diff --git "a/checkpoint-26190/trainer_state.json" "b/checkpoint-26190/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-26190/trainer_state.json"
@@ -0,0 +1,157204 @@
+{
+  "best_metric": 1.0227203369140625,
+  "best_model_checkpoint": "news-category-classifier-distilbert/checkpoint-26190",
+  "epoch": 5.0,
+  "eval_steps": 1,
+  "global_step": 26190,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.823,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.753,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.8174,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.7731,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.8113,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.741,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.7273,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.7975,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.8448,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.733,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1e-06,
+      "loss": 3.7654,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.7529,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3e-06,
+      "loss": 3.75,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.6525,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 3.7996,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.8306,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.7536,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.7642,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.7645,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.65,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.7366,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-06,
+      "loss": 3.7872,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.6725,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.7912,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 3.7664,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-06,
+      "loss": 3.7313,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.6718,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.745,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-06,
+      "loss": 3.7278,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 3.7683,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.6686,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.7453,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.7898,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.6603,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 3.6896,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.7122,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-06,
+      "loss": 3.7154,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.7559,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.7736,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.5454,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-06,
+      "loss": 3.6679,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.5774,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-06,
+      "loss": 3.6449,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 3.6707,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 3.721,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.7674,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-06,
+      "loss": 3.6828,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.7103,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.634,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 3.7189,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999044525128994e-06,
+      "loss": 3.631,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9998089050257985e-06,
+      "loss": 3.6721,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999713357538697e-06,
+      "loss": 3.6482,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996178100515966e-06,
+      "loss": 3.6807,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999522262564495e-06,
+      "loss": 3.5768,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999426715077394e-06,
+      "loss": 3.6546,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999331167590293e-06,
+      "loss": 3.6519,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999235620103192e-06,
+      "loss": 3.5507,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99914007261609e-06,
+      "loss": 3.6189,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999044525128989e-06,
+      "loss": 3.6134,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998948977641889e-06,
+      "loss": 3.5441,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998853430154787e-06,
+      "loss": 3.6359,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998757882667686e-06,
+      "loss": 3.5292,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998662335180585e-06,
+      "loss": 3.5595,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998566787693484e-06,
+      "loss": 3.6821,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9984712402063825e-06,
+      "loss": 3.432,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983756927192815e-06,
+      "loss": 3.4827,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9982801452321806e-06,
+      "loss": 3.6031,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99818459774508e-06,
+      "loss": 3.5891,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998089050257979e-06,
+      "loss": 3.5966,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997993502770878e-06,
+      "loss": 3.6144,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997897955283777e-06,
+      "loss": 3.6072,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997802407796675e-06,
+      "loss": 3.5461,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997706860309574e-06,
+      "loss": 3.5314,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997611312822473e-06,
+      "loss": 3.5826,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997515765335372e-06,
+      "loss": 3.491,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997420217848271e-06,
+      "loss": 3.4861,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99732467036117e-06,
+      "loss": 3.3482,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997229122874069e-06,
+      "loss": 3.5833,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997133575386967e-06,
+      "loss": 3.4859,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997038027899866e-06,
+      "loss": 3.5181,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996942480412765e-06,
+      "loss": 3.4885,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968469329256644e-06,
+      "loss": 3.4508,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9967513854385635e-06,
+      "loss": 3.3732,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9966558379514625e-06,
+      "loss": 3.4519,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965602904643616e-06,
+      "loss": 3.4123,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99646474297726e-06,
+      "loss": 3.5593,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996369195490159e-06,
+      "loss": 3.5539,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996273648003058e-06,
+      "loss": 3.2962,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996178100515957e-06,
+      "loss": 3.3013,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996082553028856e-06,
+      "loss": 3.6485,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995987005541755e-06,
+      "loss": 3.326,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995891458054654e-06,
+      "loss": 3.4103,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995795910567552e-06,
+      "loss": 3.3727,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995700363080451e-06,
+      "loss": 3.4636,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99560481559335e-06,
+      "loss": 3.4255,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995509268106249e-06,
+      "loss": 3.5618,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995413720619148e-06,
+      "loss": 3.5383,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995318173132047e-06,
+      "loss": 3.2373,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9952226256449456e-06,
+      "loss": 3.4197,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995127078157845e-06,
+      "loss": 3.5381,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995031530670744e-06,
+      "loss": 3.4718,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994935983183643e-06,
+      "loss": 3.5178,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994840435696542e-06,
+      "loss": 3.3171,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994744888209441e-06,
+      "loss": 3.4195,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99464934072234e-06,
+      "loss": 3.3075,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994553793235238e-06,
+      "loss": 3.4123,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994458245748137e-06,
+      "loss": 3.3492,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994362698261036e-06,
+      "loss": 3.4191,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994267150773935e-06,
+      "loss": 3.3115,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994171603286833e-06,
+      "loss": 3.28,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994076055799733e-06,
+      "loss": 3.3896,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993980508312632e-06,
+      "loss": 3.3385,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99388496082553e-06,
+      "loss": 3.3514,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9937894133384294e-06,
+      "loss": 3.2966,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9936938658513285e-06,
+      "loss": 3.2193,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9935983183642275e-06,
+      "loss": 3.4536,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993502770877126e-06,
+      "loss": 3.4856,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993407223390026e-06,
+      "loss": 3.2176,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993311675902925e-06,
+      "loss": 3.0366,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993216128415823e-06,
+      "loss": 3.5064,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993120580928722e-06,
+      "loss": 3.2942,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993025033441621e-06,
+      "loss": 3.2021,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99292948595452e-06,
+      "loss": 3.6751,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992833938467418e-06,
+      "loss": 3.3195,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992738390980318e-06,
+      "loss": 3.1699,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992642843493217e-06,
+      "loss": 3.3721,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992547296006115e-06,
+      "loss": 3.3786,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992451748519014e-06,
+      "loss": 3.2499,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992356201031913e-06,
+      "loss": 3.1661,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992260653544812e-06,
+      "loss": 3.3669,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9921651060577106e-06,
+      "loss": 3.2149,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9920695585706104e-06,
+      "loss": 3.4662,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9919740110835095e-06,
+      "loss": 3.2667,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991878463596408e-06,
+      "loss": 3.2932,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991782916109307e-06,
+      "loss": 3.1916,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991687368622206e-06,
+      "loss": 3.4094,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991591821135105e-06,
+      "loss": 3.3483,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991496273648003e-06,
+      "loss": 2.9835,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991400726160902e-06,
+      "loss": 3.178,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991305178673802e-06,
+      "loss": 3.2402,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9912096311867e-06,
+      "loss": 3.199,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991114083699599e-06,
+      "loss": 3.2952,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991018536212498e-06,
+      "loss": 3.2563,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990922988725397e-06,
+      "loss": 3.3144,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990827441238295e-06,
+      "loss": 3.0462,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9907318937511944e-06,
+      "loss": 3.4096,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906363462640935e-06,
+      "loss": 3.2858,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9905407987769925e-06,
+      "loss": 3.126,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9904452512898916e-06,
+      "loss": 3.3534,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990349703802791e-06,
+      "loss": 3.1871,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990254156315689e-06,
+      "loss": 3.202,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990158608828588e-06,
+      "loss": 3.0675,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990063061341487e-06,
+      "loss": 3.3509,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989967513854386e-06,
+      "loss": 3.3073,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989871966367285e-06,
+      "loss": 3.3301,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989776418880184e-06,
+      "loss": 3.2746,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989680871393083e-06,
+      "loss": 3.0299,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989585323905981e-06,
+      "loss": 3.3449,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98948977641888e-06,
+      "loss": 3.071,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989394228931779e-06,
+      "loss": 2.984,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989298681444678e-06,
+      "loss": 3.2445,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989203133957577e-06,
+      "loss": 3.1209,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989107586470476e-06,
+      "loss": 3.3189,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9890120389833754e-06,
+      "loss": 3.2758,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988916491496274e-06,
+      "loss": 3.2951,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988820944009173e-06,
+      "loss": 3.088,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988725396522072e-06,
+      "loss": 3.1747,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988629849034971e-06,
+      "loss": 3.193,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98853430154787e-06,
+      "loss": 3.2429,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988438754060769e-06,
+      "loss": 3.4449,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988343206573668e-06,
+      "loss": 3.2453,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988247659086566e-06,
+      "loss": 3.246,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988152111599465e-06,
+      "loss": 3.3818,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988056564112364e-06,
+      "loss": 3.2503,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987961016625263e-06,
+      "loss": 2.9093,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987865469138162e-06,
+      "loss": 3.2789,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987769921651061e-06,
+      "loss": 3.0265,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98767437416396e-06,
+      "loss": 2.9101,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9875788266768585e-06,
+      "loss": 2.9748,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9874832791897575e-06,
+      "loss": 3.3132,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9873877317026566e-06,
+      "loss": 3.138,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987292184215556e-06,
+      "loss": 3.2879,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987196636728455e-06,
+      "loss": 2.9482,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987101089241354e-06,
+      "loss": 3.1312,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987005541754253e-06,
+      "loss": 3.3396,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986909994267151e-06,
+      "loss": 2.7904,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98681444678005e-06,
+      "loss": 2.9283,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986718899292949e-06,
+      "loss": 3.1768,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986623351805848e-06,
+      "loss": 3.308,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986527804318747e-06,
+      "loss": 2.9387,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986432256831646e-06,
+      "loss": 3.4139,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986336709344545e-06,
+      "loss": 3.3419,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986241161857443e-06,
+      "loss": 2.9295,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986145614370342e-06,
+      "loss": 3.3774,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986050066883241e-06,
+      "loss": 3.0261,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9859545193961404e-06,
+      "loss": 2.9463,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9858589719090395e-06,
+      "loss": 2.9503,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9857634244219385e-06,
+      "loss": 3.0994,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985667876934837e-06,
+      "loss": 3.1089,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985572329447736e-06,
+      "loss": 3.1612,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985476781960635e-06,
+      "loss": 3.1919,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985381234473534e-06,
+      "loss": 3.2767,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985285686986433e-06,
+      "loss": 3.0394,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985190139499332e-06,
+      "loss": 2.984,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985094592012231e-06,
+      "loss": 2.9513,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984999044525129e-06,
+      "loss": 3.3238,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984903497038028e-06,
+      "loss": 3.2814,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984807949550927e-06,
+      "loss": 3.0514,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984712402063826e-06,
+      "loss": 3.048,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9846168545767244e-06,
+      "loss": 3.159,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9845213070896235e-06,
+      "loss": 3.2179,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984425759602523e-06,
+      "loss": 2.8968,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9843302121154216e-06,
+      "loss": 3.0299,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984234664628321e-06,
+      "loss": 3.1512,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98413911714122e-06,
+      "loss": 2.9927,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984043569654119e-06,
+      "loss": 2.9712,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983948022167017e-06,
+      "loss": 3.1299,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983852474679916e-06,
+      "loss": 3.0708,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983756927192816e-06,
+      "loss": 2.9308,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983661379705714e-06,
+      "loss": 2.9977,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983565832218613e-06,
+      "loss": 3.0129,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983470284731512e-06,
+      "loss": 2.8616,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983374737244411e-06,
+      "loss": 3.1054,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983279189757309e-06,
+      "loss": 2.7482,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983183642270208e-06,
+      "loss": 3.199,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983088094783108e-06,
+      "loss": 3.1299,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982992547296006e-06,
+      "loss": 3.0947,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9828969998089054e-06,
+      "loss": 2.7645,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9828014523218045e-06,
+      "loss": 3.1288,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9827059048347035e-06,
+      "loss": 3.2028,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982610357347602e-06,
+      "loss": 3.2465,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982514809860501e-06,
+      "loss": 2.9592,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982419262373401e-06,
+      "loss": 3.2002,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982323714886299e-06,
+      "loss": 2.4657,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982228167399198e-06,
+      "loss": 2.9636,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982132619912097e-06,
+      "loss": 3.3094,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982037072424996e-06,
+      "loss": 3.221,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981941524937894e-06,
+      "loss": 3.0155,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981845977450793e-06,
+      "loss": 3.0232,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981750429963692e-06,
+      "loss": 2.9887,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981654882476591e-06,
+      "loss": 3.197,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98155933498949e-06,
+      "loss": 2.6409,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981463787502389e-06,
+      "loss": 3.2644,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981368240015288e-06,
+      "loss": 2.5508,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9812726925281866e-06,
+      "loss": 3.0668,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981177145041086e-06,
+      "loss": 3.2028,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981081597553985e-06,
+      "loss": 2.9136,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980986050066884e-06,
+      "loss": 2.9139,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980890502579783e-06,
+      "loss": 3.1184,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980794955092682e-06,
+      "loss": 2.9618,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.98069940760558e-06,
+      "loss": 2.7051,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980603860118479e-06,
+      "loss": 2.9968,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980508312631378e-06,
+      "loss": 2.9412,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980412765144277e-06,
+      "loss": 2.9646,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980317217657176e-06,
+      "loss": 3.148,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980221670170075e-06,
+      "loss": 2.9666,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980126122682974e-06,
+      "loss": 3.257,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980030575195872e-06,
+      "loss": 3.1641,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979935027708771e-06,
+      "loss": 2.4049,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9798394802216704e-06,
+      "loss": 3.2577,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9797439327345695e-06,
+      "loss": 2.6859,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9796483852474685e-06,
+      "loss": 2.9539,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9795528377603676e-06,
+      "loss": 2.8831,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979457290273267e-06,
+      "loss": 2.7155,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979361742786165e-06,
+      "loss": 3.1108,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979266195299064e-06,
+      "loss": 3.064,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979170647811963e-06,
+      "loss": 2.8307,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.979075100324862e-06,
+      "loss": 2.9395,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978979552837761e-06,
+      "loss": 2.8165,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97888400535066e-06,
+      "loss": 2.9628,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978788457863559e-06,
+      "loss": 2.7856,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978692910376457e-06,
+      "loss": 2.8218,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978597362889356e-06,
+      "loss": 2.7327,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978501815402255e-06,
+      "loss": 2.8508,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978406267915154e-06,
+      "loss": 3.0315,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978310720428053e-06,
+      "loss": 2.9114,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.978215172940952e-06,
+      "loss": 2.8769,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9781196254538514e-06,
+      "loss": 3.0463,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97802407796675e-06,
+      "loss": 2.7959,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977928530479649e-06,
+      "loss": 2.8887,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977832982992548e-06,
+      "loss": 2.835,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977737435505447e-06,
+      "loss": 2.8037,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977641888018345e-06,
+      "loss": 2.8335,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977546340531245e-06,
+      "loss": 3.1785,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977450793044144e-06,
+      "loss": 2.5667,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977355245557042e-06,
+      "loss": 2.544,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977259698069941e-06,
+      "loss": 2.8199,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97716415058284e-06,
+      "loss": 2.8967,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.977068603095739e-06,
+      "loss": 2.8983,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976973055608637e-06,
+      "loss": 2.7508,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976877508121537e-06,
+      "loss": 2.5298,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9767819606344354e-06,
+      "loss": 2.9811,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9766864131473345e-06,
+      "loss": 2.8177,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9765908656602335e-06,
+      "loss": 2.4499,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9764953181731326e-06,
+      "loss": 2.5846,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976399770686032e-06,
+      "loss": 2.9581,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97630422319893e-06,
+      "loss": 3.144,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97620867571183e-06,
+      "loss": 3.0438,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976113128224728e-06,
+      "loss": 2.8643,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.976017580737627e-06,
+      "loss": 2.7185,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975922033250526e-06,
+      "loss": 2.6449,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975826485763425e-06,
+      "loss": 2.8537,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975730938276323e-06,
+      "loss": 2.5933,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975635390789222e-06,
+      "loss": 2.9121,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975539843302122e-06,
+      "loss": 2.6188,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97544429581502e-06,
+      "loss": 2.7504,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975348748327919e-06,
+      "loss": 2.8851,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975253200840818e-06,
+      "loss": 2.5322,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975157653353717e-06,
+      "loss": 2.6466,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.975062105866616e-06,
+      "loss": 2.6217,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974966558379515e-06,
+      "loss": 2.9667,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9748710108924145e-06,
+      "loss": 2.5703,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974775463405313e-06,
+      "loss": 2.6111,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974679915918212e-06,
+      "loss": 2.9395,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974584368431111e-06,
+      "loss": 2.2974,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97448882094401e-06,
+      "loss": 2.8733,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974393273456908e-06,
+      "loss": 2.4615,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974297725969807e-06,
+      "loss": 2.7818,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974202178482706e-06,
+      "loss": 2.6705,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974106630995605e-06,
+      "loss": 2.8077,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.974011083508504e-06,
+      "loss": 2.8113,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973915536021403e-06,
+      "loss": 3.1466,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973819988534302e-06,
+      "loss": 2.5994,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9737244410472004e-06,
+      "loss": 3.2421,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9736288935600995e-06,
+      "loss": 2.6774,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9735333460729985e-06,
+      "loss": 2.5481,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9734377985858976e-06,
+      "loss": 2.8017,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973342251098797e-06,
+      "loss": 2.757,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973246703611696e-06,
+      "loss": 2.4866,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973151156124595e-06,
+      "loss": 2.9139,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.973055608637493e-06,
+      "loss": 2.4555,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972960061150392e-06,
+      "loss": 2.9109,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972864513663291e-06,
+      "loss": 2.9388,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97276896617619e-06,
+      "loss": 2.4048,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972673418689089e-06,
+      "loss": 2.4312,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972577871201988e-06,
+      "loss": 2.5545,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972482323714887e-06,
+      "loss": 3.1067,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972386776227785e-06,
+      "loss": 2.8093,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.972291228740684e-06,
+      "loss": 2.5731,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.972195681253583e-06,
+      "loss": 2.6359,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.972100133766482e-06,
+      "loss": 2.8517,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9720045862793814e-06,
+      "loss": 2.5779,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9719090387922805e-06,
+      "loss": 2.5159,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9718134913051795e-06,
+      "loss": 3.3929,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971717943818078e-06,
+      "loss": 2.7878,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971622396330977e-06,
+      "loss": 2.8362,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971526848843876e-06,
+      "loss": 2.91,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971431301356775e-06,
+      "loss": 2.901,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971335753869674e-06,
+      "loss": 2.7279,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971240206382573e-06,
+      "loss": 2.1007,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.971144658895471e-06,
+      "loss": 2.919,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97104911140837e-06,
+      "loss": 2.7293,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970953563921269e-06,
+      "loss": 2.7158,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970858016434168e-06,
+      "loss": 2.7236,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970762468947066e-06,
+      "loss": 2.9423,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970666921459966e-06,
+      "loss": 2.5377,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970571373972865e-06,
+      "loss": 2.4763,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9704758264857635e-06,
+      "loss": 2.451,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9703802789986626e-06,
+      "loss": 2.332,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970284731511562e-06,
+      "loss": 2.8933,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970189184024461e-06,
+      "loss": 2.2132,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.970093636537359e-06,
+      "loss": 2.8664,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969998089050259e-06,
+      "loss": 2.5394,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969902541563158e-06,
+      "loss": 2.6666,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969806994076056e-06,
+      "loss": 2.4604,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969711446588955e-06,
+      "loss": 2.8417,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969615899101854e-06,
+      "loss": 2.9983,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969520351614753e-06,
+      "loss": 2.6717,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969424804127651e-06,
+      "loss": 2.6432,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969329256640551e-06,
+      "loss": 2.5861,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96923370915345e-06,
+      "loss": 2.9026,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969138161666348e-06,
+      "loss": 2.4419,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.969042614179247e-06,
+      "loss": 2.3774,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9689470666921464e-06,
+      "loss": 3.0139,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9688515192050455e-06,
+      "loss": 2.6877,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968755971717944e-06,
+      "loss": 3.126,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9686604242308436e-06,
+      "loss": 2.9502,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968564876743743e-06,
+      "loss": 2.4907,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968469329256641e-06,
+      "loss": 2.5898,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96837378176954e-06,
+      "loss": 2.6573,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968278234282439e-06,
+      "loss": 2.8638,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968182686795338e-06,
+      "loss": 2.9168,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.968087139308236e-06,
+      "loss": 2.5718,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967991591821136e-06,
+      "loss": 2.4016,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967896044334035e-06,
+      "loss": 2.6294,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967800496846933e-06,
+      "loss": 3.11,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967704949359832e-06,
+      "loss": 2.6956,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967609401872731e-06,
+      "loss": 2.9096,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.96751385438563e-06,
+      "loss": 3.0158,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9674183068985285e-06,
+      "loss": 2.6566,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9673227594114276e-06,
+      "loss": 2.4877,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967227211924327e-06,
+      "loss": 3.0353,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967131664437226e-06,
+      "loss": 2.3704,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967036116950125e-06,
+      "loss": 2.6801,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966940569463024e-06,
+      "loss": 2.4923,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966845021975923e-06,
+      "loss": 2.683,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966749474488821e-06,
+      "loss": 3.0196,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96665392700172e-06,
+      "loss": 2.7972,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966558379514619e-06,
+      "loss": 2.4284,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966462832027518e-06,
+      "loss": 2.5662,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966367284540417e-06,
+      "loss": 2.6584,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966271737053316e-06,
+      "loss": 3.077,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966176189566214e-06,
+      "loss": 2.3575,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966080642079113e-06,
+      "loss": 2.8027,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965985094592012e-06,
+      "loss": 2.665,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9658895471049114e-06,
+      "loss": 2.546,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9657939996178105e-06,
+      "loss": 2.8082,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9656984521307095e-06,
+      "loss": 2.9436,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9656029046436086e-06,
+      "loss": 2.627,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965507357156507e-06,
+      "loss": 2.4391,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965411809669406e-06,
+      "loss": 2.5671,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965316262182305e-06,
+      "loss": 2.5946,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965220714695204e-06,
+      "loss": 2.3694,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965125167208103e-06,
+      "loss": 2.8994,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965029619721002e-06,
+      "loss": 2.5082,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964934072233901e-06,
+      "loss": 3.0187,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964838524746799e-06,
+      "loss": 2.736,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964742977259698e-06,
+      "loss": 2.5021,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964647429772597e-06,
+      "loss": 2.2551,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964551882285496e-06,
+      "loss": 2.5971,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964456334798395e-06,
+      "loss": 3.0948,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964360787311294e-06,
+      "loss": 2.7736,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964265239824193e-06,
+      "loss": 2.6791,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964169692337092e-06,
+      "loss": 2.6263,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.964074144849991e-06,
+      "loss": 2.2192,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96397859736289e-06,
+      "loss": 2.9984,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963883049875789e-06,
+      "loss": 2.7421,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963787502388688e-06,
+      "loss": 2.7094,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963691954901587e-06,
+      "loss": 2.691,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963596407414486e-06,
+      "loss": 2.4698,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963500859927384e-06,
+      "loss": 2.8185,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963405312440283e-06,
+      "loss": 2.548,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963309764953182e-06,
+      "loss": 2.685,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963214217466081e-06,
+      "loss": 2.3123,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96311866997898e-06,
+      "loss": 2.5353,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963023122491879e-06,
+      "loss": 2.5139,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962927575004778e-06,
+      "loss": 2.7036,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9628320275176764e-06,
+      "loss": 2.7786,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9627364800305755e-06,
+      "loss": 2.4674,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9626409325434745e-06,
+      "loss": 2.3752,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9625453850563736e-06,
+      "loss": 2.7742,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962449837569273e-06,
