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.42798353909465e-06,
+      "loss": 0.2764,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4115226337448565e-06,
+      "loss": 0.3705,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.395061728395062e-06,
+      "loss": 0.2299,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.378600823045268e-06,
+      "loss": 0.2138,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.362139917695473e-06,
+      "loss": 0.2885,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.3456790123456795e-06,
+      "loss": 0.2822,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.329218106995885e-06,
+      "loss": 0.3187,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.312757201646091e-06,
+      "loss": 0.2897,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.296296296296296e-06,
+      "loss": 0.3047,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2798353909465025e-06,
+      "loss": 0.2437,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.263374485596708e-06,
+      "loss": 0.318,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.246913580246914e-06,
+      "loss": 0.3601,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.23045267489712e-06,
+      "loss": 0.3227,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2139917695473255e-06,
+      "loss": 0.2253,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 0.2683,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.181069958847737e-06,
+      "loss": 0.2662,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.164609053497943e-06,
+      "loss": 0.1885,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1253,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}