+      "loss": 2.7574,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962354290082172e-06,
+      "loss": 2.8649,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96225874259507e-06,
+      "loss": 2.7329,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.962163195107969e-06,
+      "loss": 2.6748,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.962067647620868e-06,
+      "loss": 2.5076,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961972100133767e-06,
+      "loss": 3.0249,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961876552646666e-06,
+      "loss": 2.5932,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961781005159565e-06,
+      "loss": 2.4512,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961685457672464e-06,
+      "loss": 2.4215,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961589910185362e-06,
+      "loss": 2.7762,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961494362698261e-06,
+      "loss": 2.2112,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96139881521116e-06,
+      "loss": 2.5178,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.961303267724059e-06,
+      "loss": 2.8783,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9612077202369576e-06,
+      "loss": 2.7854,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9611121727498574e-06,
+      "loss": 2.524,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9610166252627565e-06,
+      "loss": 2.3494,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960921077775655e-06,
+      "loss": 2.5517,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960825530288554e-06,
+      "loss": 2.7995,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960729982801453e-06,
+      "loss": 2.3264,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960634435314352e-06,
+      "loss": 2.6315,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.96053888782725e-06,
+      "loss": 2.4635,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960443340340149e-06,
+      "loss": 2.5757,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960347792853049e-06,
+      "loss": 2.6874,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960252245365947e-06,
+      "loss": 2.8959,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960156697878846e-06,
+      "loss": 2.7733,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960061150391745e-06,
+      "loss": 2.5978,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959965602904644e-06,
+      "loss": 2.5016,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959870055417542e-06,
+      "loss": 2.6343,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9597745079304414e-06,
+      "loss": 2.5672,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959678960443341e-06,
+      "loss": 2.0833,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9595834129562395e-06,
+      "loss": 2.6771,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9594878654691386e-06,
+      "loss": 2.6521,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959392317982038e-06,
+      "loss": 2.4557,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959296770494937e-06,
+      "loss": 2.3887,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959201223007835e-06,
+      "loss": 2.7511,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959105675520734e-06,
+      "loss": 2.5071,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959010128033634e-06,
+      "loss": 2.8615,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958914580546532e-06,
+      "loss": 2.8467,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958819033059431e-06,
+      "loss": 3.1974,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95872348557233e-06,
+      "loss": 2.4031,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958627938085229e-06,
+      "loss": 2.4945,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958532390598127e-06,
+      "loss": 2.2722,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958436843111026e-06,
+      "loss": 2.3436,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958341295623925e-06,
+      "loss": 2.6383,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958245748136824e-06,
+      "loss": 2.2793,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958150200649723e-06,
+      "loss": 2.6345,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9580546531626224e-06,
+      "loss": 2.3145,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9579591056755215e-06,
+      "loss": 2.4488,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95786355818842e-06,
+      "loss": 2.9671,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957768010701319e-06,
+      "loss": 2.558,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957672463214218e-06,
+      "loss": 2.5404,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957576915727117e-06,
+      "loss": 2.5323,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957481368240016e-06,
+      "loss": 2.26,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957385820752915e-06,
+      "loss": 2.8045,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957290273265813e-06,
+      "loss": 2.2363,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.957194725778712e-06,
+      "loss": 2.6144,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.957099178291611e-06,
+      "loss": 2.6909,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95700363080451e-06,
+      "loss": 2.3201,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956908083317409e-06,
+      "loss": 2.166,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956812535830308e-06,
+      "loss": 2.6404,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956716988343207e-06,
+      "loss": 2.5312,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9566214408561055e-06,
+      "loss": 2.6058,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9565258933690045e-06,
+      "loss": 2.4209,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9564303458819036e-06,
+      "loss": 2.2667,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956334798394803e-06,
+      "loss": 2.5762,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956239250907702e-06,
+      "loss": 2.1086,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.956143703420601e-06,
+      "loss": 2.2779,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9560481559335e-06,
+      "loss": 2.2871,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955952608446398e-06,
+      "loss": 2.4999,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955857060959297e-06,
+      "loss": 2.7144,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955761513472196e-06,
+      "loss": 2.7294,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955665965985095e-06,
+      "loss": 2.513,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955570418497994e-06,
+      "loss": 2.5427,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955474871010893e-06,
+      "loss": 2.3232,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955379323523792e-06,
+      "loss": 2.528,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95528377603669e-06,
+      "loss": 2.4422,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955188228549589e-06,
+      "loss": 2.8364,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.955092681062488e-06,
+      "loss": 2.7846,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9549971335753874e-06,
+      "loss": 2.3886,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9549015860882865e-06,
+      "loss": 2.6021,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9548060386011855e-06,
+      "loss": 2.4468,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9547104911140846e-06,
+      "loss": 2.3045,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954614943626983e-06,
+      "loss": 2.4764,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954519396139882e-06,
+      "loss": 2.1028,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954423848652781e-06,
+      "loss": 2.4423,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95432830116568e-06,
+      "loss": 2.5443,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954232753678579e-06,
+      "loss": 2.5154,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954137206191478e-06,
+      "loss": 2.4133,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.954041658704377e-06,
+      "loss": 2.7606,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953946111217275e-06,
+      "loss": 2.6885,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953850563730174e-06,
+      "loss": 2.4669,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953755016243073e-06,
+      "loss": 2.84,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953659468755972e-06,
+      "loss": 2.2959,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9535639212688705e-06,
+      "loss": 2.1971,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95346837378177e-06,
+      "loss": 2.0107,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953372826294669e-06,
+      "loss": 2.2055,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953277278807568e-06,
+      "loss": 2.8706,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953181731320467e-06,
+      "loss": 2.1254,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.953086183833366e-06,
+      "loss": 2.3707,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952990636346265e-06,
+      "loss": 2.4366,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952895088859163e-06,
+      "loss": 2.67,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952799541372063e-06,
+      "loss": 2.2394,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952703993884961e-06,
+      "loss": 2.4049,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.95260844639786e-06,
+      "loss": 2.6002,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952512898910759e-06,
+      "loss": 2.7326,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952417351423658e-06,
+      "loss": 2.5397,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.952321803936556e-06,
+      "loss": 2.199,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.952226256449455e-06,
+      "loss": 2.3859,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.952130708962355e-06,
+      "loss": 2.7793,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.952035161475253e-06,
+      "loss": 2.224,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9519396139881524e-06,
+      "loss": 1.9003,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9518440665010515e-06,
+      "loss": 2.6741,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9517485190139505e-06,
+      "loss": 2.6443,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951652971526849e-06,
+      "loss": 2.5078,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951557424039748e-06,
+      "loss": 2.6299,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951461876552648e-06,
+      "loss": 3.1298,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951366329065546e-06,
+      "loss": 2.9926,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951270781578445e-06,
+      "loss": 2.6457,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951175234091344e-06,
+      "loss": 2.3856,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.951079686604243e-06,
+      "loss": 2.4708,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950984139117141e-06,
+      "loss": 2.3439,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.95088859163004e-06,
+      "loss": 2.6788,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.95079304414294e-06,
+      "loss": 2.2127,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950697496655838e-06,
+      "loss": 2.3698,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950601949168737e-06,
+      "loss": 2.3578,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950506401681636e-06,
+      "loss": 2.3316,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950410854194535e-06,
+      "loss": 2.8014,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9503153067074336e-06,
+      "loss": 2.2576,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950219759220333e-06,
+      "loss": 1.9012,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950124211733232e-06,
+      "loss": 2.4684,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.950028664246131e-06,
+      "loss": 2.4678,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.94993311675903e-06,
+      "loss": 2.4018,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949837569271929e-06,
+      "loss": 2.255,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949742021784828e-06,
+      "loss": 1.9568,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949646474297726e-06,
+      "loss": 2.6631,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949550926810625e-06,
+      "loss": 2.4893,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949455379323524e-06,
+      "loss": 2.4028,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949359831836423e-06,
+      "loss": 2.1785,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949264284349322e-06,
+      "loss": 2.3297,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.949168736862221e-06,
+      "loss": 2.9249,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.94907318937512e-06,
+      "loss": 2.5159,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948977641888018e-06,
+      "loss": 2.3543,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9488820944009174e-06,
+      "loss": 2.1195,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9487865469138165e-06,
+      "loss": 2.3728,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9486909994267155e-06,
+      "loss": 1.9255,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9485954519396146e-06,
+      "loss": 2.4634,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948499904452514e-06,
+      "loss": 2.4303,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948404356965413e-06,
+      "loss": 2.4157,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948308809478311e-06,
+      "loss": 2.4621,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.94821326199121e-06,
+      "loss": 2.0726,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948117714504109e-06,
+      "loss": 2.2599,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.948022167017008e-06,
+      "loss": 2.7717,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947926619529907e-06,
+      "loss": 2.6099,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947831072042806e-06,
+      "loss": 2.316,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947735524555704e-06,
+      "loss": 2.3763,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947639977068603e-06,
+      "loss": 2.0537,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947544429581502e-06,
+      "loss": 2.5753,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947448882094401e-06,
+      "loss": 2.2591,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9473533346073e-06,
+      "loss": 2.2637,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947257787120199e-06,
+      "loss": 1.9834,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9471622396330984e-06,
+      "loss": 2.5291,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.947066692145997e-06,
+      "loss": 2.245,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946971144658896e-06,
+      "loss": 2.6131,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946875597171795e-06,
+      "loss": 2.3686,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946780049684694e-06,
+      "loss": 2.6174,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946684502197592e-06,
+      "loss": 2.4324,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946588954710492e-06,
+      "loss": 2.2341,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946493407223391e-06,
+      "loss": 2.2882,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946397859736289e-06,
+      "loss": 2.2523,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946302312249188e-06,
+      "loss": 2.5834,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946206764762087e-06,
+      "loss": 2.4192,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946111217274986e-06,
+      "loss": 2.119,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.946015669787884e-06,
+      "loss": 2.6326,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945920122300784e-06,
+      "loss": 2.3737,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945824574813683e-06,
+      "loss": 2.4301,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9457290273265815e-06,
+      "loss": 2.2872,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9456334798394805e-06,
+      "loss": 2.1661,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9455379323523796e-06,
+      "loss": 2.4534,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945442384865279e-06,
+      "loss": 2.2346,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945346837378177e-06,
+      "loss": 2.4299,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945251289891077e-06,
+      "loss": 2.085,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945155742403976e-06,
+      "loss": 2.5297,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.945060194916874e-06,
+      "loss": 2.3321,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944964647429773e-06,
+      "loss": 1.8928,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944869099942672e-06,
+      "loss": 2.3943,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944773552455571e-06,
+      "loss": 2.1673,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944678004968469e-06,
+      "loss": 2.647,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944582457481369e-06,
+      "loss": 2.4178,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944486909994268e-06,
+      "loss": 2.1689,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944391362507166e-06,
+      "loss": 2.2363,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944295815020065e-06,
+      "loss": 2.4538,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944200267532964e-06,
+      "loss": 2.3706,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9441047200458634e-06,
+      "loss": 2.0035,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.944009172558762e-06,
+      "loss": 2.1102,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9439136250716615e-06,
+      "loss": 2.3921,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.94381807758456e-06,
+      "loss": 2.1114,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943722530097459e-06,
+      "loss": 2.1704,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943626982610358e-06,
+      "loss": 2.2333,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943531435123257e-06,
+      "loss": 2.146,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943435887636156e-06,
+      "loss": 2.468,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943340340149054e-06,
+      "loss": 2.4766,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943244792661953e-06,
+      "loss": 2.434,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943149245174852e-06,
+      "loss": 2.785,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.943053697687751e-06,
+      "loss": 2.6478,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.94295815020065e-06,
+      "loss": 2.4233,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942862602713549e-06,
+      "loss": 2.4536,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9427670552264474e-06,
+      "loss": 2.0271,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9426715077393465e-06,
+      "loss": 1.7754,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9425759602522455e-06,
+      "loss": 2.4503,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9424804127651446e-06,
+      "loss": 2.421,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942384865278044e-06,
+      "loss": 1.8888,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942289317790943e-06,
+      "loss": 2.4258,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.942193770303842e-06,
+      "loss": 2.2302,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94209822281674e-06,
+      "loss": 1.8291,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.942002675329639e-06,
+      "loss": 2.2731,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941907127842538e-06,
+      "loss": 2.3329,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941811580355437e-06,
+      "loss": 2.4414,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941716032868336e-06,
+      "loss": 2.3304,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941620485381235e-06,
+      "loss": 2.1785,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941524937894134e-06,
+      "loss": 2.7795,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941429390407032e-06,
+      "loss": 2.5391,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941333842919931e-06,
+      "loss": 2.4555,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94123829543283e-06,
+      "loss": 2.2966,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.941142747945729e-06,
+      "loss": 2.2483,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9410472004586284e-06,
+      "loss": 2.1548,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9409516529715275e-06,
+      "loss": 2.2289,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9408561054844265e-06,
+      "loss": 2.2848,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940760557997325e-06,
+      "loss": 2.5497,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940665010510224e-06,
+      "loss": 2.6223,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940569463023123e-06,
+      "loss": 2.216,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940473915536022e-06,
+      "loss": 2.4194,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940378368048921e-06,
+      "loss": 2.5147,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94028282056182e-06,
+      "loss": 2.2092,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940187273074719e-06,
+      "loss": 2.5212,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.940091725587617e-06,
+      "loss": 2.3188,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939996178100516e-06,
+      "loss": 2.5983,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939900630613415e-06,
+      "loss": 2.3487,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939805083126314e-06,
+      "loss": 2.4278,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939709535639213e-06,
+      "loss": 2.9012,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939613988152112e-06,
+      "loss": 2.2557,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939518440665011e-06,
+      "loss": 2.2946,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9394228931779096e-06,
+      "loss": 2.3173,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939327345690809e-06,
+      "loss": 2.2136,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939231798203708e-06,
+      "loss": 2.3867,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939136250716607e-06,
+      "loss": 2.6545,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939040703229506e-06,
+      "loss": 2.2522,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938945155742405e-06,
+      "loss": 2.2361,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938849608255303e-06,
+      "loss": 2.2073,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938754060768202e-06,
+      "loss": 2.3228,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938658513281101e-06,
+      "loss": 2.1671,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938562965794e-06,
+      "loss": 2.1109,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938467418306899e-06,
+      "loss": 2.6795,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938371870819798e-06,
+      "loss": 2.6108,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938276323332697e-06,
+      "loss": 2.2991,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938180775845595e-06,
+      "loss": 2.5136,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.938085228358494e-06,
+      "loss": 2.1027,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9379896808713934e-06,
+      "loss": 1.9974,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9378941333842925e-06,
+      "loss": 2.3302,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937798585897191e-06,
+      "loss": 2.6102,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9377030384100906e-06,
+      "loss": 2.5145,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.93760749092299e-06,
+      "loss": 2.2183,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937511943435888e-06,
+      "loss": 2.4318,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937416395948787e-06,
+      "loss": 2.442,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937320848461686e-06,
+      "loss": 2.3658,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937225300974585e-06,
+      "loss": 2.3337,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.937129753487483e-06,
+      "loss": 2.2602,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.937034206000383e-06,
+      "loss": 2.2478,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936938658513282e-06,
+      "loss": 2.2339,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93684311102618e-06,
+      "loss": 1.8902,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936747563539079e-06,
+      "loss": 2.2557,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936652016051978e-06,
+      "loss": 2.4826,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936556468564877e-06,
+      "loss": 2.2467,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9364609210777755e-06,
+      "loss": 2.5915,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9363653735906745e-06,
+      "loss": 2.3597,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9362698261035744e-06,
+      "loss": 2.4299,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936174278616473e-06,
+      "loss": 1.9392,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936078731129372e-06,
+      "loss": 2.0764,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935983183642271e-06,
+      "loss": 2.1228,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93588763615517e-06,
+      "loss": 2.4671,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935792088668068e-06,
+      "loss": 2.5266,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935696541180967e-06,
+      "loss": 2.0904,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935600993693867e-06,
+      "loss": 2.2228,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935505446206765e-06,
+      "loss": 2.2731,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935409898719664e-06,
+      "loss": 1.9986,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935314351232563e-06,
+      "loss": 1.9961,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935218803745462e-06,
+      "loss": 2.3225,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93512325625836e-06,
+      "loss": 2.3499,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.935027708771259e-06,
+      "loss": 2.0775,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934932161284159e-06,
+      "loss": 2.3114,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9348366137970575e-06,
+      "loss": 2.177,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9347410663099565e-06,
+      "loss": 2.3767,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9346455188228555e-06,
+      "loss": 1.9681,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934549971335755e-06,
+      "loss": 2.5531,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934454423848653e-06,
+      "loss": 2.299,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934358876361552e-06,
+      "loss": 1.9132,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934263328874451e-06,
+      "loss": 2.4211,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93416778138735e-06,
+      "loss": 2.2428,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.934072233900249e-06,
+      "loss": 1.7429,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933976686413148e-06,
+      "loss": 2.1964,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933881138926046e-06,
+      "loss": 2.0109,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933785591438945e-06,
+      "loss": 2.1676,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933690043951844e-06,
+      "loss": 2.4191,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933594496464743e-06,
+      "loss": 1.7803,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933498948977642e-06,
+      "loss": 2.0707,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933403401490541e-06,
+      "loss": 2.1751,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93330785400344e-06,
+      "loss": 1.9783,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933212306516339e-06,
+      "loss": 2.0706,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933116759029238e-06,
+      "loss": 2.1627,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.933021211542137e-06,
+      "loss": 2.2493,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932925664055036e-06,
+      "loss": 2.2716,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932830116567935e-06,
+      "loss": 2.1091,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932734569080834e-06,
+      "loss": 2.3177,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932639021593733e-06,
+      "loss": 2.4433,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932543474106631e-06,
+      "loss": 1.6911,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.93244792661953e-06,
+      "loss": 2.0357,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932352379132429e-06,
+      "loss": 2.1959,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.932256831645328e-06,
+      "loss": 2.4767,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.932161284158227e-06,
+      "loss": 2.0074,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.932065736671126e-06,
+      "loss": 2.2884,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931970189184025e-06,
+      "loss": 2.1149,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9318746416969234e-06,
+      "loss": 2.1472,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9317790942098225e-06,
+      "loss": 2.0862,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9316835467227215e-06,
+      "loss": 2.0476,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9315879992356205e-06,
+      "loss": 2.5671,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.93149245174852e-06,
+      "loss": 2.2434,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931396904261419e-06,
+      "loss": 2.2402,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931301356774318e-06,
+      "loss": 1.6853,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931205809287216e-06,
+      "loss": 1.9729,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931110261800115e-06,
+      "loss": 2.1362,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931014714313014e-06,
+      "loss": 2.5836,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930919166825913e-06,
+      "loss": 2.2723,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930823619338812e-06,
+      "loss": 2.0481,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930728071851711e-06,
+      "loss": 2.5973,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.93063252436461e-06,
+      "loss": 2.1644,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930536976877508e-06,
+      "loss": 2.2618,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930441429390407e-06,
+      "loss": 2.0924,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930345881903306e-06,
+      "loss": 2.1623,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930250334416205e-06,
+      "loss": 2.2476,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9301547869291044e-06,
+      "loss": 2.2044,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9300592394420035e-06,
+      "loss": 1.9718,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9299636919549025e-06,
+      "loss": 2.3463,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929868144467801e-06,
+      "loss": 2.2542,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9297725969807e-06,
+      "loss": 2.4133,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929677049493599e-06,
+      "loss": 2.0758,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929581502006498e-06,
+      "loss": 2.0059,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929485954519396e-06,
+      "loss": 2.2149,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929390407032296e-06,
+      "loss": 2.027,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929294859545194e-06,
+      "loss": 2.3462,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929199312058093e-06,
+      "loss": 2.3543,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929103764570992e-06,
+      "loss": 2.1166,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929008217083891e-06,
+      "loss": 2.3131,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.92891266959679e-06,
+      "loss": 2.088,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9288171221096884e-06,
+      "loss": 2.308,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928721574622588e-06,
+      "loss": 2.3022,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9286260271354865e-06,
+      "loss": 2.4973,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9285304796483855e-06,
+      "loss": 2.3499,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928434932161285e-06,
+      "loss": 2.0834,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928339384674184e-06,
+      "loss": 1.9986,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928243837187082e-06,
+      "loss": 1.9384,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928148289699981e-06,
+      "loss": 2.2661,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928052742212881e-06,
+      "loss": 2.3604,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927957194725779e-06,
+      "loss": 2.001,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927861647238678e-06,
+      "loss": 2.0161,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927766099751577e-06,
+      "loss": 2.1697,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927670552264476e-06,
+      "loss": 2.3317,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927575004777374e-06,
+      "loss": 1.6521,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927479457290273e-06,
+      "loss": 2.3148,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927383909803173e-06,
+      "loss": 2.0725,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.927288362316071e-06,
+      "loss": 2.4909,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92719281482897e-06,
+      "loss": 2.4612,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9270972673418694e-06,
+      "loss": 2.3311,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9270017198547685e-06,
+      "loss": 1.8679,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926906172367667e-06,
+      "loss": 2.3045,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926810624880566e-06,
+      "loss": 2.0717,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926715077393466e-06,
+      "loss": 2.3152,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926619529906364e-06,
+      "loss": 1.935,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926523982419263e-06,
+      "loss": 2.6549,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926428434932162e-06,
+      "loss": 2.2563,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926332887445061e-06,
+      "loss": 2.4041,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926237339957959e-06,
+      "loss": 2.358,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926141792470858e-06,
+      "loss": 1.8682,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926046244983757e-06,
+      "loss": 2.0606,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925950697496656e-06,
+      "loss": 1.8505,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925855150009555e-06,
+      "loss": 1.9034,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925759602522454e-06,
+      "loss": 1.9813,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925664055035353e-06,
+      "loss": 2.0105,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9255685075482515e-06,
+      "loss": 2.062,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9254729600611505e-06,
+      "loss": 2.3613,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92537741257405e-06,
+      "loss": 2.2717,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925281865086949e-06,
+      "loss": 2.2543,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925186317599848e-06,
+      "loss": 1.9728,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.925090770112747e-06,
+      "loss": 2.4511,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924995222625646e-06,
+      "loss": 2.278,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924899675138544e-06,
+      "loss": 2.298,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924804127651443e-06,
+      "loss": 2.1807,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924708580164342e-06,
+      "loss": 1.6767,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924613032677241e-06,
+      "loss": 1.995,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92451748519014e-06,
+      "loss": 2.046,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924421937703039e-06,
+      "loss": 2.1683,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924326390215937e-06,
+      "loss": 2.3526,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924230842728836e-06,
+      "loss": 1.6304,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924135295241735e-06,
+      "loss": 2.5557,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9240397477546344e-06,
+      "loss": 2.0916,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9239442002675335e-06,
+      "loss": 2.3722,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9238486527804325e-06,
+      "loss": 1.9105,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9237531052933315e-06,
+      "loss": 2.1398,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92365755780623e-06,
+      "loss": 2.4175,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923562010319129e-06,
+      "loss": 2.332,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923466462832028e-06,
+      "loss": 2.0364,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923370915344927e-06,
+      "loss": 2.1665,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923275367857826e-06,
+      "loss": 2.0475,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923179820370725e-06,
+      "loss": 2.2888,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923084272883624e-06,
+      "loss": 2.0056,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922988725396522e-06,
+      "loss": 1.7627,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922893177909421e-06,
+      "loss": 2.1962,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92279763042232e-06,
+      "loss": 2.1863,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922702082935219e-06,
+      "loss": 2.2962,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9226065354481175e-06,
+      "loss": 2.0402,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922510987961017e-06,
+      "loss": 2.2587,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922415440473916e-06,
+      "loss": 2.1218,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922319892986815e-06,
+      "loss": 2.0226,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.922224345499714e-06,
+      "loss": 2.1503,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922128798012613e-06,
+      "loss": 2.0211,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.922033250525512e-06,
+      "loss": 2.2397,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92193770303841e-06,
+      "loss": 2.0056,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92184215555131e-06,
+      "loss": 2.3185,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921746608064209e-06,
+      "loss": 2.3704,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921651060577107e-06,
+      "loss": 2.1476,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921555513090006e-06,
+      "loss": 2.1208,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921459965602905e-06,
+      "loss": 2.2942,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921364418115804e-06,
+      "loss": 2.3625,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921268870628702e-06,
+      "loss": 1.9033,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921173323141602e-06,
+      "loss": 2.1443,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.921077775654501e-06,
+      "loss": 1.6957,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9209822281673994e-06,
+      "loss": 1.7378,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9208866806802985e-06,
+      "loss": 2.3352,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9207911331931975e-06,
+      "loss": 2.4386,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9206955857060965e-06,
+      "loss": 2.0537,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920600038218995e-06,
+      "loss": 1.8027,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920504490731895e-06,
+      "loss": 2.3526,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920408943244793e-06,
+      "loss": 2.0854,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920313395757692e-06,
+      "loss": 2.6321,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920217848270591e-06,
+      "loss": 2.1448,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.92012230078349e-06,
+      "loss": 1.8758,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.920026753296389e-06,
+      "loss": 2.0413,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919931205809287e-06,
+      "loss": 2.3104,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919835658322187e-06,
+      "loss": 2.2144,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919740110835085e-06,
+      "loss": 2.4758,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919644563347984e-06,
+      "loss": 1.8765,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919549015860883e-06,
+      "loss": 1.9681,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919453468373782e-06,
+      "loss": 2.0655,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9193579208866805e-06,
+      "loss": 2.623,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91926237339958e-06,
+      "loss": 2.6199,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919166825912479e-06,
+      "loss": 2.2057,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.919071278425378e-06,
+      "loss": 1.7101,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918975730938277e-06,
+      "loss": 2.2584,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918880183451176e-06,
+      "loss": 1.8789,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918784635964075e-06,
+      "loss": 2.1609,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918689088476973e-06,
+      "loss": 1.8155,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918593540989872e-06,
+      "loss": 2.1893,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918497993502771e-06,
+      "loss": 2.1465,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91840244601567e-06,
+      "loss": 1.5133,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918306898528569e-06,
+      "loss": 2.1959,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918211351041468e-06,
+      "loss": 2.1594,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918115803554367e-06,
+      "loss": 1.9794,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.918020256067265e-06,
+      "loss": 1.8904,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9179247085801644e-06,
+      "loss": 2.3575,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9178291610930635e-06,
+      "loss": 1.9514,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9177336136059625e-06,
+      "loss": 2.0544,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9176380661188615e-06,
+      "loss": 1.8028,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917542518631761e-06,
+      "loss": 2.3619,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.91744697114466e-06,
+      "loss": 2.0534,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917351423657558e-06,
+      "loss": 2.1136,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917255876170457e-06,
+      "loss": 2.4101,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.917160328683356e-06,
+      "loss": 2.3279,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.917064781196255e-06,
+      "loss": 1.9355,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916969233709154e-06,
+      "loss": 2.437,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916873686222053e-06,
+      "loss": 1.8155,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916778138734952e-06,
+      "loss": 1.9711,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91668259124785e-06,
+      "loss": 1.8394,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916587043760749e-06,
+      "loss": 1.7307,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916491496273648e-06,
+      "loss": 1.7574,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916395948786547e-06,
+      "loss": 1.5748,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916300401299446e-06,
+      "loss": 2.2211,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9162048538123454e-06,
+      "loss": 2.3939,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9161093063252445e-06,
+      "loss": 1.8482,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.916013758838143e-06,
+      "loss": 2.0218,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915918211351042e-06,
+      "loss": 2.4279,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915822663863941e-06,
+      "loss": 2.0818,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91572711637684e-06,
+      "loss": 2.3697,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915631568889739e-06,
+      "loss": 1.9592,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915536021402638e-06,
+      "loss": 2.0939,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915440473915537e-06,
+      "loss": 1.7878,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915344926428435e-06,
+      "loss": 2.1751,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915249378941334e-06,
+      "loss": 2.6785,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915153831454233e-06,
+      "loss": 2.2266,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.915058283967132e-06,
+      "loss": 2.3648,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914962736480031e-06,
+      "loss": 2.2714,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91486718899293e-06,
+      "loss": 1.9855,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9147716415058285e-06,
+      "loss": 1.8784,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9146760940187275e-06,
+      "loss": 2.0259,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9145805465316265e-06,
+      "loss": 2.1463,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914484999044526e-06,
+      "loss": 2.2401,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914389451557424e-06,
+      "loss": 2.4721,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914293904070324e-06,
+      "loss": 1.7816,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914198356583223e-06,
+      "loss": 2.2493,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.914102809096121e-06,
+      "loss": 2.2271,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91400726160902e-06,
+      "loss": 1.9384,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913911714121919e-06,
+      "loss": 2.1882,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913816166634818e-06,
+      "loss": 1.859,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913720619147716e-06,
+      "loss": 1.9887,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913625071660616e-06,
+      "loss": 2.4426,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913529524173515e-06,
+      "loss": 1.7993,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913433976686413e-06,
+      "loss": 1.4957,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913338429199312e-06,
+      "loss": 2.0157,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913242881712211e-06,
+      "loss": 1.7581,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9131473342251104e-06,
+      "loss": 1.7055,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.913051786738009e-06,
+      "loss": 2.2613,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9129562392509085e-06,
+      "loss": 2.1941,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9128606917638075e-06,
+      "loss": 2.0565,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912765144276706e-06,
+      "loss": 2.1952,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912669596789605e-06,
+      "loss": 2.0302,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912574049302504e-06,
+      "loss": 2.4967,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912478501815403e-06,
+      "loss": 2.1439,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.912382954328301e-06,
+      "loss": 1.912,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9122874068412e-06,
+      "loss": 1.8512,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9121918593541e-06,
+      "loss": 1.849,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.912096311866998e-06,
+      "loss": 1.9821,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.912000764379897e-06,
+      "loss": 1.7318,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911905216892796e-06,
+      "loss": 2.1969,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911809669405695e-06,
+      "loss": 2.1689,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9117141219185935e-06,
+      "loss": 1.6711,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9116185744314925e-06,
+      "loss": 2.5626,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911523026944392e-06,
+      "loss": 2.3531,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911427479457291e-06,
+      "loss": 2.3078,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.91133193197019e-06,
+      "loss": 2.2974,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911236384483089e-06,
+      "loss": 2.4826,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911140836995988e-06,
+      "loss": 1.7968,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.911045289508886e-06,
+      "loss": 2.4076,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910949742021785e-06,
+      "loss": 2.2726,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910854194534684e-06,
+      "loss": 1.6946,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910758647047583e-06,
+      "loss": 2.2527,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910663099560482e-06,
+      "loss": 2.4924,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910567552073381e-06,
+      "loss": 1.8554,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.91047200458628e-06,
+      "loss": 2.0266,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910376457099178e-06,
+      "loss": 2.3452,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910280909612077e-06,
+      "loss": 2.1322,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.910185362124976e-06,
+      "loss": 2.3051,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9100898146378754e-06,
+      "loss": 2.5263,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9099942671507745e-06,
+      "loss": 2.1406,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9098987196636735e-06,
+      "loss": 2.1396,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909803172176572e-06,
+      "loss": 1.4727,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909707624689471e-06,
+      "loss": 2.399,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.90961207720237e-06,
+      "loss": 1.9886,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909516529715269e-06,
+      "loss": 2.0478,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909420982228168e-06,
+      "loss": 2.2833,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909325434741067e-06,
+      "loss": 1.7529,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909229887253966e-06,
+      "loss": 2.1105,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909134339766864e-06,
+      "loss": 2.1545,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.909038792279763e-06,
+      "loss": 1.9811,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908943244792662e-06,
+      "loss": 2.6713,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908847697305561e-06,
+      "loss": 2.4452,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.90875214981846e-06,
+      "loss": 1.519,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908656602331359e-06,
+      "loss": 2.1319,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908561054844258e-06,
+      "loss": 2.2401,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9084655073571565e-06,
+      "loss": 2.0677,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908369959870056e-06,
+      "loss": 2.4309,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908274412382955e-06,
+      "loss": 1.8451,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908178864895854e-06,
+      "loss": 1.8525,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.908083317408753e-06,
+      "loss": 2.2911,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907987769921652e-06,
+      "loss": 1.9405,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907892222434551e-06,
+      "loss": 2.4027,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907796674947449e-06,
+      "loss": 2.1967,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907701127460348e-06,
+      "loss": 1.7812,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907605579973247e-06,
+      "loss": 2.4126,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907510032486146e-06,
+      "loss": 2.2103,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907414484999045e-06,
+      "loss": 2.5234,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907318937511944e-06,
+      "loss": 2.0592,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.907223390024843e-06,
+      "loss": 2.1239,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.907127842537741e-06,
+      "loss": 2.2658,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9070322950506404e-06,
+      "loss": 2.1355,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9069367475635395e-06,
+      "loss": 2.036,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9068412000764385e-06,
+      "loss": 2.2486,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9067456525893375e-06,
+      "loss": 2.1854,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906650105102237e-06,
+      "loss": 2.5705,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906554557615136e-06,
+      "loss": 2.3559,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906459010128034e-06,
+      "loss": 2.1982,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906363462640933e-06,
+      "loss": 1.8478,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906267915153832e-06,
+      "loss": 1.8515,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.906172367666731e-06,
+      "loss": 2.3389,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90607682017963e-06,
+      "loss": 2.2309,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905981272692529e-06,
+      "loss": 2.0791,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905885725205427e-06,
+      "loss": 2.0192,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905790177718326e-06,
+      "loss": 2.3092,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905694630231225e-06,
+      "loss": 2.0751,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905599082744124e-06,
+      "loss": 1.4501,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905503535257023e-06,
+      "loss": 1.6574,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9054079877699215e-06,
+      "loss": 2.2906,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9053124402828214e-06,
+      "loss": 1.8481,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90521689279572e-06,
+      "loss": 1.8545,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905121345308619e-06,
+      "loss": 2.0191,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.905025797821518e-06,
+      "loss": 1.894,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904930250334417e-06,
+      "loss": 2.184,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904834702847315e-06,
+      "loss": 1.8009,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904739155360214e-06,
+      "loss": 2.1402,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904643607873114e-06,
+      "loss": 2.2468,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904548060386012e-06,
+      "loss": 1.6946,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904452512898911e-06,
+      "loss": 1.907,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90435696541181e-06,
+      "loss": 2.6366,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904261417924709e-06,
+      "loss": 1.796,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904165870437607e-06,
+      "loss": 2.0649,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.904070322950506e-06,
+      "loss": 2.5466,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903974775463406e-06,
+      "loss": 1.7638,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9038792279763045e-06,
+      "loss": 1.9128,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9037836804892035e-06,
+      "loss": 1.9523,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9036881330021025e-06,
+      "loss": 1.8367,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903592585515002e-06,
+      "loss": 1.9438,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9034970380279e-06,
+      "loss": 1.837,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903401490540799e-06,
+      "loss": 1.7526,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903305943053699e-06,
+      "loss": 2.1059,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903210395566597e-06,
+      "loss": 1.8564,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903114848079496e-06,
+      "loss": 2.0581,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.903019300592395e-06,
+      "loss": 1.6661,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902923753105294e-06,
+      "loss": 2.0073,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902828205618192e-06,
+      "loss": 1.335,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902732658131091e-06,
+      "loss": 1.9993,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902637110643991e-06,
+      "loss": 2.3103,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902541563156889e-06,
+      "loss": 1.7535,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902446015669788e-06,
+      "loss": 2.4674,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.902350468182687e-06,
+      "loss": 1.8373,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9022549206955864e-06,
+      "loss": 1.4968,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.902159373208485e-06,
+      "loss": 1.7988,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.902063825721384e-06,
+      "loss": 1.4267,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901968278234283e-06,
+      "loss": 1.9293,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901872730747182e-06,
+      "loss": 1.9112,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901777183260081e-06,
+      "loss": 1.8407,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.90168163577298e-06,
+      "loss": 1.7124,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901586088285879e-06,
+      "loss": 2.4902,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901490540798777e-06,
+      "loss": 2.0714,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901394993311676e-06,
+      "loss": 1.4664,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901299445824575e-06,
+      "loss": 1.9502,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901203898337474e-06,
+      "loss": 2.0324,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901108350850373e-06,
+      "loss": 1.7268,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.901012803363272e-06,
+      "loss": 1.6642,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9009172558761704e-06,
+      "loss": 1.998,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9008217083890695e-06,
+      "loss": 1.8585,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9007261609019685e-06,
+      "loss": 2.3196,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9006306134148675e-06,
+      "loss": 1.9001,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900535065927767e-06,
+      "loss": 2.3094,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900439518440666e-06,
+      "loss": 2.1199,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900343970953565e-06,
+      "loss": 2.1991,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900248423466463e-06,
+      "loss": 2.0413,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900152875979362e-06,
+      "loss": 2.1493,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.900057328492261e-06,
+      "loss": 2.0434,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.89996178100516e-06,
+      "loss": 2.085,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899866233518059e-06,
+      "loss": 1.9474,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899770686030958e-06,
+      "loss": 2.3314,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899675138543857e-06,
+      "loss": 2.2855,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899579591056755e-06,
+      "loss": 2.001,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899484043569654e-06,
+      "loss": 2.0789,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899388496082553e-06,
+      "loss": 2.2006,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.899292948595452e-06,
+      "loss": 1.909,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8991974011083514e-06,
+      "loss": 1.9962,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8991018536212505e-06,
+      "loss": 2.068,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8990063061341495e-06,
+      "loss": 2.0701,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898910758647048e-06,
+      "loss": 1.9675,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898815211159947e-06,
+      "loss": 2.2151,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898719663672846e-06,
+      "loss": 1.8412,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898624116185745e-06,
+      "loss": 1.671,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898528568698643e-06,
+      "loss": 2.0545,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898433021211543e-06,
+      "loss": 2.5074,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898337473724442e-06,
+      "loss": 1.7467,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.89824192623734e-06,
+      "loss": 2.4624,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898146378750239e-06,
+      "loss": 1.695,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.898050831263138e-06,
+      "loss": 2.4132,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897955283776037e-06,
+      "loss": 2.3648,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8978597362889354e-06,
+      "loss": 1.9344,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897764188801835e-06,
+      "loss": 1.591,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897668641314734e-06,
+      "loss": 2.2968,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8975730938276325e-06,
+      "loss": 2.3324,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897477546340532e-06,
+      "loss": 1.6897,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897381998853431e-06,
+      "loss": 2.0646,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.89728645136633e-06,
+      "loss": 1.9582,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897190903879228e-06,
+      "loss": 1.6804,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.897095356392128e-06,
+      "loss": 1.9385,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896999808905027e-06,
+      "loss": 2.172,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896904261417925e-06,
+      "loss": 1.6791,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896808713930824e-06,
+      "loss": 1.3082,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896713166443723e-06,
+      "loss": 2.1027,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896617618956622e-06,
+      "loss": 2.015,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89652207146952e-06,
+      "loss": 1.8836,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89642652398242e-06,
+      "loss": 2.3876,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896330976495318e-06,
+      "loss": 1.7105,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.896235429008217e-06,
+      "loss": 2.1863,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8961398815211164e-06,
+      "loss": 2.2125,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8960443340340155e-06,
+      "loss": 1.8211,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895948786546914e-06,
+      "loss": 2.1313,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895853239059813e-06,
+      "loss": 2.2623,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895757691572713e-06,
+      "loss": 1.8296,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895662144085611e-06,
+      "loss": 1.8434,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89556659659851e-06,
+      "loss": 1.7523,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895471049111409e-06,
+      "loss": 2.4504,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895375501624308e-06,
+      "loss": 1.9684,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895279954137206e-06,
+      "loss": 2.0721,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895184406650105e-06,
+      "loss": 1.8427,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.895088859163004e-06,
+      "loss": 2.1967,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894993311675903e-06,
+      "loss": 1.9367,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894897764188802e-06,
+      "loss": 1.9746,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894802216701701e-06,
+      "loss": 2.1645,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8947066692146e-06,
+      "loss": 1.7627,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8946111217274985e-06,
+      "loss": 1.8552,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8945155742403975e-06,
+      "loss": 2.3038,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894420026753297e-06,
+      "loss": 2.1021,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894324479266196e-06,
+      "loss": 1.9341,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894228931779095e-06,
+      "loss": 2.1327,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894133384291994e-06,
+      "loss": 1.928,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.894037836804893e-06,
+      "loss": 2.4714,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893942289317791e-06,
+      "loss": 1.9696,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.89384674183069e-06,
+      "loss": 1.9622,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893751194343589e-06,
+      "loss": 2.0723,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893655646856488e-06,
+      "loss": 2.6488,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893560099369387e-06,
+      "loss": 1.7589,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893464551882286e-06,
+      "loss": 1.7536,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893369004395185e-06,
+      "loss": 2.0549,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893273456908083e-06,
+      "loss": 1.7186,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.893177909420982e-06,
+      "loss": 2.1261,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8930823619338814e-06,
+      "loss": 2.023,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8929868144467805e-06,
+      "loss": 2.0973,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8928912669596795e-06,
+      "loss": 2.0956,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8927957194725785e-06,
+      "loss": 1.9709,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892700171985478e-06,
+      "loss": 1.976,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892604624498376e-06,
+      "loss": 1.826,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892509077011275e-06,
+      "loss": 2.5195,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892413529524174e-06,
+      "loss": 1.6855,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892317982037073e-06,
+      "loss": 1.5041,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.892222434549972e-06,
+      "loss": 1.8757,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.892126887062871e-06,
+      "loss": 1.8417,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.89203133957577e-06,
+      "loss": 2.0483,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891935792088668e-06,
+      "loss": 1.8085,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891840244601567e-06,
+      "loss": 1.7933,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891744697114466e-06,
+      "loss": 1.7784,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891649149627365e-06,
+      "loss": 1.6626,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891553602140264e-06,
+      "loss": 2.0576,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891458054653163e-06,
+      "loss": 2.3171,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891362507166062e-06,
+      "loss": 1.5671,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891266959678961e-06,
+      "loss": 2.0007,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.89117141219186e-06,
+      "loss": 2.0132,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.891075864704759e-06,
+      "loss": 1.9313,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890980317217658e-06,
+      "loss": 1.8045,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890884769730557e-06,
+      "loss": 2.5774,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890789222243456e-06,
+      "loss": 2.1642,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890693674756354e-06,
+      "loss": 1.6987,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890598127269253e-06,
+      "loss": 1.6544,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890502579782152e-06,
+      "loss": 1.8325,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890407032295051e-06,
+      "loss": 2.0142,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890311484807949e-06,
+      "loss": 1.7861,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890215937320849e-06,
+      "loss": 1.469,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.890120389833748e-06,
+      "loss": 1.8044,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8900248423466464e-06,
+      "loss": 1.6797,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8899292948595455e-06,
+      "loss": 1.9556,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8898337473724445e-06,
+      "loss": 1.9055,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8897381998853435e-06,
+      "loss": 1.6043,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889642652398242e-06,
+      "loss": 2.344,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889547104911142e-06,
+      "loss": 2.0561,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889451557424041e-06,
+      "loss": 1.9665,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889356009936939e-06,
+      "loss": 2.0009,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889260462449838e-06,
+      "loss": 1.6678,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889164914962737e-06,
+      "loss": 1.8268,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.889069367475636e-06,
+      "loss": 2.0841,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888973819988534e-06,
+      "loss": 1.8056,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888878272501434e-06,
+      "loss": 1.7162,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888782725014333e-06,
+      "loss": 1.7584,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888687177527231e-06,
+      "loss": 1.9625,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.88859163004013e-06,
+      "loss": 2.6535,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888496082553029e-06,
+      "loss": 1.9996,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888400535065928e-06,
+      "loss": 2.042,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888304987578827e-06,
+      "loss": 1.9084,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888209440091726e-06,
+      "loss": 2.0954,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8881138926046255e-06,
+      "loss": 1.7035,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.888018345117524e-06,
+      "loss": 1.7284,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887922797630423e-06,
+      "loss": 1.6469,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887827250143322e-06,
+      "loss": 1.6306,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887731702656221e-06,
+      "loss": 1.9381,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887636155169119e-06,
+      "loss": 1.7869,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887540607682018e-06,
+      "loss": 1.859,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887445060194917e-06,
+      "loss": 1.8156,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887349512707816e-06,
+      "loss": 2.3783,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.887253965220715e-06,
+      "loss": 1.9243,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.887158417733614e-06,
+      "loss": 2.0586,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.887062870246513e-06,
+      "loss": 1.9106,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8869673227594114e-06,
+      "loss": 2.2004,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8868717752723105e-06,
+      "loss": 1.4518,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8867762277852095e-06,
+      "loss": 1.8559,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8866806802981085e-06,
+      "loss": 1.755,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886585132811008e-06,
+      "loss": 2.1236,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886489585323907e-06,
+      "loss": 1.9106,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886394037836805e-06,
+      "loss": 2.4691,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886298490349704e-06,
+      "loss": 2.1143,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886202942862603e-06,
+      "loss": 1.6192,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886107395375502e-06,
+      "loss": 2.0939,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.886011847888401e-06,
+      "loss": 1.4328,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8859163004013e-06,
+      "loss": 1.7287,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885820752914199e-06,
+      "loss": 2.0411,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885725205427097e-06,
+      "loss": 2.1075,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885629657939996e-06,
+      "loss": 1.6995,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885534110452895e-06,
+      "loss": 1.9215,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885438562965794e-06,
+      "loss": 1.8241,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.885343015478693e-06,
+      "loss": 1.8385,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8852474679915924e-06,
+      "loss": 1.7548,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8851519205044915e-06,
+      "loss": 1.73,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.88505637301739e-06,
+      "loss": 1.9014,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884960825530289e-06,
+      "loss": 2.142,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884865278043188e-06,
+      "loss": 2.1645,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884769730556087e-06,
+      "loss": 2.3346,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884674183068986e-06,
+      "loss": 2.2978,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884578635581885e-06,
+      "loss": 1.9579,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884483088094784e-06,
+      "loss": 2.113,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884387540607682e-06,
+      "loss": 1.8684,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884291993120581e-06,
+      "loss": 1.7332,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.88419644563348e-06,
+      "loss": 2.2135,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884100898146379e-06,
+      "loss": 2.2478,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884005350659278e-06,
+      "loss": 1.7405,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883909803172177e-06,
+      "loss": 1.9808,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883814255685076e-06,
+      "loss": 1.7419,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8837187081979745e-06,
+      "loss": 1.9062,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8836231607108735e-06,
+      "loss": 1.7612,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883527613223773e-06,
+      "loss": 2.0255,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883432065736672e-06,
+      "loss": 1.6999,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883336518249571e-06,
+      "loss": 2.2434,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.88324097076247e-06,
+      "loss": 2.3168,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883145423275369e-06,
+      "loss": 1.9405,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.883049875788267e-06,
+      "loss": 1.9516,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882954328301166e-06,
+      "loss": 1.6234,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882858780814065e-06,
+      "loss": 1.9488,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882763233326964e-06,
+      "loss": 1.4424,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882667685839863e-06,
+      "loss": 1.7664,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882572138352762e-06,
+      "loss": 2.0337,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.88247659086566e-06,
+      "loss": 1.8579,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882381043378559e-06,
+      "loss": 1.9261,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882285495891458e-06,
+      "loss": 1.9556,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882189948404357e-06,
+      "loss": 2.4494,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8820944009172565e-06,
+      "loss": 1.9738,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8819988534301555e-06,
+      "loss": 1.6671,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8819033059430545e-06,
+      "loss": 2.268,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881807758455953e-06,
+      "loss": 1.875,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881712210968852e-06,
+      "loss": 1.8952,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881616663481751e-06,
+      "loss": 2.2395,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.88152111599465e-06,
+      "loss": 1.4615,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881425568507548e-06,
+      "loss": 1.7847,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881330021020447e-06,
+      "loss": 2.0788,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881234473533347e-06,
+      "loss": 1.714,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881138926046245e-06,
+      "loss": 1.6808,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.881043378559144e-06,
+      "loss": 2.2025,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880947831072043e-06,
+      "loss": 1.5454,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880852283584942e-06,
+      "loss": 2.0604,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8807567360978405e-06,
+      "loss": 2.1111,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8806611886107395e-06,
+      "loss": 1.4412,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880565641123639e-06,
+      "loss": 2.2576,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880470093636538e-06,
+      "loss": 1.9204,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880374546149437e-06,
+      "loss": 1.9681,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880278998662336e-06,
+      "loss": 2.2558,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880183451175235e-06,
+      "loss": 1.9091,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.880087903688133e-06,
+      "loss": 1.8185,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879992356201032e-06,
+      "loss": 2.6443,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879896808713932e-06,
+      "loss": 1.697,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.87980126122683e-06,
+      "loss": 1.6166,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879705713739729e-06,
+      "loss": 1.6888,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879610166252628e-06,
+      "loss": 1.8915,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879514618765527e-06,
+      "loss": 1.7053,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879419071278425e-06,
+      "loss": 1.9558,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879323523791324e-06,
+      "loss": 2.0875,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879227976304224e-06,
+      "loss": 1.7114,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.879132428817122e-06,
+      "loss": 2.0251,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8790368813300215e-06,
+      "loss": 1.7689,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8789413338429205e-06,
+      "loss": 2.001,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8788457863558195e-06,
+      "loss": 1.5583,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878750238868718e-06,
+      "loss": 1.527,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878654691381617e-06,
+      "loss": 1.9066,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878559143894517e-06,
+      "loss": 2.0667,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878463596407415e-06,
+      "loss": 1.4622,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878368048920314e-06,
+      "loss": 2.013,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878272501433213e-06,
+      "loss": 1.9451,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.878176953946112e-06,
+      "loss": 1.7269,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.87808140645901e-06,
+      "loss": 1.8173,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877985858971909e-06,
+      "loss": 2.172,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877890311484808e-06,
+      "loss": 1.6943,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877794763997707e-06,
+      "loss": 2.0045,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877699216510606e-06,
+      "loss": 1.182,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877603669023505e-06,
+      "loss": 1.8253,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877508121536404e-06,
+      "loss": 1.9779,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8774125740493026e-06,
+      "loss": 1.948,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877317026562202e-06,
+      "loss": 1.6977,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.877221479075101e-06,
+      "loss": 1.8849,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.877125931588e-06,
+      "loss": 1.9971,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.877030384100899e-06,
+      "loss": 2.1501,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876934836613798e-06,
+      "loss": 1.6685,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876839289126696e-06,
+      "loss": 2.0252,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876743741639595e-06,
+      "loss": 2.1437,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876648194152494e-06,
+      "loss": 1.7406,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876552646665393e-06,
+      "loss": 1.8631,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876457099178292e-06,
+      "loss": 2.4821,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876361551691191e-06,
+      "loss": 2.0063,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.87626600420409e-06,
+      "loss": 1.8428,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876170456716988e-06,
+      "loss": 1.4851,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.876074909229887e-06,
+      "loss": 1.7409,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8759793617427865e-06,
+      "loss": 1.8285,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8758838142556855e-06,
+      "loss": 1.8175,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8757882667685845e-06,
+      "loss": 2.0811,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8756927192814836e-06,
+      "loss": 1.6638,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875597171794383e-06,
+      "loss": 1.7681,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875501624307281e-06,
+      "loss": 2.1163,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.87540607682018e-06,
+      "loss": 2.5426,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875310529333079e-06,
+      "loss": 1.8612,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875214981845978e-06,
+      "loss": 2.0301,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875119434358877e-06,
+      "loss": 1.8061,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.875023886871776e-06,
+      "loss": 2.3864,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874928339384675e-06,
+      "loss": 2.4679,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874832791897573e-06,
+      "loss": 1.8644,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874737244410472e-06,
+      "loss": 1.8918,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874641696923371e-06,
+      "loss": 2.0497,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.87454614943627e-06,
+      "loss": 1.8186,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8744506019491685e-06,
+      "loss": 1.7986,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874355054462068e-06,
+      "loss": 1.7901,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8742595069749675e-06,
+      "loss": 1.4459,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874163959487866e-06,
+      "loss": 2.1941,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.874068412000765e-06,
+      "loss": 1.6953,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873972864513664e-06,
+      "loss": 1.519,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873877317026563e-06,
+      "loss": 1.6232,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873781769539461e-06,
+      "loss": 2.3027,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873686222052361e-06,
+      "loss": 2.1535,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.87359067456526e-06,
+      "loss": 1.6204,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873495127078158e-06,
+      "loss": 2.0842,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873399579591057e-06,
+      "loss": 1.3232,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873304032103956e-06,
+      "loss": 1.9288,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873208484616855e-06,
+      "loss": 1.924,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873112937129753e-06,
+      "loss": 2.0998,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.873017389642653e-06,
+      "loss": 1.7261,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8729218421555515e-06,
+      "loss": 1.8796,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8728262946684505e-06,
+      "loss": 1.7471,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8727307471813495e-06,
+      "loss": 1.6815,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8726351996942486e-06,
+      "loss": 1.7605,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872539652207148e-06,
+      "loss": 1.8112,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872444104720046e-06,
+      "loss": 1.8638,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872348557232946e-06,
+      "loss": 1.9038,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872253009745844e-06,
+      "loss": 1.8427,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872157462258743e-06,
+      "loss": 2.4185,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.872061914771642e-06,
+      "loss": 1.8426,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871966367284541e-06,
+      "loss": 1.7221,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871870819797439e-06,
+      "loss": 1.6617,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871775272310338e-06,
+      "loss": 1.5384,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871679724823238e-06,
+      "loss": 1.493,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871584177336136e-06,
+      "loss": 1.9534,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871488629849035e-06,
+      "loss": 1.9658,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871393082361934e-06,
+      "loss": 2.0281,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871297534874833e-06,
+      "loss": 1.6052,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871201987387732e-06,
+      "loss": 2.0446,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.871106439900631e-06,
+      "loss": 1.9082,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87101089241353e-06,
+      "loss": 2.2934,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870915344926429e-06,
+      "loss": 1.8264,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870819797439328e-06,
+      "loss": 1.9633,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870724249952227e-06,
+      "loss": 1.6368,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870628702465126e-06,
+      "loss": 2.4411,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870533154978024e-06,
+      "loss": 1.6984,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870437607490923e-06,
+      "loss": 1.682,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870342060003822e-06,
+      "loss": 1.5671,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870246512516721e-06,
+      "loss": 2.2976,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.87015096502962e-06,
+      "loss": 1.8548,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.870055417542519e-06,
+      "loss": 2.0961,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869959870055418e-06,
+      "loss": 1.722,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8698643225683165e-06,
+      "loss": 1.9834,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8697687750812155e-06,
+      "loss": 1.9907,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8696732275941145e-06,
+      "loss": 1.9686,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8695776801070136e-06,
+      "loss": 1.5907,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869482132619913e-06,
+      "loss": 1.4622,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869386585132812e-06,
+      "loss": 1.3538,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869291037645711e-06,
+      "loss": 2.0923,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869195490158609e-06,
+      "loss": 1.8342,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869099942671508e-06,
+      "loss": 1.6292,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.869004395184407e-06,
+      "loss": 2.0471,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868908847697306e-06,
+      "loss": 1.8023,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868813300210205e-06,
+      "loss": 2.2017,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868717752723104e-06,
+      "loss": 1.6006,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868622205236003e-06,
+      "loss": 2.2587,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868526657748901e-06,
+      "loss": 2.2053,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8684311102618e-06,
+      "loss": 1.7592,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868335562774699e-06,
+      "loss": 1.8244,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.868240015287598e-06,
+      "loss": 1.215,
+      "step": 1429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8681444678004975e-06,
+      "loss": 1.7509,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8680489203133965e-06,
+      "loss": 1.8761,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867953372826295e-06,
+      "loss": 2.3502,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867857825339194e-06,
+      "loss": 1.8761,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867762277852093e-06,
+      "loss": 1.6065,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867666730364992e-06,
+      "loss": 1.932,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867571182877891e-06,
+      "loss": 2.5216,
+      "step": 1436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.86747563539079e-06,
+      "loss": 2.6042,
+      "step": 1437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867380087903689e-06,
+      "loss": 1.4173,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867284540416587e-06,
+      "loss": 1.6869,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.867188992929486e-06,
+      "loss": 1.4674,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.867093445442385e-06,
+      "loss": 1.6679,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866997897955284e-06,
+      "loss": 2.0543,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866902350468182e-06,
+      "loss": 1.5843,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866806802981082e-06,
+      "loss": 1.9933,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866711255493981e-06,
+      "loss": 1.7598,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8666157080068795e-06,
+      "loss": 2.2028,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8665201605197786e-06,
+      "loss": 1.5571,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866424613032678e-06,
+      "loss": 2.1512,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866329065545577e-06,
+      "loss": 1.4331,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866233518058475e-06,
+      "loss": 1.5138,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866137970571375e-06,
+      "loss": 1.8892,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866042423084274e-06,
+      "loss": 1.6785,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865946875597172e-06,
+      "loss": 2.3131,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865851328110071e-06,
+      "loss": 1.644,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86575578062297e-06,
+      "loss": 1.6606,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865660233135869e-06,
+      "loss": 1.4846,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865564685648767e-06,
+      "loss": 1.2915,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865469138161667e-06,
+      "loss": 1.469,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865373590674566e-06,
+      "loss": 2.0511,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865278043187464e-06,
+      "loss": 2.0293,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865182495700363e-06,
+      "loss": 1.6083,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8650869482132625e-06,
+      "loss": 1.7336,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8649914007261615e-06,
+      "loss": 1.7005,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86489585323906e-06,
+      "loss": 1.8243,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8648003057519596e-06,
+      "loss": 1.8458,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864704758264859e-06,
+      "loss": 1.4822,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864609210777757e-06,
+      "loss": 1.7579,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864513663290656e-06,
+      "loss": 1.2284,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864418115803555e-06,
+      "loss": 1.7901,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864322568316454e-06,
+      "loss": 1.9208,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864227020829352e-06,
+      "loss": 2.1349,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.864131473342251e-06,
+      "loss": 1.8836,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86403592585515e-06,
+      "loss": 1.7835,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863940378368049e-06,
+      "loss": 2.0267,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863844830880948e-06,
+      "loss": 1.6795,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863749283393847e-06,
+      "loss": 1.3938,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863653735906746e-06,
+      "loss": 2.0242,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8635581884196445e-06,
+      "loss": 1.7365,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8634626409325436e-06,
+      "loss": 1.5882,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863367093445443e-06,
+      "loss": 1.6707,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863271545958342e-06,
+      "loss": 1.713,
+      "step": 1481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.863175998471241e-06,
+      "loss": 1.711,
+      "step": 1482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86308045098414e-06,
+      "loss": 1.5901,
+      "step": 1483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862984903497038e-06,
+      "loss": 1.8361,
+      "step": 1484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862889356009937e-06,
+      "loss": 1.9421,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862793808522836e-06,
+      "loss": 1.5743,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862698261035735e-06,
+      "loss": 1.6498,
+      "step": 1487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862602713548634e-06,
+      "loss": 1.6617,
+      "step": 1488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862507166061533e-06,
+      "loss": 1.769,
+      "step": 1489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862411618574432e-06,
+      "loss": 1.4585,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86231607108733e-06,
+      "loss": 2.1531,
+      "step": 1491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862220523600229e-06,
+      "loss": 1.9737,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.862124976113128e-06,
+      "loss": 2.1042,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8620294286260275e-06,
+      "loss": 2.3392,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8619338811389265e-06,
+      "loss": 1.8931,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8618383336518255e-06,
+      "loss": 2.2495,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8617427861647246e-06,
+      "loss": 2.0269,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861647238677623e-06,
+      "loss": 1.6859,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861551691190522e-06,
+      "loss": 1.751,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861456143703421e-06,
+      "loss": 1.9544,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86136059621632e-06,
+      "loss": 1.727,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861265048729219e-06,
+      "loss": 1.5463,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861169501242118e-06,
+      "loss": 1.4107,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.861073953755017e-06,
+      "loss": 1.5757,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860978406267915e-06,
+      "loss": 1.4594,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860882858780814e-06,
+      "loss": 1.7189,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860787311293713e-06,
+      "loss": 2.0877,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860691763806612e-06,
+      "loss": 1.7128,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860596216319511e-06,
+      "loss": 1.5029,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.86050066883241e-06,
+      "loss": 1.7026,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860405121345309e-06,
+      "loss": 1.5966,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860309573858208e-06,
+      "loss": 1.7059,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860214026371107e-06,
+      "loss": 2.2573,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860118478884006e-06,
+      "loss": 2.2614,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.860022931396905e-06,
+      "loss": 1.8261,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859927383909804e-06,
+      "loss": 2.116,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859831836422703e-06,
+      "loss": 2.0113,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859736288935602e-06,
+      "loss": 2.2319,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8596407414485e-06,
+      "loss": 2.0883,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859545193961399e-06,
+      "loss": 1.724,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859449646474298e-06,
+      "loss": 1.5518,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859354098987197e-06,
+      "loss": 1.4604,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859258551500096e-06,
+      "loss": 2.0836,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859163004012995e-06,
+      "loss": 1.696,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.859067456525894e-06,
+      "loss": 2.0664,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8589719090387925e-06,
+      "loss": 1.4776,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8588763615516915e-06,
+      "loss": 1.914,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8587808140645905e-06,
+      "loss": 2.1085,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8586852665774896e-06,
+      "loss": 1.6943,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858589719090389e-06,
+      "loss": 2.2168,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858494171603288e-06,
+      "loss": 1.8251,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858398624116186e-06,
+      "loss": 1.5739,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858303076629085e-06,
+      "loss": 1.4645,
+      "step": 1533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858207529141984e-06,
+      "loss": 1.7576,
+      "step": 1534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858111981654883e-06,
+      "loss": 2.1359,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858016434167781e-06,
+      "loss": 1.7332,
+      "step": 1536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857920886680681e-06,
+      "loss": 1.794,
+      "step": 1537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.85782533919358e-06,
+      "loss": 2.0321,
+      "step": 1538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857729791706478e-06,
+      "loss": 2.1398,
+      "step": 1539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857634244219377e-06,
+      "loss": 1.8804,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857538696732276e-06,
+      "loss": 1.7891,
+      "step": 1541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857443149245175e-06,
+      "loss": 1.8001,
+      "step": 1542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8573476017580736e-06,
+      "loss": 2.1164,
+      "step": 1543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857252054270973e-06,
+      "loss": 1.441,
+      "step": 1544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8571565067838725e-06,
+      "loss": 1.8011,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.857060959296771e-06,
+      "loss": 1.5803,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85696541180967e-06,
+      "loss": 1.8274,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856869864322569e-06,
+      "loss": 1.298,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856774316835468e-06,
+      "loss": 2.0966,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856678769348366e-06,
+      "loss": 2.1147,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856583221861265e-06,
+      "loss": 1.7286,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856487674374165e-06,
+      "loss": 1.4683,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856392126887063e-06,
+      "loss": 1.803,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856296579399962e-06,
+      "loss": 2.0144,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856201031912861e-06,
+      "loss": 1.8523,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85610548442576e-06,
+      "loss": 1.8027,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.856009936938658e-06,
+      "loss": 1.9736,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8559143894515575e-06,
+      "loss": 1.9556,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.855818841964457e-06,
+      "loss": 2.2003,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8557232944773555e-06,
+      "loss": 1.8835,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8556277469902546e-06,
+      "loss": 2.0132,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.855532199503154e-06,
+      "loss": 1.786,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.855436652016053e-06,
+      "loss": 1.7793,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.855341104528951e-06,
+      "loss": 1.8974,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85524555704185e-06,
+      "loss": 1.7615,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85515000955475e-06,
+      "loss": 2.1206,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.855054462067648e-06,
+      "loss": 1.8098,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854958914580547e-06,
+      "loss": 1.8207,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854863367093446e-06,
+      "loss": 2.0352,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854767819606345e-06,
+      "loss": 1.8562,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854672272119243e-06,
+      "loss": 1.6214,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854576724632142e-06,
+      "loss": 1.7544,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854481177145041e-06,
+      "loss": 1.3988,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85438562965794e-06,
+      "loss": 1.9778,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854290082170839e-06,
+      "loss": 2.0077,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8541945346837385e-06,
+      "loss": 2.1319,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8540989871966375e-06,
+      "loss": 1.576,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.854003439709536e-06,
+      "loss": 1.8206,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853907892222435e-06,
+      "loss": 1.7833,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853812344735334e-06,
+      "loss": 1.3498,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853716797248233e-06,
+      "loss": 1.8682,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853621249761132e-06,
+      "loss": 1.7568,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853525702274031e-06,
+      "loss": 2.0988,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853430154786929e-06,
+      "loss": 1.6009,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853334607299828e-06,
+      "loss": 1.6171,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853239059812727e-06,
+      "loss": 1.4396,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853143512325626e-06,
+      "loss": 2.196,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.853047964838525e-06,
+      "loss": 1.8323,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852952417351424e-06,
+      "loss": 1.8373,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852856869864323e-06,
+      "loss": 2.1734,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8527613223772215e-06,
+      "loss": 1.7169,
+      "step": 1591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8526657748901205e-06,
+      "loss": 1.7377,
+      "step": 1592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8525702274030196e-06,
+      "loss": 2.0429,
+      "step": 1593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852474679915919e-06,
+      "loss": 2.1937,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852379132428818e-06,
+      "loss": 1.7672,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852283584941717e-06,
+      "loss": 2.2153,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852188037454616e-06,
+      "loss": 1.9141,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.852092489967514e-06,
+      "loss": 1.8545,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851996942480413e-06,
+      "loss": 1.65,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851901394993312e-06,
+      "loss": 1.8708,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851805847506211e-06,
+      "loss": 1.7889,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.85171030001911e-06,
+      "loss": 1.8117,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851614752532009e-06,
+      "loss": 1.8216,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851519205044908e-06,
+      "loss": 1.5218,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851423657557806e-06,
+      "loss": 1.8686,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851328110070705e-06,
+      "loss": 1.5891,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.851232562583604e-06,
+      "loss": 2.2062,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8511370150965035e-06,
+      "loss": 2.127,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8510414676094025e-06,
+      "loss": 1.6649,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8509459201223015e-06,
+      "loss": 1.7416,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8508503726352006e-06,
+      "loss": 1.5888,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850754825148099e-06,
+      "loss": 1.6771,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850659277660998e-06,
+      "loss": 1.7894,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850563730173897e-06,
+      "loss": 1.6962,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850468182686796e-06,
+      "loss": 1.5563,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850372635199694e-06,
+      "loss": 1.5148,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850277087712594e-06,
+      "loss": 1.6712,
+      "step": 1617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850181540225493e-06,
+      "loss": 2.21,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.850085992738391e-06,
+      "loss": 1.7354,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.84999044525129e-06,
+      "loss": 2.0932,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849894897764189e-06,
+      "loss": 2.2406,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849799350277088e-06,
+      "loss": 1.4649,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8497038027899865e-06,
+      "loss": 2.0388,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849608255302886e-06,
+      "loss": 1.9227,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8495127078157846e-06,
+      "loss": 0.9385,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849417160328684e-06,
+      "loss": 2.0513,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849321612841583e-06,
+      "loss": 1.6556,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849226065354482e-06,
+      "loss": 1.4091,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849130517867381e-06,
+      "loss": 1.8207,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.849034970380279e-06,
+      "loss": 1.6756,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848939422893179e-06,
+      "loss": 1.5915,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848843875406077e-06,
+      "loss": 1.7816,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848748327918976e-06,
+      "loss": 1.3694,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848652780431875e-06,
+      "loss": 1.6001,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848557232944774e-06,
+      "loss": 1.4844,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848461685457672e-06,
+      "loss": 1.5925,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848366137970571e-06,
+      "loss": 2.0685,
+      "step": 1637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848270590483471e-06,
+      "loss": 1.2564,
+      "step": 1638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.848175042996369e-06,
+      "loss": 1.9362,
+      "step": 1639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8480794955092685e-06,
+      "loss": 1.5579,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8479839480221675e-06,
+      "loss": 1.8589,
+      "step": 1641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8478884005350665e-06,
+      "loss": 1.3162,
+      "step": 1642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847792853047965e-06,
+      "loss": 1.8732,
+      "step": 1643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847697305560864e-06,
+      "loss": 1.5469,
+      "step": 1644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847601758073764e-06,
+      "loss": 2.0266,
+      "step": 1645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847506210586662e-06,
+      "loss": 2.1633,
+      "step": 1646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847410663099561e-06,
+      "loss": 1.86,
+      "step": 1647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.84731511561246e-06,
+      "loss": 2.2842,
+      "step": 1648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.847219568125359e-06,
+      "loss": 1.6477,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.847124020638257e-06,
+      "loss": 1.9214,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.847028473151156e-06,
+      "loss": 2.1004,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846932925664055e-06,
+      "loss": 1.7314,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846837378176954e-06,
+      "loss": 1.6465,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846741830689853e-06,
+      "loss": 1.6005,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846646283202752e-06,
+      "loss": 1.6508,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846550735715651e-06,
+      "loss": 1.8642,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8464551882285496e-06,
+      "loss": 1.7074,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846359640741449e-06,
+      "loss": 1.6174,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846264093254348e-06,
+      "loss": 1.9444,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846168545767247e-06,
+      "loss": 2.0926,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.846072998280146e-06,
+      "loss": 1.6245,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845977450793045e-06,
+      "loss": 1.8984,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845881903305944e-06,
+      "loss": 1.835,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845786355818842e-06,
+      "loss": 1.8784,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845690808331741e-06,
+      "loss": 1.8922,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84559526084464e-06,
+      "loss": 1.9849,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845499713357539e-06,
+      "loss": 1.6778,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845404165870438e-06,
+      "loss": 1.6102,
+      "step": 1668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845308618383337e-06,
+      "loss": 1.4176,
+      "step": 1669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845213070896236e-06,
+      "loss": 1.7833,
+      "step": 1670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.845117523409134e-06,
+      "loss": 1.8964,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8450219759220335e-06,
+      "loss": 1.3537,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8449264284349325e-06,
+      "loss": 2.0949,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8448308809478315e-06,
+      "loss": 1.3683,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8447353334607306e-06,
+      "loss": 1.6041,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84463978597363e-06,
+      "loss": 1.8834,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844544238486528e-06,
+      "loss": 1.9891,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844448690999427e-06,
+      "loss": 1.2296,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844353143512326e-06,
+      "loss": 1.5713,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844257596025225e-06,
+      "loss": 2.1593,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844162048538124e-06,
+      "loss": 1.989,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.844066501051023e-06,
+      "loss": 1.7054,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843970953563922e-06,
+      "loss": 1.8498,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84387540607682e-06,
+      "loss": 1.9995,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843779858589719e-06,
+      "loss": 1.4802,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843684311102618e-06,
+      "loss": 1.8249,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843588763615517e-06,
+      "loss": 1.9323,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8434932161284155e-06,
+      "loss": 1.2434,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843397668641315e-06,
+      "loss": 1.4411,
+      "step": 1689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8433021211542145e-06,
+      "loss": 1.6283,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843206573667113e-06,
+      "loss": 1.8934,
+      "step": 1691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843111026180012e-06,
+      "loss": 1.8306,
+      "step": 1692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.843015478692911e-06,
+      "loss": 1.5413,
+      "step": 1693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.84291993120581e-06,
+      "loss": 1.5914,
+      "step": 1694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842824383718708e-06,
+      "loss": 1.6789,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842728836231608e-06,
+      "loss": 1.3471,
+      "step": 1696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842633288744507e-06,
+      "loss": 2.2764,
+      "step": 1697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842537741257405e-06,
+      "loss": 1.7959,
+      "step": 1698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842442193770304e-06,
+      "loss": 2.2957,
+      "step": 1699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842346646283203e-06,
+      "loss": 2.1735,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842251098796102e-06,
+      "loss": 2.0688,
+      "step": 1701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842155551309e-06,
+      "loss": 1.9037,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8420600038219e-06,
+      "loss": 2.0582,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841964456334799e-06,
+      "loss": 1.7503,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8418689088476975e-06,
+      "loss": 1.8289,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8417733613605965e-06,
+      "loss": 1.5083,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8416778138734956e-06,
+      "loss": 1.4421,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841582266386395e-06,
+      "loss": 1.8519,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841486718899293e-06,
+      "loss": 1.5916,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841391171412193e-06,
+      "loss": 1.461,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841295623925092e-06,
+      "loss": 1.6471,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84120007643799e-06,
+      "loss": 2.0366,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841104528950889e-06,
+      "loss": 1.8152,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.841008981463788e-06,
+      "loss": 1.632,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840913433976687e-06,
+      "loss": 1.9215,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840817886489585e-06,
+      "loss": 1.3578,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840722339002485e-06,
+      "loss": 2.1601,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840626791515384e-06,
+      "loss": 1.5531,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840531244028282e-06,
+      "loss": 1.6873,
+      "step": 1719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840435696541181e-06,
+      "loss": 1.8774,
+      "step": 1720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.84034014905408e-06,
+      "loss": 2.2963,
+      "step": 1721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8402446015669795e-06,
+      "loss": 1.7889,
+      "step": 1722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840149054079878e-06,
+      "loss": 1.4705,
+      "step": 1723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.840053506592777e-06,
+      "loss": 1.4953,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839957959105676e-06,
+      "loss": 1.9537,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839862411618575e-06,
+      "loss": 1.3316,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839766864131474e-06,
+      "loss": 1.5647,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839671316644373e-06,
+      "loss": 1.9193,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839575769157271e-06,
+      "loss": 1.7219,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83948022167017e-06,
+      "loss": 1.7636,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839384674183069e-06,
+      "loss": 1.8616,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839289126695968e-06,
+      "loss": 1.5109,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839193579208867e-06,
+      "loss": 1.5278,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839098031721766e-06,
+      "loss": 1.3336,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.839002484234665e-06,
+      "loss": 1.7417,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8389069367475635e-06,
+      "loss": 1.6447,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8388113892604625e-06,
+      "loss": 1.5546,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8387158417733615e-06,
+      "loss": 1.5395,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8386202942862606e-06,
+      "loss": 1.8195,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83852474679916e-06,
+      "loss": 1.7628,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.838429199312059e-06,
+      "loss": 1.5445,
+      "step": 1741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.838333651824958e-06,
+      "loss": 1.4321,
+      "step": 1742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.838238104337856e-06,
+      "loss": 2.0598,
+      "step": 1743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.838142556850755e-06,
+      "loss": 1.876,
+      "step": 1744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.838047009363654e-06,
+      "loss": 1.8001,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837951461876553e-06,
+      "loss": 1.8717,
+      "step": 1746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837855914389452e-06,
+      "loss": 1.7828,
+      "step": 1747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837760366902351e-06,
+      "loss": 1.9188,
+      "step": 1748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.83766481941525e-06,
+      "loss": 1.4296,
+      "step": 1749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837569271928148e-06,
+      "loss": 1.5595,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837473724441047e-06,
+      "loss": 1.572,
+      "step": 1751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837378176953946e-06,
+      "loss": 1.6879,
+      "step": 1752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837282629466845e-06,
+      "loss": 1.6797,
+      "step": 1753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8371870819797445e-06,
+      "loss": 2.1585,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8370915344926435e-06,
+      "loss": 1.5606,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8369959870055425e-06,
+      "loss": 1.4285,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836900439518441e-06,
+      "loss": 1.6195,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83680489203134e-06,
+      "loss": 1.6108,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836709344544239e-06,
+      "loss": 1.7308,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836613797057138e-06,
+      "loss": 1.8068,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836518249570037e-06,
+      "loss": 1.832,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836422702082936e-06,
+      "loss": 1.3363,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836327154595835e-06,
+      "loss": 1.534,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836231607108733e-06,
+      "loss": 1.8223,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836136059621632e-06,
+      "loss": 1.6535,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.836040512134531e-06,
+      "loss": 1.2871,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83594496464743e-06,
+      "loss": 1.4719,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835849417160329e-06,
+      "loss": 1.9082,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835753869673228e-06,
+      "loss": 1.7641,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835658322186127e-06,
+      "loss": 1.7358,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8355627746990256e-06,
+      "loss": 2.1205,
+      "step": 1771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835467227211925e-06,
+      "loss": 1.82,
+      "step": 1772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835371679724824e-06,
+      "loss": 1.6722,
+      "step": 1773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835276132237723e-06,
+      "loss": 1.5821,
+      "step": 1774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835180584750622e-06,
+      "loss": 1.7951,
+      "step": 1775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.835085037263521e-06,
+      "loss": 2.2201,
+      "step": 1776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834989489776419e-06,
+      "loss": 1.6371,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834893942289318e-06,
+      "loss": 1.8607,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834798394802217e-06,
+      "loss": 1.6553,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834702847315116e-06,
+      "loss": 1.504,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834607299828015e-06,
+      "loss": 2.1523,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834511752340914e-06,
+      "loss": 1.5009,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834416204853813e-06,
+      "loss": 1.3735,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.834320657366711e-06,
+      "loss": 1.4356,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.83422510987961e-06,
+      "loss": 1.9251,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8341295623925095e-06,
+      "loss": 2.0411,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8340340149054085e-06,
+      "loss": 1.6343,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833938467418307e-06,
+      "loss": 1.4677,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8338429199312066e-06,
+      "loss": 1.433,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833747372444106e-06,
+      "loss": 1.4383,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833651824957004e-06,
+      "loss": 2.2828,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833556277469903e-06,
+      "loss": 1.6444,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833460729982802e-06,
+      "loss": 2.1992,
+      "step": 1793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833365182495701e-06,
+      "loss": 1.6656,
+      "step": 1794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833269635008599e-06,
+      "loss": 1.6216,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833174087521498e-06,
+      "loss": 2.1818,
+      "step": 1796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.833078540034398e-06,
+      "loss": 1.4862,
+      "step": 1797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832982992547296e-06,
+      "loss": 1.6244,
+      "step": 1798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832887445060195e-06,
+      "loss": 1.7855,
+      "step": 1799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832791897573094e-06,
+      "loss": 1.7449,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832696350085993e-06,
+      "loss": 1.9184,
+      "step": 1801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8326008025988915e-06,
+      "loss": 1.5038,
+      "step": 1802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8325052551117906e-06,
+      "loss": 2.1346,
+      "step": 1803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8324097076246905e-06,
+      "loss": 1.3548,
+      "step": 1804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832314160137589e-06,
+      "loss": 1.8102,
+      "step": 1805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832218612650488e-06,
+      "loss": 1.9431,
+      "step": 1806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832123065163387e-06,
+      "loss": 1.5651,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.832027517676286e-06,
+      "loss": 1.675,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831931970189184e-06,
+      "loss": 1.8749,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831836422702083e-06,
+      "loss": 1.2872,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831740875214983e-06,
+      "loss": 1.5076,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831645327727881e-06,
+      "loss": 1.6035,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.83154978024078e-06,
+      "loss": 2.1959,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831454232753679e-06,
+      "loss": 1.6337,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831358685266578e-06,
+      "loss": 1.1982,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831263137779476e-06,
+      "loss": 1.883,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.831167590292375e-06,
+      "loss": 1.7658,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8310720428052745e-06,
+      "loss": 1.7555,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8309764953181735e-06,
+      "loss": 1.4567,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8308809478310725e-06,
+      "loss": 1.5186,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8307854003439716e-06,
+      "loss": 1.6144,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830689852856871e-06,
+      "loss": 1.3649,
+      "step": 1822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830594305369769e-06,
+      "loss": 1.5374,
+      "step": 1823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830498757882668e-06,
+      "loss": 1.8066,
+      "step": 1824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830403210395567e-06,
+      "loss": 1.434,
+      "step": 1825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830307662908466e-06,
+      "loss": 1.828,
+      "step": 1826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830212115421365e-06,
+      "loss": 1.6423,
+      "step": 1827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830116567934264e-06,
+      "loss": 1.7131,
+      "step": 1828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.830021020447162e-06,
+      "loss": 2.2969,
+      "step": 1829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829925472960061e-06,
+      "loss": 1.1942,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.82982992547296e-06,
+      "loss": 1.9425,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829734377985859e-06,
+      "loss": 1.4955,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829638830498758e-06,
+      "loss": 1.5315,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829543283011657e-06,
+      "loss": 1.489,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829447735524556e-06,
+      "loss": 1.657,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829352188037455e-06,
+      "loss": 2.0393,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829256640550354e-06,
+      "loss": 1.6636,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829161093063253e-06,
+      "loss": 2.0821,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.829065545576152e-06,
+      "loss": 1.7285,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828969998089051e-06,
+      "loss": 1.9306,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.82887445060195e-06,
+      "loss": 1.754,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828778903114849e-06,
+      "loss": 2.2583,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828683355627747e-06,
+      "loss": 1.356,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828587808140646e-06,
+      "loss": 1.4774,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828492260653545e-06,
+      "loss": 2.0231,
+      "step": 1845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828396713166444e-06,
+      "loss": 1.3932,
+      "step": 1846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828301165679343e-06,
+      "loss": 1.9975,
+      "step": 1847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828205618192242e-06,
+      "loss": 2.3963,
+      "step": 1848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.828110070705141e-06,
+      "loss": 1.6696,
+      "step": 1849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8280145232180395e-06,
+      "loss": 1.588,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8279189757309385e-06,
+      "loss": 1.7253,
+      "step": 1851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8278234282438375e-06,
+      "loss": 1.4258,
+      "step": 1852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8277278807567366e-06,
+      "loss": 1.4607,
+      "step": 1853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827632333269636e-06,
+      "loss": 1.6479,
+      "step": 1854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827536785782535e-06,
+      "loss": 1.8729,
+      "step": 1855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827441238295434e-06,
+      "loss": 1.288,
+      "step": 1856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827345690808332e-06,
+      "loss": 1.6898,
+      "step": 1857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827250143321231e-06,
+      "loss": 1.8407,
+      "step": 1858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.82715459583413e-06,
+      "loss": 1.9414,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.827059048347029e-06,
+      "loss": 2.2945,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826963500859928e-06,
+      "loss": 1.7691,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826867953372827e-06,
+      "loss": 1.7555,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826772405885726e-06,
+      "loss": 1.2867,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826676858398624e-06,
+      "loss": 1.7873,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826581310911523e-06,
+      "loss": 2.0523,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826485763424422e-06,
+      "loss": 1.4581,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826390215937321e-06,
+      "loss": 1.629,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82629466845022e-06,
+      "loss": 1.7698,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8261991209631195e-06,
+      "loss": 1.7506,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826103573476018e-06,
+      "loss": 1.6376,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.826008025988917e-06,
+      "loss": 1.9375,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825912478501816e-06,
+      "loss": 1.8792,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825816931014715e-06,
+      "loss": 1.7471,
+      "step": 1873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825721383527614e-06,
+      "loss": 1.7875,
+      "step": 1874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825625836040512e-06,
+      "loss": 1.6787,
+      "step": 1875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825530288553412e-06,
+      "loss": 1.5893,
+      "step": 1876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82543474106631e-06,
+      "loss": 1.532,
+      "step": 1877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825339193579209e-06,
+      "loss": 1.6113,
+      "step": 1878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825243646092108e-06,
+      "loss": 1.9582,
+      "step": 1879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825148098605007e-06,
+      "loss": 1.466,
+      "step": 1880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825052551117905e-06,
+      "loss": 1.9245,
+      "step": 1881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8249570036308044e-06,
+      "loss": 1.6835,
+      "step": 1882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824861456143704e-06,
+      "loss": 1.7082,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8247659086566025e-06,
+      "loss": 1.7336,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8246703611695016e-06,
+      "loss": 1.7783,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824574813682401e-06,
+      "loss": 1.5722,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8244792661953e-06,
+      "loss": 1.7903,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824383718708198e-06,
+      "loss": 1.5902,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824288171221097e-06,
+      "loss": 1.6251,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824192623733997e-06,
+      "loss": 1.7785,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824097076246895e-06,
+      "loss": 2.188,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.824001528759794e-06,
+      "loss": 1.6617,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823905981272693e-06,
+      "loss": 1.4178,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823810433785592e-06,
+      "loss": 1.3709,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82371488629849e-06,
+      "loss": 1.642,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823619338811389e-06,
+      "loss": 0.9792,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823523791324289e-06,
+      "loss": 1.7971,
+      "step": 1897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823428243837187e-06,
+      "loss": 1.5055,
+      "step": 1898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823332696350086e-06,
+      "loss": 1.5867,
+      "step": 1899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8232371488629854e-06,
+      "loss": 0.9698,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8231416013758845e-06,
+      "loss": 1.3623,
+      "step": 1901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.823046053888783e-06,
+      "loss": 1.956,
+      "step": 1902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822950506401682e-06,
+      "loss": 1.8879,
+      "step": 1903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822854958914581e-06,
+      "loss": 1.979,
+      "step": 1904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82275941142748e-06,
+      "loss": 1.4016,
+      "step": 1905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822663863940379e-06,
+      "loss": 2.066,
+      "step": 1906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822568316453278e-06,
+      "loss": 1.4738,
+      "step": 1907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822472768966177e-06,
+      "loss": 1.489,
+      "step": 1908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822377221479075e-06,
+      "loss": 1.8789,
+      "step": 1909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822281673991974e-06,
+      "loss": 1.4038,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822186126504873e-06,
+      "loss": 1.8072,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.822090579017772e-06,
+      "loss": 1.7642,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821995031530671e-06,
+      "loss": 1.4604,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.82189948404357e-06,
+      "loss": 1.2275,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821803936556469e-06,
+      "loss": 1.5876,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8217083890693675e-06,
+      "loss": 1.5914,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8216128415822666e-06,
+      "loss": 1.7496,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821517294095166e-06,
+      "loss": 1.8346,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821421746608065e-06,
+      "loss": 1.3188,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821326199120964e-06,
+      "loss": 1.7558,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821230651633863e-06,
+      "loss": 1.9629,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.821135104146762e-06,
+      "loss": 1.7626,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.82103955665966e-06,
+      "loss": 2.055,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820944009172559e-06,
+      "loss": 1.6153,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820848461685458e-06,
+      "loss": 1.5755,
+      "step": 1925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820752914198357e-06,
+      "loss": 1.7374,
+      "step": 1926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820657366711256e-06,
+      "loss": 1.4732,
+      "step": 1927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820561819224155e-06,
+      "loss": 1.9429,
+      "step": 1928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820466271737053e-06,
+      "loss": 1.8607,
+      "step": 1929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820370724249952e-06,
+      "loss": 2.0967,
+      "step": 1930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.820275176762851e-06,
+      "loss": 1.4461,
+      "step": 1931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8201796292757504e-06,
+      "loss": 1.5973,
+      "step": 1932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8200840817886495e-06,
+      "loss": 2.0645,
+      "step": 1933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8199885343015485e-06,
+      "loss": 1.6901,
+      "step": 1934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8198929868144476e-06,
+      "loss": 1.5373,
+      "step": 1935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819797439327346e-06,
+      "loss": 1.6818,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819701891840245e-06,
+      "loss": 1.7392,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819606344353144e-06,
+      "loss": 1.7565,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819510796866043e-06,
+      "loss": 2.2269,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819415249378941e-06,
+      "loss": 1.6651,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819319701891841e-06,
+      "loss": 1.8189,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.81922415440474e-06,
+      "loss": 1.5769,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819128606917638e-06,
+      "loss": 2.0373,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.819033059430537e-06,
+      "loss": 1.3319,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818937511943436e-06,
+      "loss": 1.7848,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818841964456335e-06,
+      "loss": 1.5367,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8187464169692335e-06,
+      "loss": 1.4695,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818650869482133e-06,
+      "loss": 1.4133,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818555321995032e-06,
+      "loss": 1.9072,
+      "step": 1949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818459774507931e-06,
+      "loss": 1.7344,
+      "step": 1950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.81836422702083e-06,
+      "loss": 2.0271,
+      "step": 1951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818268679533729e-06,
+      "loss": 1.6812,
+      "step": 1952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818173132046628e-06,
+      "loss": 1.2872,
+      "step": 1953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818077584559526e-06,
+      "loss": 1.4861,
+      "step": 1954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817982037072426e-06,
+      "loss": 1.6704,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817886489585325e-06,
+      "loss": 1.5281,
+      "step": 1956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817790942098223e-06,
+      "loss": 1.6831,
+      "step": 1957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817695394611122e-06,
+      "loss": 1.8625,
+      "step": 1958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817599847124021e-06,
+      "loss": 1.9078,
+      "step": 1959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.81750429963692e-06,
+      "loss": 1.6993,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817408752149818e-06,
+      "loss": 1.258,
+      "step": 1961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817313204662718e-06,
+      "loss": 1.6499,
+      "step": 1962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817217657175617e-06,
+      "loss": 2.1107,
+      "step": 1963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8171221096885154e-06,
+      "loss": 1.6774,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8170265622014145e-06,
+      "loss": 1.527,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8169310147143135e-06,
+      "loss": 1.7377,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8168354672272126e-06,
+      "loss": 1.7732,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816739919740111e-06,
+      "loss": 1.559,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816644372253011e-06,
+      "loss": 1.4182,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816548824765909e-06,
+      "loss": 1.6505,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816453277278808e-06,
+      "loss": 1.5918,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816357729791707e-06,
+      "loss": 1.4547,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816262182304606e-06,
+      "loss": 1.7734,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816166634817505e-06,
+      "loss": 1.6712,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.816071087330403e-06,
+      "loss": 1.4172,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815975539843302e-06,
+      "loss": 1.9945,
+      "step": 1976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815879992356201e-06,
+      "loss": 1.5162,
+      "step": 1977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8157844448691e-06,
+      "loss": 1.4533,
+      "step": 1978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815688897381999e-06,
+      "loss": 1.6941,
+      "step": 1979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815593349894898e-06,
+      "loss": 1.8614,
+      "step": 1980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8154978024077966e-06,
+      "loss": 1.5684,
+      "step": 1981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815402254920696e-06,
+      "loss": 1.2586,
+      "step": 1982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815306707433595e-06,
+      "loss": 1.5356,
+      "step": 1983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815211159946494e-06,
+      "loss": 1.5382,
+      "step": 1984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815115612459393e-06,
+      "loss": 1.4079,
+      "step": 1985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.815020064972292e-06,
+      "loss": 1.7867,
+      "step": 1986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814924517485191e-06,
+      "loss": 1.6183,
+      "step": 1987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814828969998089e-06,
+      "loss": 1.6244,
+      "step": 1988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814733422510988e-06,
+      "loss": 1.761,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814637875023887e-06,
+      "loss": 2.53,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814542327536786e-06,
+      "loss": 1.3473,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814446780049685e-06,
+      "loss": 1.8148,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814351232562584e-06,
+      "loss": 1.9381,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814255685075483e-06,
+      "loss": 1.8068,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.814160137588381e-06,
+      "loss": 1.5754,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8140645901012804e-06,
+      "loss": 1.911,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8139690426141795e-06,
+      "loss": 1.62,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8138734951270785e-06,
+      "loss": 1.2891,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8137779476399776e-06,
+      "loss": 1.7075,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813682400152877e-06,
+      "loss": 1.7693,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813586852665776e-06,
+      "loss": 1.5813,
+      "step": 2001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813491305178674e-06,
+      "loss": 1.7634,
+      "step": 2002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813395757691573e-06,
+      "loss": 1.1515,
+      "step": 2003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813300210204472e-06,
+      "loss": 1.3782,
+      "step": 2004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813204662717371e-06,
+      "loss": 1.7791,
+      "step": 2005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.81310911523027e-06,
+      "loss": 1.6996,
+      "step": 2006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.813013567743169e-06,
+      "loss": 1.4758,
+      "step": 2007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812918020256068e-06,
+      "loss": 1.2968,
+      "step": 2008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812822472768966e-06,
+      "loss": 1.5321,
+      "step": 2009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812726925281865e-06,
+      "loss": 1.4893,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812631377794764e-06,
+      "loss": 1.9061,
+      "step": 2011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812535830307663e-06,
+      "loss": 1.7089,
+      "step": 2012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812440282820562e-06,
+      "loss": 1.7204,
+      "step": 2013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8123447353334614e-06,
+      "loss": 1.4969,
+      "step": 2014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8122491878463605e-06,
+      "loss": 1.6306,
+      "step": 2015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.812153640359259e-06,
+      "loss": 1.9269,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.812058092872158e-06,
+      "loss": 1.7714,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811962545385057e-06,
+      "loss": 1.6865,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811866997897956e-06,
+      "loss": 1.9437,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811771450410855e-06,
+      "loss": 1.6151,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811675902923754e-06,
+      "loss": 1.7124,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811580355436652e-06,
+      "loss": 1.994,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811484807949551e-06,
+      "loss": 1.7195,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.81138926046245e-06,
+      "loss": 1.6355,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811293712975349e-06,
+      "loss": 1.0686,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811198165488248e-06,
+      "loss": 1.9346,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811102618001147e-06,
+      "loss": 1.4615,
+      "step": 2027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.811007070514046e-06,
+      "loss": 1.4986,
+      "step": 2028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8109115230269445e-06,
+      "loss": 1.341,
+      "step": 2029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8108159755398435e-06,
+      "loss": 1.8268,
+      "step": 2030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8107204280527426e-06,
+      "loss": 1.946,
+      "step": 2031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.810624880565642e-06,
+      "loss": 1.7022,
+      "step": 2032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.81052933307854e-06,
+      "loss": 1.7637,
+      "step": 2033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.81043378559144e-06,
+      "loss": 1.4667,
+      "step": 2034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.810338238104339e-06,
+      "loss": 1.7792,
+      "step": 2035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.810242690617237e-06,
+      "loss": 1.6728,
+      "step": 2036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.810147143130136e-06,
+      "loss": 1.8696,
+      "step": 2037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.810051595643035e-06,
+      "loss": 1.8028,
+      "step": 2038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809956048155934e-06,
+      "loss": 1.7829,
+      "step": 2039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809860500668832e-06,
+      "loss": 1.6616,
+      "step": 2040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809764953181732e-06,
+      "loss": 1.8211,
+      "step": 2041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809669405694631e-06,
+      "loss": 1.8081,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809573858207529e-06,
+      "loss": 1.3174,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809478310720428e-06,
+      "loss": 1.5837,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809382763233327e-06,
+      "loss": 1.8574,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8092872157462264e-06,
+      "loss": 1.76,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809191668259125e-06,
+      "loss": 1.8717,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.809096120772024e-06,
+      "loss": 1.4295,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8090005732849236e-06,
+      "loss": 1.7163,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808905025797822e-06,
+      "loss": 1.5684,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808809478310721e-06,
+      "loss": 1.6037,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.80871393082362e-06,
+      "loss": 1.5858,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808618383336519e-06,
+      "loss": 1.4861,
+      "step": 2053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808522835849417e-06,
+      "loss": 1.914,
+      "step": 2054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808427288362316e-06,
+      "loss": 1.6143,
+      "step": 2055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808331740875216e-06,
+      "loss": 1.8382,
+      "step": 2056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808236193388114e-06,
+      "loss": 1.3723,
+      "step": 2057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808140645901013e-06,
+      "loss": 1.6078,
+      "step": 2058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.808045098413912e-06,
+      "loss": 2.0312,
+      "step": 2059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807949550926811e-06,
+      "loss": 1.9732,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8078540034397095e-06,
+      "loss": 1.9269,
+      "step": 2061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8077584559526085e-06,
+      "loss": 1.4535,
+      "step": 2062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8076629084655076e-06,
+      "loss": 1.9184,
+      "step": 2063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807567360978407e-06,
+      "loss": 1.4261,
+      "step": 2064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807471813491306e-06,
+      "loss": 1.5428,
+      "step": 2065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807376266004205e-06,
+      "loss": 2.0434,
+      "step": 2066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807280718517104e-06,
+      "loss": 1.7833,
+      "step": 2067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807185171030002e-06,
+      "loss": 1.5958,
+      "step": 2068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807089623542901e-06,
+      "loss": 1.7946,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8069940760558e-06,
+      "loss": 2.0937,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806898528568699e-06,
+      "loss": 1.9303,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806802981081598e-06,
+      "loss": 1.4543,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806707433594497e-06,
+      "loss": 1.8708,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806611886107395e-06,
+      "loss": 1.4336,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806516338620294e-06,
+      "loss": 1.6657,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806420791133193e-06,
+      "loss": 1.8206,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806325243646092e-06,
+      "loss": 1.5853,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8062296961589914e-06,
+      "loss": 1.3511,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8061341486718905e-06,
+      "loss": 1.8505,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8060386011847895e-06,
+      "loss": 1.5395,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805943053697688e-06,
+      "loss": 1.3433,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805847506210587e-06,
+      "loss": 1.8379,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805751958723486e-06,
+      "loss": 1.2005,
+      "step": 2083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805656411236385e-06,
+      "loss": 1.4637,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805560863749284e-06,
+      "loss": 1.9413,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805465316262183e-06,
+      "loss": 2.1393,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805369768775082e-06,
+      "loss": 2.0231,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.80527422128798e-06,
+      "loss": 1.8121,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805178673800879e-06,
+      "loss": 1.8804,
+      "step": 2089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.805083126313778e-06,
+      "loss": 1.8834,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804987578826677e-06,
+      "loss": 1.671,
+      "step": 2091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804892031339576e-06,
+      "loss": 1.7699,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804796483852475e-06,
+      "loss": 1.4813,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804700936365374e-06,
+      "loss": 1.5398,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8046053888782726e-06,
+      "loss": 1.8033,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804509841391172e-06,
+      "loss": 1.7479,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804414293904071e-06,
+      "loss": 2.1217,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.80431874641697e-06,
+      "loss": 1.7928,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804223198929869e-06,
+      "loss": 1.8883,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804127651442768e-06,
+      "loss": 1.708,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.804032103955667e-06,
+      "loss": 1.9453,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803936556468565e-06,
+      "loss": 1.6292,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803841008981464e-06,
+      "loss": 2.0468,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803745461494363e-06,
+      "loss": 1.3719,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803649914007262e-06,
+      "loss": 1.5094,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803554366520161e-06,
+      "loss": 1.5989,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.80345881903306e-06,
+      "loss": 2.3265,
+      "step": 2107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803363271545959e-06,
+      "loss": 1.812,
+      "step": 2108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.803267724058857e-06,
+      "loss": 1.4153,
+      "step": 2109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8031721765717564e-06,
+      "loss": 1.8021,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8030766290846555e-06,
+      "loss": 1.909,
+      "step": 2111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8029810815975545e-06,
+      "loss": 1.73,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8028855341104536e-06,
+      "loss": 1.2868,
+      "step": 2113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802789986623353e-06,
+      "loss": 1.4613,
+      "step": 2114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802694439136252e-06,
+      "loss": 1.6676,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.80259889164915e-06,
+      "loss": 1.9647,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802503344162049e-06,
+      "loss": 1.4087,
+      "step": 2117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802407796674948e-06,
+      "loss": 1.2674,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802312249187847e-06,
+      "loss": 1.8569,
+      "step": 2119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802216701700745e-06,
+      "loss": 1.5884,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.802121154213645e-06,
+      "loss": 1.9511,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.802025606726543e-06,
+      "loss": 1.8047,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801930059239442e-06,
+      "loss": 1.3518,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801834511752341e-06,
+      "loss": 1.9662,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.80173896426524e-06,
+      "loss": 1.6802,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8016434167781385e-06,
+      "loss": 1.6186,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8015478692910376e-06,
+      "loss": 1.3709,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8014523218039374e-06,
+      "loss": 1.8428,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801356774316836e-06,
+      "loss": 1.5851,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801261226829735e-06,
+      "loss": 1.3211,
+      "step": 2130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801165679342634e-06,
+      "loss": 1.7393,
+      "step": 2131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801070131855533e-06,
+      "loss": 1.9644,
+      "step": 2132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800974584368431e-06,
+      "loss": 1.6228,
+      "step": 2133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.80087903688133e-06,
+      "loss": 1.5504,
+      "step": 2134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.80078348939423e-06,
+      "loss": 1.5142,
+      "step": 2135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800687941907128e-06,
+      "loss": 1.7698,
+      "step": 2136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800592394420027e-06,
+      "loss": 1.6986,
+      "step": 2137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800496846932926e-06,
+      "loss": 1.7406,
+      "step": 2138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800401299445825e-06,
+      "loss": 1.5467,
+      "step": 2139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800305751958723e-06,
+      "loss": 2.14,
+      "step": 2140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800210204471622e-06,
+      "loss": 1.7072,
+      "step": 2141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.800114656984522e-06,
+      "loss": 1.3385,
+      "step": 2142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8000191094974205e-06,
+      "loss": 1.5411,
+      "step": 2143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7999235620103195e-06,
+      "loss": 1.7097,
+      "step": 2144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7998280145232186e-06,
+      "loss": 1.2904,
+      "step": 2145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799732467036118e-06,
+      "loss": 1.5975,
+      "step": 2146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799636919549016e-06,
+      "loss": 1.7612,
+      "step": 2147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799541372061915e-06,
+      "loss": 1.8749,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799445824574815e-06,
+      "loss": 1.6876,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799350277087713e-06,
+      "loss": 1.4286,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799254729600612e-06,
+      "loss": 1.2119,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.799159182113511e-06,
+      "loss": 1.425,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.79906363462641e-06,
+      "loss": 1.7534,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798968087139308e-06,
+      "loss": 1.3267,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798872539652207e-06,
+      "loss": 1.7142,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798776992165106e-06,
+      "loss": 1.3868,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798681444678005e-06,
+      "loss": 1.4303,
+      "step": 2157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798585897190904e-06,
+      "loss": 1.4861,
+      "step": 2158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798490349703803e-06,
+      "loss": 1.6537,
+      "step": 2159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7983948022167024e-06,
+      "loss": 1.7152,
+      "step": 2160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798299254729601e-06,
+      "loss": 1.4097,
+      "step": 2161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7982037072425e-06,
+      "loss": 2.1741,
+      "step": 2162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798108159755399e-06,
+      "loss": 1.7904,
+      "step": 2163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.798012612268298e-06,
+      "loss": 1.3407,
+      "step": 2164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797917064781197e-06,
+      "loss": 1.7519,
+      "step": 2165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797821517294096e-06,
+      "loss": 1.7062,
+      "step": 2166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797725969806995e-06,
+      "loss": 1.9317,
+      "step": 2167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797630422319893e-06,
+      "loss": 1.1136,
+      "step": 2168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797534874832792e-06,
+      "loss": 1.4951,
+      "step": 2169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797439327345691e-06,
+      "loss": 1.5775,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.79734377985859e-06,
+      "loss": 2.3228,
+      "step": 2171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797248232371489e-06,
+      "loss": 1.5451,
+      "step": 2172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797152684884388e-06,
+      "loss": 1.2546,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7970571373972864e-06,
+      "loss": 1.4488,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7969615899101855e-06,
+      "loss": 1.2791,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7968660424230845e-06,
+      "loss": 1.7453,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7967704949359836e-06,
+      "loss": 1.5827,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796674947448883e-06,
+      "loss": 1.8007,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796579399961782e-06,
+      "loss": 1.8997,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796483852474681e-06,
+      "loss": 1.4214,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796388304987579e-06,
+      "loss": 1.7877,
+      "step": 2181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796292757500478e-06,
+      "loss": 1.5259,
+      "step": 2182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796197210013377e-06,
+      "loss": 1.364,
+      "step": 2183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796101662526276e-06,
+      "loss": 2.0277,
+      "step": 2184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.796006115039175e-06,
+      "loss": 1.955,
+      "step": 2185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795910567552074e-06,
+      "loss": 1.7146,
+      "step": 2186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795815020064973e-06,
+      "loss": 1.4583,
+      "step": 2187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795719472577871e-06,
+      "loss": 1.6889,
+      "step": 2188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.79562392509077e-06,
+      "loss": 1.4308,
+      "step": 2189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795528377603669e-06,
+      "loss": 1.6527,
+      "step": 2190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795432830116568e-06,
+      "loss": 1.6615,
+      "step": 2191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795337282629467e-06,
+      "loss": 1.5643,
+      "step": 2192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7952417351423665e-06,
+      "loss": 1.4049,
+      "step": 2193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7951461876552655e-06,
+      "loss": 1.672,
+      "step": 2194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.795050640168164e-06,
+      "loss": 1.329,
+      "step": 2195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794955092681063e-06,
+      "loss": 1.4069,
+      "step": 2196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794859545193962e-06,
+      "loss": 1.3278,
+      "step": 2197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794763997706861e-06,
+      "loss": 1.6796,
+      "step": 2198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794668450219759e-06,
+      "loss": 1.8734,
+      "step": 2199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794572902732659e-06,
+      "loss": 1.3379,
+      "step": 2200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794477355245558e-06,
+      "loss": 1.7591,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794381807758456e-06,
+      "loss": 1.4355,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794286260271355e-06,
+      "loss": 2.0849,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794190712784254e-06,
+      "loss": 1.7521,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.794095165297153e-06,
+      "loss": 1.7423,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7939996178100514e-06,
+      "loss": 1.6423,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793904070322951e-06,
+      "loss": 1.4187,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.79380852283585e-06,
+      "loss": 1.4274,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7937129753487486e-06,
+      "loss": 1.5574,
+      "step": 2209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793617427861648e-06,
+      "loss": 1.8908,
+      "step": 2210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793521880374547e-06,
+      "loss": 2.0915,
+      "step": 2211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793426332887446e-06,
+      "loss": 1.6302,
+      "step": 2212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793330785400344e-06,
+      "loss": 1.5541,
+      "step": 2213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793235237913244e-06,
+      "loss": 1.4222,
+      "step": 2214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793139690426142e-06,
+      "loss": 1.6232,
+      "step": 2215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.793044142939041e-06,
+      "loss": 1.5052,
+      "step": 2216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.79294859545194e-06,
+      "loss": 1.2283,
+      "step": 2217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792853047964839e-06,
+      "loss": 1.5501,
+      "step": 2218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792757500477738e-06,
+      "loss": 2.023,
+      "step": 2219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792661952990636e-06,
+      "loss": 1.6769,
+      "step": 2220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792566405503536e-06,
+      "loss": 1.5649,
+      "step": 2221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792470858016434e-06,
+      "loss": 1.2573,
+      "step": 2222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792375310529333e-06,
+      "loss": 2.0256,
+      "step": 2223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7922797630422324e-06,
+      "loss": 1.2251,
+      "step": 2224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7921842155551315e-06,
+      "loss": 1.568,
+      "step": 2225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.79208866806803e-06,
+      "loss": 1.9866,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791993120580929e-06,
+      "loss": 1.5969,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791897573093828e-06,
+      "loss": 1.9028,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791802025606727e-06,
+      "loss": 1.453,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791706478119626e-06,
+      "loss": 1.7969,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791610930632525e-06,
+      "loss": 1.1915,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791515383145424e-06,
+      "loss": 1.2988,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791419835658322e-06,
+      "loss": 1.1411,
+      "step": 2233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791324288171221e-06,
+      "loss": 2.0472,
+      "step": 2234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.79122874068412e-06,
+      "loss": 1.8337,
+      "step": 2235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791133193197019e-06,
+      "loss": 1.2469,
+      "step": 2236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.791037645709918e-06,
+      "loss": 1.1275,
+      "step": 2237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790942098222817e-06,
+      "loss": 1.1078,
+      "step": 2238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790846550735716e-06,
+      "loss": 1.6779,
+      "step": 2239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7907510032486145e-06,
+      "loss": 2.0108,
+      "step": 2240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7906554557615136e-06,
+      "loss": 1.7155,
+      "step": 2241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790559908274413e-06,
+      "loss": 1.5873,
+      "step": 2242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790464360787312e-06,
+      "loss": 1.8446,
+      "step": 2243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790368813300211e-06,
+      "loss": 1.8919,
+      "step": 2244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.79027326581311e-06,
+      "loss": 1.7137,
+      "step": 2245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790177718326009e-06,
+      "loss": 1.929,
+      "step": 2246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790082170838907e-06,
+      "loss": 1.907,
+      "step": 2247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789986623351806e-06,
+      "loss": 1.8208,
+      "step": 2248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789891075864705e-06,
+      "loss": 2.3098,
+      "step": 2249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789795528377604e-06,
+      "loss": 1.0131,
+      "step": 2250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789699980890503e-06,
+      "loss": 1.7351,
+      "step": 2251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789604433403402e-06,
+      "loss": 1.7284,
+      "step": 2252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789508885916301e-06,
+      "loss": 1.228,
+      "step": 2253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789413338429199e-06,
+      "loss": 1.5168,
+      "step": 2254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.789317790942098e-06,
+      "loss": 1.9826,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7892222434549974e-06,
+      "loss": 1.4747,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7891266959678965e-06,
+      "loss": 1.9051,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7890311484807955e-06,
+      "loss": 1.3707,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7889356009936946e-06,
+      "loss": 1.4562,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788840053506594e-06,
+      "loss": 1.6673,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788744506019492e-06,
+      "loss": 2.0822,
+      "step": 2261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788648958532391e-06,
+      "loss": 1.9966,
+      "step": 2262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.78855341104529e-06,
+      "loss": 1.8611,
+      "step": 2263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788457863558189e-06,
+      "loss": 1.4235,
+      "step": 2264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788362316071088e-06,
+      "loss": 1.6404,
+      "step": 2265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788266768583987e-06,
+      "loss": 1.9496,
+      "step": 2266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788171221096885e-06,
+      "loss": 1.3207,
+      "step": 2267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.788075673609784e-06,
+      "loss": 1.4718,
+      "step": 2268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787980126122683e-06,
+      "loss": 1.6215,
+      "step": 2269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787884578635582e-06,
+      "loss": 1.8602,
+      "step": 2270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787789031148481e-06,
+      "loss": 1.5827,
+      "step": 2271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.78769348366138e-06,
+      "loss": 1.5123,
+      "step": 2272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787597936174279e-06,
+      "loss": 1.1705,
+      "step": 2273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787502388687178e-06,
+      "loss": 1.3708,
+      "step": 2274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787406841200077e-06,
+      "loss": 1.6029,
+      "step": 2275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787311293712976e-06,
+      "loss": 1.2848,
+      "step": 2276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787215746225875e-06,
+      "loss": 1.5611,
+      "step": 2277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787120198738773e-06,
+      "loss": 1.387,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.787024651251673e-06,
+      "loss": 1.4516,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786929103764572e-06,
+      "loss": 2.1667,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78683355627747e-06,
+      "loss": 1.8052,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786738008790369e-06,
+      "loss": 1.9568,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786642461303268e-06,
+      "loss": 1.6374,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786546913816167e-06,
+      "loss": 1.3352,
+      "step": 2284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786451366329065e-06,
+      "loss": 1.7651,
+      "step": 2285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786355818841965e-06,
+      "loss": 1.731,
+      "step": 2286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.786260271354864e-06,
+      "loss": 1.5912,
+      "step": 2287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7861647238677624e-06,
+      "loss": 1.7041,
+      "step": 2288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7860691763806615e-06,
+      "loss": 0.9175,
+      "step": 2289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7859736288935605e-06,
+      "loss": 2.1384,
+      "step": 2290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7858780814064596e-06,
+      "loss": 1.3304,
+      "step": 2291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785782533919358e-06,
+      "loss": 1.8517,
+      "step": 2292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785686986432258e-06,
+      "loss": 1.6907,
+      "step": 2293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785591438945157e-06,
+      "loss": 1.581,
+      "step": 2294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785495891458055e-06,
+      "loss": 1.4898,
+      "step": 2295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785400343970954e-06,
+      "loss": 1.9996,
+      "step": 2296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785304796483853e-06,
+      "loss": 1.6218,
+      "step": 2297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785209248996752e-06,
+      "loss": 1.9233,
+      "step": 2298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78511370150965e-06,
+      "loss": 1.3782,
+      "step": 2299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.785018154022549e-06,
+      "loss": 1.4524,
+      "step": 2300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784922606535449e-06,
+      "loss": 1.5143,
+      "step": 2301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784827059048347e-06,
+      "loss": 2.271,
+      "step": 2302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784731511561246e-06,
+      "loss": 1.8758,
+      "step": 2303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784635964074145e-06,
+      "loss": 1.4261,
+      "step": 2304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784540416587044e-06,
+      "loss": 1.7653,
+      "step": 2305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784444869099943e-06,
+      "loss": 1.7686,
+      "step": 2306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784349321612842e-06,
+      "loss": 1.4043,
+      "step": 2307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7842537741257415e-06,
+      "loss": 1.4872,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78415822663864e-06,
+      "loss": 1.6528,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.784062679151539e-06,
+      "loss": 1.2051,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783967131664438e-06,
+      "loss": 1.6436,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783871584177337e-06,
+      "loss": 1.823,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783776036690235e-06,
+      "loss": 1.8435,
+      "step": 2313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783680489203134e-06,
+      "loss": 1.8053,
+      "step": 2314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783584941716033e-06,
+      "loss": 1.7562,
+      "step": 2315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783489394228932e-06,
+      "loss": 1.2708,
+      "step": 2316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783393846741831e-06,
+      "loss": 1.5989,
+      "step": 2317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78329829925473e-06,
+      "loss": 2.0407,
+      "step": 2318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.783202751767629e-06,
+      "loss": 1.7298,
+      "step": 2319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7831072042805274e-06,
+      "loss": 1.73,
+      "step": 2320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7830116567934265e-06,
+      "loss": 1.0835,
+      "step": 2321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7829161093063255e-06,
+      "loss": 1.7656,
+      "step": 2322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7828205618192246e-06,
+      "loss": 1.8119,
+      "step": 2323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782725014332124e-06,
+      "loss": 1.7451,
+      "step": 2324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782629466845023e-06,
+      "loss": 1.6105,
+      "step": 2325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782533919357921e-06,
+      "loss": 1.584,
+      "step": 2326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.78243837187082e-06,
+      "loss": 1.448,
+      "step": 2327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782342824383719e-06,
+      "loss": 1.6106,
+      "step": 2328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782247276896618e-06,
+      "loss": 1.7916,
+      "step": 2329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.782151729409517e-06,
+      "loss": 1.3559,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.782056181922416e-06,
+      "loss": 1.2171,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781960634435315e-06,
+      "loss": 1.7854,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781865086948213e-06,
+      "loss": 1.1406,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781769539461112e-06,
+      "loss": 1.561,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781673991974011e-06,
+      "loss": 2.0133,
+      "step": 2335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78157844448691e-06,
+      "loss": 1.5394,
+      "step": 2336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781482896999809e-06,
+      "loss": 1.9894,
+      "step": 2337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7813873495127084e-06,
+      "loss": 1.68,
+      "step": 2338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7812918020256075e-06,
+      "loss": 1.6129,
+      "step": 2339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781196254538506e-06,
+      "loss": 1.3821,
+      "step": 2340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781100707051405e-06,
+      "loss": 1.6534,
+      "step": 2341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781005159564304e-06,
+      "loss": 2.0436,
+      "step": 2342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780909612077203e-06,
+      "loss": 1.6967,
+      "step": 2343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780814064590102e-06,
+      "loss": 1.6166,
+      "step": 2344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780718517103001e-06,
+      "loss": 1.4059,
+      "step": 2345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7806229696159e-06,
+      "loss": 1.3083,
+      "step": 2346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780527422128798e-06,
+      "loss": 1.5519,
+      "step": 2347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780431874641697e-06,
+      "loss": 1.4311,
+      "step": 2348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780336327154596e-06,
+      "loss": 1.3615,
+      "step": 2349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780240779667495e-06,
+      "loss": 1.7526,
+      "step": 2350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780145232180394e-06,
+      "loss": 1.296,
+      "step": 2351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.780049684693293e-06,
+      "loss": 1.6341,
+      "step": 2352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779954137206192e-06,
+      "loss": 2.033,
+      "step": 2353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7798585897190905e-06,
+      "loss": 1.7986,
+      "step": 2354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7797630422319896e-06,
+      "loss": 1.2812,
+      "step": 2355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779667494744889e-06,
+      "loss": 2.3146,
+      "step": 2356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779571947257788e-06,
+      "loss": 1.7481,
+      "step": 2357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779476399770687e-06,
+      "loss": 1.7143,
+      "step": 2358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779380852283586e-06,
+      "loss": 1.7381,
+      "step": 2359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779285304796485e-06,
+      "loss": 1.9643,
+      "step": 2360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779189757309383e-06,
+      "loss": 2.0941,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.779094209822282e-06,
+      "loss": 1.5047,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778998662335181e-06,
+      "loss": 1.8136,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.77890311484808e-06,
+      "loss": 1.4976,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778807567360979e-06,
+      "loss": 1.6539,
+      "step": 2365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778712019873878e-06,
+      "loss": 2.0166,
+      "step": 2366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778616472386776e-06,
+      "loss": 1.7055,
+      "step": 2367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778520924899675e-06,
+      "loss": 1.3548,
+      "step": 2368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778425377412574e-06,
+      "loss": 1.4197,
+      "step": 2369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7783298299254734e-06,
+      "loss": 1.8935,
+      "step": 2370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7782342824383725e-06,
+      "loss": 1.452,
+      "step": 2371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.778138734951271e-06,
+      "loss": 1.4466,
+      "step": 2372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7780431874641706e-06,
+      "loss": 1.3372,
+      "step": 2373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777947639977069e-06,
+      "loss": 1.5483,
+      "step": 2374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777852092489968e-06,
+      "loss": 2.127,
+      "step": 2375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777756545002867e-06,
+      "loss": 1.6044,
+      "step": 2376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777660997515766e-06,
+      "loss": 1.3148,
+      "step": 2377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777565450028664e-06,
+      "loss": 1.4911,
+      "step": 2378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777469902541563e-06,
+      "loss": 1.8016,
+      "step": 2379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777374355054463e-06,
+      "loss": 1.3516,
+      "step": 2380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777278807567361e-06,
+      "loss": 1.9239,
+      "step": 2381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.77718326008026e-06,
+      "loss": 1.6548,
+      "step": 2382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777087712593159e-06,
+      "loss": 1.8414,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776992165106058e-06,
+      "loss": 1.6686,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7768966176189565e-06,
+      "loss": 1.7481,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7768010701318555e-06,
+      "loss": 1.6475,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776705522644755e-06,
+      "loss": 1.2447,
+      "step": 2387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776609975157654e-06,
+      "loss": 1.8855,
+      "step": 2388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776514427670553e-06,
+      "loss": 2.1351,
+      "step": 2389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776418880183452e-06,
+      "loss": 1.4701,
+      "step": 2390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776323332696351e-06,
+      "loss": 1.7666,
+      "step": 2391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776227785209249e-06,
+      "loss": 1.5468,
+      "step": 2392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776132237722148e-06,
+      "loss": 1.3809,
+      "step": 2393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776036690235048e-06,
+      "loss": 1.5331,
+      "step": 2394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775941142747946e-06,
+      "loss": 1.8383,
+      "step": 2395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775845595260845e-06,
+      "loss": 0.9496,
+      "step": 2396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775750047773744e-06,
+      "loss": 1.6518,
+      "step": 2397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775654500286643e-06,
+      "loss": 1.3802,
+      "step": 2398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775558952799541e-06,
+      "loss": 1.5418,
+      "step": 2399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.77546340531244e-06,
+      "loss": 1.5792,
+      "step": 2400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.77536785782534e-06,
+      "loss": 1.4326,
+      "step": 2401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7752723103382384e-06,
+      "loss": 1.7046,
+      "step": 2402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7751767628511375e-06,
+      "loss": 1.7104,
+      "step": 2403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7750812153640365e-06,
+      "loss": 1.4882,
+      "step": 2404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7749856678769356e-06,
+      "loss": 1.9036,
+      "step": 2405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774890120389834e-06,
+      "loss": 1.3121,
+      "step": 2406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774794572902733e-06,
+      "loss": 1.3445,
+      "step": 2407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774699025415632e-06,
+      "loss": 1.5229,
+      "step": 2408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774603477928531e-06,
+      "loss": 1.4569,
+      "step": 2409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.77450793044143e-06,
+      "loss": 1.2375,
+      "step": 2410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774412382954329e-06,
+      "loss": 1.6026,
+      "step": 2411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774316835467228e-06,
+      "loss": 1.7066,
+      "step": 2412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774221287980126e-06,
+      "loss": 1.3117,
+      "step": 2413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774125740493025e-06,
+      "loss": 1.2266,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774030193005924e-06,
+      "loss": 1.5082,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773934645518823e-06,
+      "loss": 1.6176,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773839098031722e-06,
+      "loss": 1.7173,
+      "step": 2417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773743550544621e-06,
+      "loss": 1.4839,
+      "step": 2418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7736480030575196e-06,
+      "loss": 1.8134,
+      "step": 2419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773552455570419e-06,
+      "loss": 1.6784,
+      "step": 2420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773456908083318e-06,
+      "loss": 1.9627,
+      "step": 2421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773361360596217e-06,
+      "loss": 1.2555,
+      "step": 2422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773265813109116e-06,
+      "loss": 1.5607,
+      "step": 2423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773170265622015e-06,
+      "loss": 1.3776,
+      "step": 2424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773074718134914e-06,
+      "loss": 0.9611,
+      "step": 2425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772979170647812e-06,
+      "loss": 1.832,
+      "step": 2426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772883623160711e-06,
+      "loss": 1.8449,
+      "step": 2427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.77278807567361e-06,
+      "loss": 1.1854,
+      "step": 2428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772692528186509e-06,
+      "loss": 1.5365,
+      "step": 2429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772596980699408e-06,
+      "loss": 1.6107,
+      "step": 2430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772501433212307e-06,
+      "loss": 1.447,
+      "step": 2431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772405885725206e-06,
+      "loss": 1.3071,
+      "step": 2432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772310338238104e-06,
+      "loss": 1.5763,
+      "step": 2433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7722147907510034e-06,
+      "loss": 1.9659,
+      "step": 2434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7721192432639025e-06,
+      "loss": 1.4725,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7720236957768015e-06,
+      "loss": 1.6609,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7719281482897006e-06,
+      "loss": 1.4258,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7718326008026e-06,
+      "loss": 1.4721,
+      "step": 2438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771737053315499e-06,
+      "loss": 2.2996,
+      "step": 2439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771641505828397e-06,
+      "loss": 1.1758,
+      "step": 2440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771545958341296e-06,
+      "loss": 1.6991,
+      "step": 2441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771450410854195e-06,
+      "loss": 2.0086,
+      "step": 2442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771354863367094e-06,
+      "loss": 1.481,
+      "step": 2443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771259315879992e-06,
+      "loss": 1.9495,
+      "step": 2444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771163768392892e-06,
+      "loss": 1.7229,
+      "step": 2445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.771068220905791e-06,
+      "loss": 1.8573,
+      "step": 2446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770972673418689e-06,
+      "loss": 1.5464,
+      "step": 2447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770877125931588e-06,
+      "loss": 1.1587,
+      "step": 2448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770781578444487e-06,
+      "loss": 1.6822,
+      "step": 2449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770686030957386e-06,
+      "loss": 1.5459,
+      "step": 2450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7705904834702846e-06,
+      "loss": 1.5867,
+      "step": 2451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7704949359831844e-06,
+      "loss": 1.203,
+      "step": 2452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7703993884960835e-06,
+      "loss": 1.8096,
+      "step": 2453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770303841008982e-06,
+      "loss": 1.785,
+      "step": 2454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770208293521881e-06,
+      "loss": 1.5342,
+      "step": 2455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.77011274603478e-06,
+      "loss": 1.7046,
+      "step": 2456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.770017198547679e-06,
+      "loss": 1.6444,
+      "step": 2457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769921651060577e-06,
+      "loss": 1.9344,
+      "step": 2458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769826103573477e-06,
+      "loss": 1.6035,
+      "step": 2459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769730556086375e-06,
+      "loss": 1.2827,
+      "step": 2460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769635008599274e-06,
+      "loss": 1.5092,
+      "step": 2461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769539461112173e-06,
+      "loss": 1.6942,
+      "step": 2462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769443913625072e-06,
+      "loss": 1.9143,
+      "step": 2463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769348366137971e-06,
+      "loss": 1.8055,
+      "step": 2464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769252818650869e-06,
+      "loss": 2.1214,
+      "step": 2465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.769157271163769e-06,
+      "loss": 1.5344,
+      "step": 2466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7690617236766675e-06,
+      "loss": 1.6669,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7689661761895665e-06,
+      "loss": 1.4716,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7688706287024656e-06,
+      "loss": 1.3922,
+      "step": 2469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768775081215365e-06,
+      "loss": 1.7183,
+      "step": 2470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768679533728263e-06,
+      "loss": 1.4172,
+      "step": 2471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768583986241162e-06,
+      "loss": 1.6949,
+      "step": 2472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768488438754062e-06,
+      "loss": 1.5067,
+      "step": 2473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76839289126696e-06,
+      "loss": 1.7237,
+      "step": 2474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768297343779859e-06,
+      "loss": 1.0451,
+      "step": 2475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768201796292758e-06,
+      "loss": 1.8526,
+      "step": 2476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768106248805657e-06,
+      "loss": 1.4224,
+      "step": 2477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.768010701318555e-06,
+      "loss": 1.5618,
+      "step": 2478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767915153831454e-06,
+      "loss": 1.8847,
+      "step": 2479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767819606344353e-06,
+      "loss": 1.5452,
+      "step": 2480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767724058857252e-06,
+      "loss": 1.6935,
+      "step": 2481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767628511370151e-06,
+      "loss": 1.635,
+      "step": 2482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76753296388305e-06,
+      "loss": 2.1369,
+      "step": 2483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7674374163959494e-06,
+      "loss": 1.6872,
+      "step": 2484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767341868908848e-06,
+      "loss": 1.6108,
+      "step": 2485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767246321421747e-06,
+      "loss": 1.6048,
+      "step": 2486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767150773934646e-06,
+      "loss": 2.115,
+      "step": 2487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.767055226447545e-06,
+      "loss": 1.5826,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766959678960444e-06,
+      "loss": 1.4433,
+      "step": 2489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766864131473343e-06,
+      "loss": 1.3204,
+      "step": 2490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766768583986242e-06,
+      "loss": 1.0033,
+      "step": 2491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76667303649914e-06,
+      "loss": 1.5586,
+      "step": 2492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766577489012039e-06,
+      "loss": 1.3615,
+      "step": 2493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766481941524938e-06,
+      "loss": 1.6615,
+      "step": 2494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766386394037837e-06,
+      "loss": 1.682,
+      "step": 2495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766290846550736e-06,
+      "loss": 1.6484,
+      "step": 2496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766195299063635e-06,
+      "loss": 2.5037,
+      "step": 2497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.766099751576534e-06,
+      "loss": 1.5456,
+      "step": 2498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7660042040894325e-06,
+      "loss": 1.307,
+      "step": 2499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7659086566023315e-06,
+      "loss": 1.5017,
+      "step": 2500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7658131091152306e-06,
+      "loss": 1.4423,
+      "step": 2501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76571756162813e-06,
+      "loss": 1.7414,
+      "step": 2502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765622014141029e-06,
+      "loss": 1.3953,
+      "step": 2503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765526466653928e-06,
+      "loss": 1.5908,
+      "step": 2504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765430919166827e-06,
+      "loss": 1.8437,
+      "step": 2505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765335371679725e-06,
+      "loss": 2.0378,
+      "step": 2506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765239824192624e-06,
+      "loss": 1.3679,
+      "step": 2507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765144276705523e-06,
+      "loss": 1.2671,
+      "step": 2508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.765048729218422e-06,
+      "loss": 1.3776,
+      "step": 2509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764953181731321e-06,
+      "loss": 1.3614,
+      "step": 2510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76485763424422e-06,
+      "loss": 1.4075,
+      "step": 2511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764762086757119e-06,
+      "loss": 1.4613,
+      "step": 2512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764666539270017e-06,
+      "loss": 1.5316,
+      "step": 2513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764570991782916e-06,
+      "loss": 1.1645,
+      "step": 2514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764475444295815e-06,
+      "loss": 1.6575,
+      "step": 2515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7643798968087144e-06,
+      "loss": 1.8168,
+      "step": 2516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7642843493216135e-06,
+      "loss": 1.683,
+      "step": 2517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7641888018345125e-06,
+      "loss": 1.6131,
+      "step": 2518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.764093254347411e-06,
+      "loss": 2.1454,
+      "step": 2519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76399770686031e-06,
+      "loss": 1.569,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763902159373209e-06,
+      "loss": 1.7597,
+      "step": 2521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763806611886108e-06,
+      "loss": 1.3445,
+      "step": 2522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763711064399006e-06,
+      "loss": 1.3305,
+      "step": 2523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763615516911906e-06,
+      "loss": 1.1238,
+      "step": 2524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763519969424805e-06,
+      "loss": 1.5166,
+      "step": 2525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763424421937703e-06,
+      "loss": 1.4677,
+      "step": 2526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763328874450602e-06,
+      "loss": 2.1081,
+      "step": 2527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763233326963501e-06,
+      "loss": 1.2288,
+      "step": 2528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7631377794764e-06,
+      "loss": 1.2374,
+      "step": 2529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7630422319892984e-06,
+      "loss": 1.9604,
+      "step": 2530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762946684502198e-06,
+      "loss": 1.4137,
+      "step": 2531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762851137015097e-06,
+      "loss": 1.7658,
+      "step": 2532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7627555895279956e-06,
+      "loss": 1.6533,
+      "step": 2533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762660042040895e-06,
+      "loss": 1.5724,
+      "step": 2534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762564494553794e-06,
+      "loss": 1.9006,
+      "step": 2535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762468947066693e-06,
+      "loss": 1.098,
+      "step": 2536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762373399579591e-06,
+      "loss": 1.817,
+      "step": 2537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762277852092491e-06,
+      "loss": 2.0968,
+      "step": 2538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76218230460539e-06,
+      "loss": 1.4326,
+      "step": 2539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.762086757118288e-06,
+      "loss": 1.324,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761991209631187e-06,
+      "loss": 1.4976,
+      "step": 2541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761895662144086e-06,
+      "loss": 1.5011,
+      "step": 2542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761800114656985e-06,
+      "loss": 1.1082,
+      "step": 2543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761704567169883e-06,
+      "loss": 1.8327,
+      "step": 2544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761609019682783e-06,
+      "loss": 1.2289,
+      "step": 2545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761513472195682e-06,
+      "loss": 1.8278,
+      "step": 2546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.76141792470858e-06,
+      "loss": 1.7672,
+      "step": 2547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7613223772214794e-06,
+      "loss": 1.4669,
+      "step": 2548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7612268297343785e-06,
+      "loss": 1.2778,
+      "step": 2549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7611312822472775e-06,
+      "loss": 1.5948,
+      "step": 2550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.761035734760176e-06,
+      "loss": 1.2625,
+      "step": 2551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760940187273075e-06,
+      "loss": 1.4785,
+      "step": 2552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760844639785975e-06,
+      "loss": 2.0648,
+      "step": 2553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760749092298873e-06,
+      "loss": 1.4136,
+      "step": 2554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760653544811772e-06,
+      "loss": 1.7617,
+      "step": 2555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760557997324671e-06,
+      "loss": 1.5712,
+      "step": 2556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.76046244983757e-06,
+      "loss": 1.6349,
+      "step": 2557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760366902350468e-06,
+      "loss": 1.4453,
+      "step": 2558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760271354863367e-06,
+      "loss": 1.6344,
+      "step": 2559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760175807376266e-06,
+      "loss": 1.5331,
+      "step": 2560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.760080259889165e-06,
+      "loss": 1.6324,
+      "step": 2561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759984712402064e-06,
+      "loss": 1.4773,
+      "step": 2562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759889164914963e-06,
+      "loss": 1.7072,
+      "step": 2563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759793617427862e-06,
+      "loss": 1.6119,
+      "step": 2564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7596980699407606e-06,
+      "loss": 1.4615,
+      "step": 2565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.75960252245366e-06,
+      "loss": 1.396,
+      "step": 2566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759506974966559e-06,
+      "loss": 1.2027,
+      "step": 2567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759411427479458e-06,
+      "loss": 2.0734,
+      "step": 2568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759315879992357e-06,
+      "loss": 1.9581,
+      "step": 2569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759220332505256e-06,
+      "loss": 1.2358,
+      "step": 2570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759124785018154e-06,
+      "loss": 1.3531,
+      "step": 2571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.759029237531053e-06,
+      "loss": 1.2681,
+      "step": 2572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758933690043952e-06,
+      "loss": 1.7479,
+      "step": 2573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758838142556851e-06,
+      "loss": 1.5506,
+      "step": 2574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.75874259506975e-06,
+      "loss": 2.0153,
+      "step": 2575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758647047582649e-06,
+      "loss": 1.4546,
+      "step": 2576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758551500095548e-06,
+      "loss": 1.5578,
+      "step": 2577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758455952608446e-06,
+      "loss": 1.5427,
+      "step": 2578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.758360405121345e-06,
+      "loss": 1.7561,
+      "step": 2579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7582648576342444e-06,
+      "loss": 1.2965,
+      "step": 2580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7581693101471435e-06,
+      "loss": 1.2294,
+      "step": 2581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7580737626600425e-06,
+      "loss": 1.4046,
+      "step": 2582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7579782151729416e-06,
+      "loss": 1.7038,
+      "step": 2583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757882667685841e-06,
+      "loss": 1.7058,
+      "step": 2584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757787120198739e-06,
+      "loss": 1.6457,
+      "step": 2585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757691572711638e-06,
+      "loss": 2.0628,
+      "step": 2586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757596025224537e-06,
+      "loss": 1.8416,
+      "step": 2587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757500477737436e-06,
+      "loss": 1.412,
+      "step": 2588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757404930250335e-06,
+      "loss": 1.3138,
+      "step": 2589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757309382763234e-06,
+      "loss": 1.6125,
+      "step": 2590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757213835276133e-06,
+      "loss": 1.3691,
+      "step": 2591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.757118287789031e-06,
+      "loss": 1.3871,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.75702274030193e-06,
+      "loss": 1.2882,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756927192814829e-06,
+      "loss": 1.6668,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756831645327728e-06,
+      "loss": 1.6144,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756736097840627e-06,
+      "loss": 1.6414,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756640550353526e-06,
+      "loss": 1.7851,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7565450028664254e-06,
+      "loss": 1.493,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756449455379324e-06,
+      "loss": 1.5321,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756353907892223e-06,
+      "loss": 1.832,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756258360405122e-06,
+      "loss": 1.556,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.756162812918021e-06,
+      "loss": 1.9074,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.75606726543092e-06,
+      "loss": 1.8815,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755971717943819e-06,
+      "loss": 1.6033,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755876170456718e-06,
+      "loss": 2.039,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755780622969616e-06,
+      "loss": 1.7445,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755685075482515e-06,
+      "loss": 1.7314,
+      "step": 2607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755589527995414e-06,
+      "loss": 1.1839,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755493980508313e-06,
+      "loss": 1.7035,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755398433021212e-06,
+      "loss": 1.4847,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.755302885534111e-06,
+      "loss": 1.2335,
+      "step": 2611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7552073380470094e-06,
+      "loss": 1.5863,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7551117905599085e-06,
+      "loss": 1.8292,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7550162430728075e-06,
+      "loss": 1.677,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7549206955857066e-06,
+      "loss": 1.6887,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754825148098606e-06,
+      "loss": 1.7003,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754729600611505e-06,
+      "loss": 1.5296,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754634053124404e-06,
+      "loss": 1.8169,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754538505637302e-06,
+      "loss": 1.7358,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754442958150201e-06,
+      "loss": 1.4922,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7543474106631e-06,
+      "loss": 1.587,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754251863175999e-06,
+      "loss": 1.7079,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754156315688897e-06,
+      "loss": 1.4107,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.754060768201796e-06,
+      "loss": 1.5917,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753965220714696e-06,
+      "loss": 1.4478,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753869673227594e-06,
+      "loss": 1.6604,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753774125740493e-06,
+      "loss": 1.5922,
+      "step": 2627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753678578253392e-06,
+      "loss": 1.4141,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753583030766291e-06,
+      "loss": 1.3673,
+      "step": 2629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.75348748327919e-06,
+      "loss": 1.6873,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753391935792089e-06,
+      "loss": 1.9407,
+      "step": 2631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7532963883049885e-06,
+      "loss": 1.6604,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753200840817887e-06,
+      "loss": 1.6073,
+      "step": 2633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753105293330786e-06,
+      "loss": 1.3151,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.753009745843685e-06,
+      "loss": 1.1625,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752914198356584e-06,
+      "loss": 1.4114,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752818650869482e-06,
+      "loss": 1.5916,
+      "step": 2637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752723103382381e-06,
+      "loss": 1.6833,
+      "step": 2638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752627555895281e-06,
+      "loss": 1.7598,
+      "step": 2639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752532008408179e-06,
+      "loss": 1.3948,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752436460921078e-06,
+      "loss": 1.6955,
+      "step": 2641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752340913433977e-06,
+      "loss": 1.5699,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752245365946876e-06,
+      "loss": 1.9857,
+      "step": 2643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7521498184597744e-06,
+      "loss": 1.3067,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7520542709726735e-06,
+      "loss": 1.6308,
+      "step": 2645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751958723485573e-06,
+      "loss": 1.7416,
+      "step": 2646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7518631759984716e-06,
+      "loss": 1.4802,
+      "step": 2647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751767628511371e-06,
+      "loss": 1.5583,
+      "step": 2648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.75167208102427e-06,
+      "loss": 1.2287,
+      "step": 2649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751576533537169e-06,
+      "loss": 1.3826,
+      "step": 2650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751480986050067e-06,
+      "loss": 1.669,
+      "step": 2651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751385438562966e-06,
+      "loss": 1.7105,
+      "step": 2652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751289891075865e-06,
+      "loss": 1.5842,
+      "step": 2653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751194343588764e-06,
+      "loss": 1.6777,
+      "step": 2654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751098796101663e-06,
+      "loss": 1.4145,
+      "step": 2655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.751003248614562e-06,
+      "loss": 1.5891,
+      "step": 2656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750907701127461e-06,
+      "loss": 1.3432,
+      "step": 2657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750812153640359e-06,
+      "loss": 1.7577,
+      "step": 2658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750716606153258e-06,
+      "loss": 1.6334,
+      "step": 2659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750621058666157e-06,
+      "loss": 1.92,
+      "step": 2660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750525511179056e-06,
+      "loss": 1.2962,
+      "step": 2661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7504299636919554e-06,
+      "loss": 1.6463,
+      "step": 2662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7503344162048545e-06,
+      "loss": 1.5738,
+      "step": 2663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750238868717753e-06,
+      "loss": 1.705,
+      "step": 2664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750143321230652e-06,
+      "loss": 1.8099,
+      "step": 2665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.750047773743551e-06,
+      "loss": 1.5248,
+      "step": 2666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.74995222625645e-06,
+      "loss": 1.9362,
+      "step": 2667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749856678769349e-06,
+      "loss": 1.2394,
+      "step": 2668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749761131282248e-06,
+      "loss": 1.7702,
+      "step": 2669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749665583795147e-06,
+      "loss": 1.1967,
+      "step": 2670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749570036308045e-06,
+      "loss": 1.4261,
+      "step": 2671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749474488820944e-06,
+      "loss": 1.4783,
+      "step": 2672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749378941333843e-06,
+      "loss": 1.3877,
+      "step": 2673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749283393846742e-06,
+      "loss": 1.5202,
+      "step": 2674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.749187846359641e-06,
+      "loss": 1.4506,
+      "step": 2675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.74909229887254e-06,
+      "loss": 1.436,
+      "step": 2676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748996751385439e-06,
+      "loss": 1.5825,
+      "step": 2677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7489012038983375e-06,
+      "loss": 1.7773,
+      "step": 2678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7488056564112365e-06,
+      "loss": 1.4656,
+      "step": 2679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748710108924136e-06,
+      "loss": 1.7636,
+      "step": 2680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748614561437035e-06,
+      "loss": 1.306,
+      "step": 2681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748519013949934e-06,
+      "loss": 1.2564,
+      "step": 2682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748423466462833e-06,
+      "loss": 1.6724,
+      "step": 2683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748327918975732e-06,
+      "loss": 1.8496,
+      "step": 2684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.74823237148863e-06,
+      "loss": 1.5718,
+      "step": 2685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748136824001529e-06,
+      "loss": 1.5285,
+      "step": 2686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.748041276514428e-06,
+      "loss": 1.3742,
+      "step": 2687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747945729027327e-06,
+      "loss": 1.4875,
+      "step": 2688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747850181540226e-06,
+      "loss": 1.8599,
+      "step": 2689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747754634053125e-06,
+      "loss": 1.4978,
+      "step": 2690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747659086566024e-06,
+      "loss": 1.3875,
+      "step": 2691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747563539078922e-06,
+      "loss": 1.8457,
+      "step": 2692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747467991591821e-06,
+      "loss": 1.1818,
+      "step": 2693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7473724441047204e-06,
+      "loss": 1.9781,
+      "step": 2694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7472768966176195e-06,
+      "loss": 1.5548,
+      "step": 2695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.747181349130518e-06,
+      "loss": 1.6575,
+      "step": 2696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7470858016434175e-06,
+      "loss": 1.3003,
+      "step": 2697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746990254156317e-06,
+      "loss": 1.439,
+      "step": 2698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746894706669215e-06,
+      "loss": 1.6124,
+      "step": 2699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746799159182114e-06,
+      "loss": 1.6861,
+      "step": 2700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746703611695013e-06,
+      "loss": 1.3394,
+      "step": 2701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746608064207912e-06,
+      "loss": 1.8633,
+      "step": 2702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74651251672081e-06,
+      "loss": 1.3331,
+      "step": 2703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74641696923371e-06,
+      "loss": 1.3116,
+      "step": 2704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746321421746609e-06,
+      "loss": 1.6237,
+      "step": 2705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746225874259507e-06,
+      "loss": 1.4866,
+      "step": 2706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746130326772406e-06,
+      "loss": 1.5414,
+      "step": 2707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.746034779285305e-06,
+      "loss": 1.3193,
+      "step": 2708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745939231798204e-06,
+      "loss": 1.3056,
+      "step": 2709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7458436843111025e-06,
+      "loss": 1.5593,
+      "step": 2710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745748136824002e-06,
+      "loss": 1.7693,
+      "step": 2711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745652589336901e-06,
+      "loss": 1.781,
+      "step": 2712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7455570418498e-06,
+      "loss": 1.496,
+      "step": 2713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745461494362699e-06,
+      "loss": 1.8228,
+      "step": 2714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745365946875598e-06,
+      "loss": 1.8228,
+      "step": 2715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745270399388496e-06,
+      "loss": 1.5794,
+      "step": 2716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745174851901395e-06,
+      "loss": 1.0512,
+      "step": 2717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.745079304414295e-06,
+      "loss": 1.6735,
+      "step": 2718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744983756927193e-06,
+      "loss": 1.6912,
+      "step": 2719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744888209440092e-06,
+      "loss": 1.5228,
+      "step": 2720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744792661952991e-06,
+      "loss": 1.353,
+      "step": 2721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74469711446589e-06,
+      "loss": 1.3558,
+      "step": 2722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744601566978788e-06,
+      "loss": 1.4848,
+      "step": 2723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744506019491687e-06,
+      "loss": 1.2076,
+      "step": 2724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.744410472004587e-06,
+      "loss": 1.5016,
+      "step": 2725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7443149245174854e-06,
+      "loss": 1.4934,
+      "step": 2726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7442193770303845e-06,
+      "loss": 1.615,
+      "step": 2727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7441238295432835e-06,
+      "loss": 1.4262,
+      "step": 2728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7440282820561825e-06,
+      "loss": 1.6291,
+      "step": 2729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743932734569081e-06,
+      "loss": 1.8607,
+      "step": 2730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74383718708198e-06,
+      "loss": 1.4666,
+      "step": 2731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743741639594879e-06,
+      "loss": 1.6818,
+      "step": 2732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743646092107778e-06,
+      "loss": 2.1896,
+      "step": 2733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743550544620677e-06,
+      "loss": 1.7571,
+      "step": 2734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743454997133576e-06,
+      "loss": 1.1456,
+      "step": 2735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743359449646475e-06,
+      "loss": 1.4564,
+      "step": 2736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743263902159373e-06,
+      "loss": 1.3628,
+      "step": 2737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743168354672272e-06,
+      "loss": 1.7492,
+      "step": 2738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743072807185171e-06,
+      "loss": 1.8761,
+      "step": 2739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.74297725969807e-06,
+      "loss": 1.7207,
+      "step": 2740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742881712210969e-06,
+      "loss": 1.6536,
+      "step": 2741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742786164723868e-06,
+      "loss": 1.5134,
+      "step": 2742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742690617236767e-06,
+      "loss": 1.5894,
+      "step": 2743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742595069749666e-06,
+      "loss": 1.7508,
+      "step": 2744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742499522262565e-06,
+      "loss": 2.0465,
+      "step": 2745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742403974775464e-06,
+      "loss": 1.7573,
+      "step": 2746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742308427288363e-06,
+      "loss": 1.5944,
+      "step": 2747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742212879801262e-06,
+      "loss": 1.5874,
+      "step": 2748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742117332314161e-06,
+      "loss": 1.7482,
+      "step": 2749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74202178482706e-06,
+      "loss": 1.7237,
+      "step": 2750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741926237339958e-06,
+      "loss": 1.5705,
+      "step": 2751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741830689852857e-06,
+      "loss": 1.5375,
+      "step": 2752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741735142365756e-06,
+      "loss": 1.4207,
+      "step": 2753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741639594878655e-06,
+      "loss": 1.3204,
+      "step": 2754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741544047391554e-06,
+      "loss": 1.375,
+      "step": 2755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741448499904453e-06,
+      "loss": 1.5894,
+      "step": 2756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741352952417352e-06,
+      "loss": 1.5431,
+      "step": 2757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7412574049302504e-06,
+      "loss": 1.6423,
+      "step": 2758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7411618574431495e-06,
+      "loss": 1.7914,
+      "step": 2759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7410663099560485e-06,
+      "loss": 1.4404,
+      "step": 2760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7409707624689475e-06,
+      "loss": 1.3151,
+      "step": 2761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740875214981847e-06,
+      "loss": 1.4574,
+      "step": 2762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740779667494746e-06,
+      "loss": 1.4905,
+      "step": 2763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740684120007644e-06,
+      "loss": 1.499,
+      "step": 2764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740588572520543e-06,
+      "loss": 1.2375,
+      "step": 2765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740493025033442e-06,
+      "loss": 1.2407,
+      "step": 2766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740397477546341e-06,
+      "loss": 1.4582,
+      "step": 2767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74030193005924e-06,
+      "loss": 1.7359,
+      "step": 2768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740206382572139e-06,
+      "loss": 1.538,
+      "step": 2769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740110835085038e-06,
+      "loss": 1.4778,
+      "step": 2770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740015287597936e-06,
+      "loss": 1.5491,
+      "step": 2771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739919740110835e-06,
+      "loss": 1.6418,
+      "step": 2772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739824192623734e-06,
+      "loss": 1.3709,
+      "step": 2773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739728645136633e-06,
+      "loss": 1.6617,
+      "step": 2774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7396330976495315e-06,
+      "loss": 2.2328,
+      "step": 2775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7395375501624314e-06,
+      "loss": 1.8186,
+      "step": 2776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7394420026753305e-06,
+      "loss": 1.6857,
+      "step": 2777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739346455188229e-06,
+      "loss": 1.5707,
+      "step": 2778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739250907701128e-06,
+      "loss": 1.1869,
+      "step": 2779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739155360214027e-06,
+      "loss": 1.6982,
+      "step": 2780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.739059812726926e-06,
+      "loss": 1.1495,
+      "step": 2781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738964265239824e-06,
+      "loss": 1.484,
+      "step": 2782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738868717752724e-06,
+      "loss": 1.5429,
+      "step": 2783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738773170265623e-06,
+      "loss": 1.586,
+      "step": 2784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738677622778521e-06,
+      "loss": 1.7883,
+      "step": 2785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73858207529142e-06,
+      "loss": 1.6512,
+      "step": 2786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738486527804319e-06,
+      "loss": 1.4269,
+      "step": 2787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738390980317218e-06,
+      "loss": 1.3972,
+      "step": 2788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738295432830116e-06,
+      "loss": 1.6515,
+      "step": 2789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738199885343016e-06,
+      "loss": 1.448,
+      "step": 2790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738104337855915e-06,
+      "loss": 1.434,
+      "step": 2791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7380087903688135e-06,
+      "loss": 1.6193,
+      "step": 2792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7379132428817125e-06,
+      "loss": 1.6507,
+      "step": 2793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737817695394612e-06,
+      "loss": 1.3982,
+      "step": 2794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737722147907511e-06,
+      "loss": 1.1539,
+      "step": 2795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737626600420409e-06,
+      "loss": 1.0392,
+      "step": 2796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737531052933309e-06,
+      "loss": 1.6143,
+      "step": 2797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737435505446208e-06,
+      "loss": 1.2267,
+      "step": 2798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737339957959106e-06,
+      "loss": 1.0124,
+      "step": 2799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737244410472005e-06,
+      "loss": 1.3246,
+      "step": 2800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737148862984904e-06,
+      "loss": 1.5878,
+      "step": 2801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737053315497803e-06,
+      "loss": 1.5333,
+      "step": 2802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736957768010701e-06,
+      "loss": 1.865,
+      "step": 2803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7368622205236e-06,
+      "loss": 1.4178,
+      "step": 2804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736766673036499e-06,
+      "loss": 1.5569,
+      "step": 2805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736671125549398e-06,
+      "loss": 1.4266,
+      "step": 2806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736575578062297e-06,
+      "loss": 1.2539,
+      "step": 2807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7364800305751964e-06,
+      "loss": 1.5981,
+      "step": 2808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7363844830880955e-06,
+      "loss": 1.3755,
+      "step": 2809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736288935600994e-06,
+      "loss": 1.3697,
+      "step": 2810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736193388113893e-06,
+      "loss": 1.9288,
+      "step": 2811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736097840626792e-06,
+      "loss": 1.556,
+      "step": 2812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.736002293139691e-06,
+      "loss": 1.726,
+      "step": 2813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.73590674565259e-06,
+      "loss": 1.7464,
+      "step": 2814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735811198165489e-06,
+      "loss": 1.6707,
+      "step": 2815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735715650678387e-06,
+      "loss": 2.2814,
+      "step": 2816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735620103191286e-06,
+      "loss": 1.3869,
+      "step": 2817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735524555704185e-06,
+      "loss": 1.9166,
+      "step": 2818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735429008217084e-06,
+      "loss": 1.6939,
+      "step": 2819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735333460729983e-06,
+      "loss": 1.6578,
+      "step": 2820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735237913242882e-06,
+      "loss": 1.965,
+      "step": 2821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.735142365755781e-06,
+      "loss": 1.2964,
+      "step": 2822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7350468182686795e-06,
+      "loss": 1.8517,
+      "step": 2823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7349512707815785e-06,
+      "loss": 1.5566,
+      "step": 2824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7348557232944775e-06,
+      "loss": 2.1373,
+      "step": 2825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734760175807377e-06,
+      "loss": 1.383,
+      "step": 2826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734664628320276e-06,
+      "loss": 1.713,
+      "step": 2827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734569080833175e-06,
+      "loss": 1.2304,
+      "step": 2828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734473533346074e-06,
+      "loss": 1.0965,
+      "step": 2829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734377985858972e-06,
+      "loss": 1.6968,
+      "step": 2830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734282438371871e-06,
+      "loss": 1.4229,
+      "step": 2831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.73418689088477e-06,
+      "loss": 1.0602,
+      "step": 2832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734091343397669e-06,
+      "loss": 1.4922,
+      "step": 2833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733995795910568e-06,
+      "loss": 1.2756,
+      "step": 2834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733900248423467e-06,
+      "loss": 1.8383,
+      "step": 2835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733804700936366e-06,
+      "loss": 1.4228,
+      "step": 2836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733709153449264e-06,
+      "loss": 1.3746,
+      "step": 2837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733613605962163e-06,
+      "loss": 1.742,
+      "step": 2838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733518058475062e-06,
+      "loss": 1.3044,
+      "step": 2839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7334225109879614e-06,
+      "loss": 1.614,
+      "step": 2840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7333269635008605e-06,
+      "loss": 1.8148,
+      "step": 2841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7332314160137595e-06,
+      "loss": 1.7126,
+      "step": 2842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7331358685266585e-06,
+      "loss": 1.5747,
+      "step": 2843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733040321039557e-06,
+      "loss": 1.4199,
+      "step": 2844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732944773552456e-06,
+      "loss": 1.382,
+      "step": 2845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732849226065355e-06,
+      "loss": 1.5598,
+      "step": 2846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732753678578254e-06,
+      "loss": 1.711,
+      "step": 2847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732658131091153e-06,
+      "loss": 1.7746,
+      "step": 2848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732562583604052e-06,
+      "loss": 1.3028,
+      "step": 2849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732467036116951e-06,
+      "loss": 1.2634,
+      "step": 2850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732371488629849e-06,
+      "loss": 1.5144,
+      "step": 2851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732275941142748e-06,
+      "loss": 1.3672,
+      "step": 2852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732180393655647e-06,
+      "loss": 1.3134,
+      "step": 2853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732084846168546e-06,
+      "loss": 1.4196,
+      "step": 2854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731989298681445e-06,
+      "loss": 1.4186,
+      "step": 2855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731893751194344e-06,
+      "loss": 1.3189,
+      "step": 2856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7317982037072425e-06,
+      "loss": 1.4901,
+      "step": 2857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731702656220142e-06,
+      "loss": 1.4631,
+      "step": 2858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731607108733041e-06,
+      "loss": 1.5358,
+      "step": 2859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73151156124594e-06,
+      "loss": 1.257,
+      "step": 2860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731416013758839e-06,
+      "loss": 2.0688,
+      "step": 2861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731320466271738e-06,
+      "loss": 1.5232,
+      "step": 2862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731224918784637e-06,
+      "loss": 1.9291,
+      "step": 2863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731129371297535e-06,
+      "loss": 1.4542,
+      "step": 2864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731033823810434e-06,
+      "loss": 1.6986,
+      "step": 2865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730938276323333e-06,
+      "loss": 1.2907,
+      "step": 2866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730842728836232e-06,
+      "loss": 2.1946,
+      "step": 2867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73074718134913e-06,
+      "loss": 1.2066,
+      "step": 2868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.73065163386203e-06,
+      "loss": 1.5312,
+      "step": 2869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730556086374929e-06,
+      "loss": 1.5314,
+      "step": 2870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730460538887827e-06,
+      "loss": 1.2517,
+      "step": 2871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7303649914007264e-06,
+      "loss": 1.0243,
+      "step": 2872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7302694439136255e-06,
+      "loss": 1.5191,
+      "step": 2873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7301738964265245e-06,
+      "loss": 1.6494,
+      "step": 2874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730078348939423e-06,
+      "loss": 1.6829,
+      "step": 2875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729982801452322e-06,
+      "loss": 1.2941,
+      "step": 2876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729887253965222e-06,
+      "loss": 1.1301,
+      "step": 2877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.72979170647812e-06,
+      "loss": 1.3851,
+      "step": 2878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729696158991019e-06,
+      "loss": 1.5842,
+      "step": 2879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729600611503918e-06,
+      "loss": 1.5629,
+      "step": 2880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729505064016817e-06,
+      "loss": 1.5483,
+      "step": 2881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729409516529715e-06,
+      "loss": 1.3486,
+      "step": 2882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729313969042614e-06,
+      "loss": 1.4935,
+      "step": 2883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729218421555514e-06,
+      "loss": 1.6223,
+      "step": 2884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729122874068412e-06,
+      "loss": 1.7758,
+      "step": 2885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729027326581311e-06,
+      "loss": 2.5306,
+      "step": 2886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.72893177909421e-06,
+      "loss": 2.1616,
+      "step": 2887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728836231607109e-06,
+      "loss": 1.3927,
+      "step": 2888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7287406841200075e-06,
+      "loss": 1.3371,
+      "step": 2889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728645136632907e-06,
+      "loss": 1.3601,
+      "step": 2890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7285495891458065e-06,
+      "loss": 1.3203,
+      "step": 2891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728454041658705e-06,
+      "loss": 1.6706,
+      "step": 2892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728358494171604e-06,
+      "loss": 1.7048,
+      "step": 2893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728262946684503e-06,
+      "loss": 1.6247,
+      "step": 2894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728167399197402e-06,
+      "loss": 1.574,
+      "step": 2895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7280718517103e-06,
+      "loss": 1.5128,
+      "step": 2896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727976304223199e-06,
+      "loss": 1.6504,
+      "step": 2897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727880756736099e-06,
+      "loss": 1.2404,
+      "step": 2898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727785209248997e-06,
+      "loss": 1.6973,
+      "step": 2899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727689661761896e-06,
+      "loss": 1.5191,
+      "step": 2900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727594114274795e-06,
+      "loss": 1.398,
+      "step": 2901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727498566787694e-06,
+      "loss": 1.7999,
+      "step": 2902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727403019300592e-06,
+      "loss": 1.7605,
+      "step": 2903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7273074718134914e-06,
+      "loss": 1.4787,
+      "step": 2904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7272119243263905e-06,
+      "loss": 1.3781,
+      "step": 2905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7271163768392895e-06,
+      "loss": 1.5227,
+      "step": 2906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7270208293521885e-06,
+      "loss": 1.7773,
+      "step": 2907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726925281865088e-06,
+      "loss": 1.4713,
+      "step": 2908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726829734377987e-06,
+      "loss": 1.1595,
+      "step": 2909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726734186890885e-06,
+      "loss": 1.5102,
+      "step": 2910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726638639403784e-06,
+      "loss": 1.4575,
+      "step": 2911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726543091916683e-06,
+      "loss": 1.0687,
+      "step": 2912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726447544429582e-06,
+      "loss": 1.7055,
+      "step": 2913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726351996942481e-06,
+      "loss": 1.4591,
+      "step": 2914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.72625644945538e-06,
+      "loss": 1.1448,
+      "step": 2915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726160901968278e-06,
+      "loss": 1.5842,
+      "step": 2916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.726065354481177e-06,
+      "loss": 1.3441,
+      "step": 2917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725969806994076e-06,
+      "loss": 1.3743,
+      "step": 2918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725874259506975e-06,
+      "loss": 2.0097,
+      "step": 2919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725778712019874e-06,
+      "loss": 1.5855,
+      "step": 2920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725683164532773e-06,
+      "loss": 1.5041,
+      "step": 2921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7255876170456724e-06,
+      "loss": 1.5613,
+      "step": 2922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725492069558571e-06,
+      "loss": 1.2053,
+      "step": 2923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.72539652207147e-06,
+      "loss": 1.4317,
+      "step": 2924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725300974584369e-06,
+      "loss": 1.5043,
+      "step": 2925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725205427097268e-06,
+      "loss": 1.6308,
+      "step": 2926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725109879610167e-06,
+      "loss": 1.2669,
+      "step": 2927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.725014332123066e-06,
+      "loss": 1.42,
+      "step": 2928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724918784635965e-06,
+      "loss": 1.585,
+      "step": 2929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724823237148863e-06,
+      "loss": 1.6156,
+      "step": 2930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724727689661762e-06,
+      "loss": 1.2525,
+      "step": 2931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724632142174661e-06,
+      "loss": 1.9098,
+      "step": 2932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.72453659468756e-06,
+      "loss": 1.9655,
+      "step": 2933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724441047200459e-06,
+      "loss": 1.743,
+      "step": 2934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724345499713358e-06,
+      "loss": 1.622,
+      "step": 2935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.724249952226257e-06,
+      "loss": 1.4378,
+      "step": 2936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7241544047391555e-06,
+      "loss": 1.4611,
+      "step": 2937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7240588572520545e-06,
+      "loss": 1.6111,
+      "step": 2938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7239633097649535e-06,
+      "loss": 1.4986,
+      "step": 2939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723867762277853e-06,
+      "loss": 1.4535,
+      "step": 2940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723772214790752e-06,
+      "loss": 1.1367,
+      "step": 2941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723676667303651e-06,
+      "loss": 1.9009,
+      "step": 2942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.72358111981655e-06,
+      "loss": 1.347,
+      "step": 2943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723485572329448e-06,
+      "loss": 1.7711,
+      "step": 2944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723390024842347e-06,
+      "loss": 1.4092,
+      "step": 2945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723294477355246e-06,
+      "loss": 1.1697,
+      "step": 2946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723198929868145e-06,
+      "loss": 1.6656,
+      "step": 2947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723103382381043e-06,
+      "loss": 1.6738,
+      "step": 2948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.723007834893943e-06,
+      "loss": 2.0779,
+      "step": 2949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722912287406842e-06,
+      "loss": 1.3904,
+      "step": 2950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.72281673991974e-06,
+      "loss": 1.2803,
+      "step": 2951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722721192432639e-06,
+      "loss": 1.4601,
+      "step": 2952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722625644945538e-06,
+      "loss": 1.5836,
+      "step": 2953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7225300974584374e-06,
+      "loss": 1.3787,
+      "step": 2954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722434549971336e-06,
+      "loss": 1.4111,
+      "step": 2955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7223390024842355e-06,
+      "loss": 1.351,
+      "step": 2956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722243454997134e-06,
+      "loss": 1.6006,
+      "step": 2957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722147907510033e-06,
+      "loss": 1.7316,
+      "step": 2958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722052360022932e-06,
+      "loss": 1.5995,
+      "step": 2959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721956812535831e-06,
+      "loss": 1.5629,
+      "step": 2960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72186126504873e-06,
+      "loss": 1.7211,
+      "step": 2961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721765717561628e-06,
+      "loss": 1.638,
+      "step": 2962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721670170074528e-06,
+      "loss": 1.1836,
+      "step": 2963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721574622587426e-06,
+      "loss": 1.263,
+      "step": 2964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721479075100325e-06,
+      "loss": 1.5751,
+      "step": 2965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721383527613224e-06,
+      "loss": 1.7982,
+      "step": 2966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.721287980126123e-06,
+      "loss": 1.4056,
+      "step": 2967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7211924326390214e-06,
+      "loss": 1.2985,
+      "step": 2968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7210968851519205e-06,
+      "loss": 1.2979,
+      "step": 2969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72100133766482e-06,
+      "loss": 1.4505,
+      "step": 2970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7209057901777185e-06,
+      "loss": 1.7816,
+      "step": 2971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720810242690618e-06,
+      "loss": 1.3799,
+      "step": 2972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720714695203517e-06,
+      "loss": 1.6752,
+      "step": 2973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720619147716416e-06,
+      "loss": 1.2219,
+      "step": 2974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720523600229314e-06,
+      "loss": 1.4297,
+      "step": 2975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720428052742213e-06,
+      "loss": 1.6404,
+      "step": 2976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720332505255113e-06,
+      "loss": 2.0253,
+      "step": 2977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720236957768011e-06,
+      "loss": 1.2961,
+      "step": 2978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72014141028091e-06,
+      "loss": 1.7925,
+      "step": 2979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720045862793809e-06,
+      "loss": 2.081,
+      "step": 2980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719950315306708e-06,
+      "loss": 1.2623,
+      "step": 2981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719854767819606e-06,
+      "loss": 1.6375,
+      "step": 2982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719759220332505e-06,
+      "loss": 1.6423,
+      "step": 2983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719663672845404e-06,
+      "loss": 1.6434,
+      "step": 2984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719568125358303e-06,
+      "loss": 2.1399,
+      "step": 2985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7194725778712024e-06,
+      "loss": 1.4913,
+      "step": 2986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7193770303841015e-06,
+      "loss": 1.3452,
+      "step": 2987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7192814828970005e-06,
+      "loss": 1.3565,
+      "step": 2988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719185935409899e-06,
+      "loss": 1.6377,
+      "step": 2989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.719090387922798e-06,
+      "loss": 1.3512,
+      "step": 2990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718994840435697e-06,
+      "loss": 1.5426,
+      "step": 2991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718899292948596e-06,
+      "loss": 1.7275,
+      "step": 2992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718803745461495e-06,
+      "loss": 1.5249,
+      "step": 2993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718708197974394e-06,
+      "loss": 1.469,
+      "step": 2994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718612650487293e-06,
+      "loss": 1.6503,
+      "step": 2995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718517103000191e-06,
+      "loss": 1.5416,
+      "step": 2996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71842155551309e-06,
+      "loss": 1.8234,
+      "step": 2997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718326008025989e-06,
+      "loss": 1.715,
+      "step": 2998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718230460538888e-06,
+      "loss": 2.1206,
+      "step": 2999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718134913051787e-06,
+      "loss": 1.3964,
+      "step": 3000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.718039365564686e-06,
+      "loss": 1.4969,
+      "step": 3001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717943818077585e-06,
+      "loss": 1.2067,
+      "step": 3002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7178482705904835e-06,
+      "loss": 1.3244,
+      "step": 3003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717752723103383e-06,
+      "loss": 1.3326,
+      "step": 3004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717657175616282e-06,
+      "loss": 1.0422,
+      "step": 3005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717561628129181e-06,
+      "loss": 1.6627,
+      "step": 3006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.71746608064208e-06,
+      "loss": 1.1898,
+      "step": 3007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717370533154979e-06,
+      "loss": 1.5786,
+      "step": 3008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717274985667877e-06,
+      "loss": 1.9368,
+      "step": 3009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717179438180776e-06,
+      "loss": 1.1737,
+      "step": 3010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.717083890693675e-06,
+      "loss": 1.9695,
+      "step": 3011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716988343206574e-06,
+      "loss": 1.5919,
+      "step": 3012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716892795719473e-06,
+      "loss": 1.717,
+      "step": 3013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716797248232372e-06,
+      "loss": 1.654,
+      "step": 3014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716701700745271e-06,
+      "loss": 1.6028,
+      "step": 3015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716606153258169e-06,
+      "loss": 1.2425,
+      "step": 3016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716510605771068e-06,
+      "loss": 2.0936,
+      "step": 3017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7164150582839674e-06,
+      "loss": 1.6502,
+      "step": 3018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7163195107968665e-06,
+      "loss": 1.8819,
+      "step": 3019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716223963309765e-06,
+      "loss": 1.6159,
+      "step": 3020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7161284158226645e-06,
+      "loss": 1.403,
+      "step": 3021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.716032868335564e-06,
+      "loss": 1.3457,
+      "step": 3022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715937320848462e-06,
+      "loss": 1.183,
+      "step": 3023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715841773361361e-06,
+      "loss": 1.4612,
+      "step": 3024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.71574622587426e-06,
+      "loss": 1.9381,
+      "step": 3025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715650678387159e-06,
+      "loss": 1.3971,
+      "step": 3026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715555130900057e-06,
+      "loss": 1.4642,
+      "step": 3027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715459583412957e-06,
+      "loss": 1.7565,
+      "step": 3028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715364035925856e-06,
+      "loss": 1.7661,
+      "step": 3029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715268488438754e-06,
+      "loss": 1.6434,
+      "step": 3030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715172940951653e-06,
+      "loss": 1.3018,
+      "step": 3031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.715077393464552e-06,
+      "loss": 1.4716,
+      "step": 3032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714981845977451e-06,
+      "loss": 1.48,
+      "step": 3033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7148862984903495e-06,
+      "loss": 1.2253,
+      "step": 3034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714790751003249e-06,
+      "loss": 1.5983,
+      "step": 3035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7146952035161484e-06,
+      "loss": 1.2537,
+      "step": 3036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714599656029047e-06,
+      "loss": 1.7025,
+      "step": 3037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714504108541946e-06,
+      "loss": 1.7101,
+      "step": 3038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714408561054845e-06,
+      "loss": 1.5254,
+      "step": 3039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714313013567744e-06,
+      "loss": 1.0957,
+      "step": 3040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714217466080642e-06,
+      "loss": 1.4023,
+      "step": 3041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714121918593542e-06,
+      "loss": 1.4029,
+      "step": 3042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.714026371106441e-06,
+      "loss": 1.4725,
+      "step": 3043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713930823619339e-06,
+      "loss": 1.5316,
+      "step": 3044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713835276132238e-06,
+      "loss": 1.5856,
+      "step": 3045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713739728645137e-06,
+      "loss": 1.0829,
+      "step": 3046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713644181158036e-06,
+      "loss": 1.471,
+      "step": 3047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713548633670934e-06,
+      "loss": 1.5028,
+      "step": 3048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.713453086183834e-06,
+      "loss": 1.5375,
+      "step": 3049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7133575386967324e-06,
+      "loss": 1.7614,
+      "step": 3050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7132619912096315e-06,
+      "loss": 1.9438,
+      "step": 3051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7131664437225305e-06,
+      "loss": 1.8517,
+      "step": 3052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7130708962354295e-06,
+      "loss": 1.5436,
+      "step": 3053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712975348748329e-06,
+      "loss": 1.5805,
+      "step": 3054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712879801261227e-06,
+      "loss": 1.3471,
+      "step": 3055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712784253774126e-06,
+      "loss": 2.092,
+      "step": 3056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712688706287025e-06,
+      "loss": 1.1753,
+      "step": 3057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712593158799924e-06,
+      "loss": 1.7523,
+      "step": 3058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712497611312823e-06,
+      "loss": 1.079,
+      "step": 3059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712402063825722e-06,
+      "loss": 1.3142,
+      "step": 3060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.71230651633862e-06,
+      "loss": 1.3327,
+      "step": 3061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712210968851519e-06,
+      "loss": 1.3669,
+      "step": 3062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712115421364418e-06,
+      "loss": 1.8453,
+      "step": 3063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712019873877317e-06,
+      "loss": 1.8463,
+      "step": 3064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711924326390216e-06,
+      "loss": 1.5759,
+      "step": 3065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711828778903115e-06,
+      "loss": 2.0643,
+      "step": 3066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711733231416014e-06,
+      "loss": 1.7403,
+      "step": 3067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711637683928913e-06,
+      "loss": 1.5805,
+      "step": 3068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711542136441812e-06,
+      "loss": 1.4158,
+      "step": 3069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711446588954711e-06,
+      "loss": 1.554,
+      "step": 3070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.71135104146761e-06,
+      "loss": 1.4115,
+      "step": 3071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711255493980509e-06,
+      "loss": 1.466,
+      "step": 3072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711159946493408e-06,
+      "loss": 1.2954,
+      "step": 3073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.711064399006307e-06,
+      "loss": 1.7619,
+      "step": 3074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710968851519205e-06,
+      "loss": 2.1291,
+      "step": 3075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710873304032104e-06,
+      "loss": 1.3071,
+      "step": 3076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710777756545003e-06,
+      "loss": 1.2673,
+      "step": 3077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710682209057902e-06,
+      "loss": 1.7451,
+      "step": 3078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710586661570801e-06,
+      "loss": 1.2283,
+      "step": 3079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7104911140837e-06,
+      "loss": 1.3005,
+      "step": 3080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.710395566596599e-06,
+      "loss": 1.2847,
+      "step": 3081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7103000191094974e-06,
+      "loss": 1.5803,
+      "step": 3082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7102044716223965e-06,
+      "loss": 1.3881,
+      "step": 3083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7101089241352955e-06,
+      "loss": 1.2239,
+      "step": 3084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7100133766481945e-06,
+      "loss": 1.9557,
+      "step": 3085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709917829161094e-06,
+      "loss": 1.4068,
+      "step": 3086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709822281673993e-06,
+      "loss": 1.5206,
+      "step": 3087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709726734186892e-06,
+      "loss": 1.5131,
+      "step": 3088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.70963118669979e-06,
+      "loss": 1.1753,
+      "step": 3089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709535639212689e-06,
+      "loss": 1.9118,
+      "step": 3090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709440091725588e-06,
+      "loss": 1.3397,
+      "step": 3091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709344544238487e-06,
+      "loss": 1.6168,
+      "step": 3092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709248996751386e-06,
+      "loss": 1.6782,
+      "step": 3093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709153449264285e-06,
+      "loss": 1.5842,
+      "step": 3094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.709057901777184e-06,
+      "loss": 1.7034,
+      "step": 3095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708962354290082e-06,
+      "loss": 1.7938,
+      "step": 3096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708866806802981e-06,
+      "loss": 1.6205,
+      "step": 3097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.70877125931588e-06,
+      "loss": 1.3822,
+      "step": 3098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708675711828779e-06,
+      "loss": 1.6966,
+      "step": 3099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7085801643416784e-06,
+      "loss": 1.2781,
+      "step": 3100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7084846168545775e-06,
+      "loss": 1.1901,
+      "step": 3101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7083890693674765e-06,
+      "loss": 1.6864,
+      "step": 3102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708293521880375e-06,
+      "loss": 1.2305,
+      "step": 3103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708197974393274e-06,
+      "loss": 1.6301,
+      "step": 3104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708102426906173e-06,
+      "loss": 1.7023,
+      "step": 3105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.708006879419072e-06,
+      "loss": 2.0813,
+      "step": 3106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707911331931971e-06,
+      "loss": 1.4931,
+      "step": 3107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.70781578444487e-06,
+      "loss": 1.1436,
+      "step": 3108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707720236957768e-06,
+      "loss": 1.2116,
+      "step": 3109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707624689470667e-06,
+      "loss": 1.8993,
+      "step": 3110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707529141983566e-06,
+      "loss": 1.5456,
+      "step": 3111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707433594496465e-06,
+      "loss": 1.7715,
+      "step": 3112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707338047009363e-06,
+      "loss": 1.5207,
+      "step": 3113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707242499522263e-06,
+      "loss": 1.3221,
+      "step": 3114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707146952035162e-06,
+      "loss": 2.1641,
+      "step": 3115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7070514045480605e-06,
+      "loss": 1.5002,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7069558570609595e-06,
+      "loss": 1.8523,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706860309573859e-06,
+      "loss": 1.3801,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706764762086758e-06,
+      "loss": 1.5087,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706669214599656e-06,
+      "loss": 1.4292,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706573667112556e-06,
+      "loss": 1.4603,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706478119625455e-06,
+      "loss": 1.631,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706382572138353e-06,
+      "loss": 1.6816,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706287024651252e-06,
+      "loss": 1.2627,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706191477164151e-06,
+      "loss": 1.0339,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70609592967705e-06,
+      "loss": 1.8458,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.706000382189948e-06,
+      "loss": 1.6737,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705904834702847e-06,
+      "loss": 1.6411,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705809287215747e-06,
+      "loss": 1.6742,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705713739728645e-06,
+      "loss": 1.4072,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705618192241544e-06,
+      "loss": 1.3777,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7055226447544434e-06,
+      "loss": 1.3223,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7054270972673425e-06,
+      "loss": 1.9392,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705331549780241e-06,
+      "loss": 1.4715,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70523600229314e-06,
+      "loss": 1.4381,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70514045480604e-06,
+      "loss": 1.3265,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.705044907318938e-06,
+      "loss": 0.8509,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704949359831837e-06,
+      "loss": 1.6227,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704853812344736e-06,
+      "loss": 1.4869,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704758264857635e-06,
+      "loss": 1.4873,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704662717370533e-06,
+      "loss": 1.5935,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704567169883432e-06,
+      "loss": 1.2054,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704471622396332e-06,
+      "loss": 0.947,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70437607490923e-06,
+      "loss": 1.2135,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704280527422129e-06,
+      "loss": 1.7985,
+      "step": 3145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704184979935028e-06,
+      "loss": 1.456,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.704089432447927e-06,
+      "loss": 1.8417,
+      "step": 3147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7039938849608255e-06,
+      "loss": 1.2257,
+      "step": 3148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7038983374737245e-06,
+      "loss": 1.4951,
+      "step": 3149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703802789986624e-06,
+      "loss": 1.2493,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703707242499523e-06,
+      "loss": 1.5452,
+      "step": 3151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703611695012422e-06,
+      "loss": 1.5937,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703516147525321e-06,
+      "loss": 1.5889,
+      "step": 3153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70342060003822e-06,
+      "loss": 0.9714,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703325052551118e-06,
+      "loss": 2.1084,
+      "step": 3155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703229505064017e-06,
+      "loss": 1.4045,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703133957576916e-06,
+      "loss": 1.4678,
+      "step": 3157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.703038410089815e-06,
+      "loss": 1.4773,
+      "step": 3158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702942862602714e-06,
+      "loss": 0.9857,
+      "step": 3159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702847315115613e-06,
+      "loss": 1.8989,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702751767628511e-06,
+      "loss": 1.7945,
+      "step": 3161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70265622014141e-06,
+      "loss": 1.661,
+      "step": 3162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702560672654309e-06,
+      "loss": 1.4769,
+      "step": 3163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7024651251672084e-06,
+      "loss": 1.5428,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7023695776801075e-06,
+      "loss": 1.4309,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7022740301930065e-06,
+      "loss": 2.13,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7021784827059055e-06,
+      "loss": 1.7075,
+      "step": 3167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.702082935218804e-06,
+      "loss": 1.2801,
+      "step": 3168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701987387731703e-06,
+      "loss": 1.5576,
+      "step": 3169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701891840244602e-06,
+      "loss": 1.0626,
+      "step": 3170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701796292757501e-06,
+      "loss": 1.5943,
+      "step": 3171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7017007452704e-06,
+      "loss": 1.5936,
+      "step": 3172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701605197783299e-06,
+      "loss": 1.6047,
+      "step": 3173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701509650296198e-06,
+      "loss": 1.4613,
+      "step": 3174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701414102809096e-06,
+      "loss": 1.1455,
+      "step": 3175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701318555321995e-06,
+      "loss": 1.4191,
+      "step": 3176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701223007834894e-06,
+      "loss": 1.6232,
+      "step": 3177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701127460347793e-06,
+      "loss": 1.2456,
+      "step": 3178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701031912860692e-06,
+      "loss": 1.2358,
+      "step": 3179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700936365373591e-06,
+      "loss": 1.5774,
+      "step": 3180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.70084081788649e-06,
+      "loss": 1.1409,
+      "step": 3181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700745270399389e-06,
+      "loss": 1.4455,
+      "step": 3182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700649722912288e-06,
+      "loss": 1.9759,
+      "step": 3183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700554175425187e-06,
+      "loss": 1.4237,
+      "step": 3184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700458627938086e-06,
+      "loss": 1.7224,
+      "step": 3185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700363080450985e-06,
+      "loss": 1.4076,
+      "step": 3186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700267532963884e-06,
+      "loss": 1.3821,
+      "step": 3187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700171985476783e-06,
+      "loss": 1.688,
+      "step": 3188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.700076437989681e-06,
+      "loss": 1.563,
+      "step": 3189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.69998089050258e-06,
+      "loss": 1.371,
+      "step": 3190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699885343015479e-06,
+      "loss": 1.7775,
+      "step": 3191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699789795528378e-06,
+      "loss": 1.3535,
+      "step": 3192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699694248041277e-06,
+      "loss": 2.2019,
+      "step": 3193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699598700554176e-06,
+      "loss": 1.6473,
+      "step": 3194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699503153067075e-06,
+      "loss": 1.7507,
+      "step": 3195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6994076055799734e-06,
+      "loss": 1.6631,
+      "step": 3196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6993120580928725e-06,
+      "loss": 1.33,
+      "step": 3197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6992165106057715e-06,
+      "loss": 1.2235,
+      "step": 3198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6991209631186705e-06,
+      "loss": 1.5224,
+      "step": 3199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.699025415631569e-06,
+      "loss": 1.6631,
+      "step": 3200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698929868144469e-06,
+      "loss": 1.4317,
+      "step": 3201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698834320657367e-06,
+      "loss": 1.5302,
+      "step": 3202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698738773170266e-06,
+      "loss": 1.3491,
+      "step": 3203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698643225683165e-06,
+      "loss": 1.4143,
+      "step": 3204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698547678196064e-06,
+      "loss": 1.3401,
+      "step": 3205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698452130708963e-06,
+      "loss": 1.4358,
+      "step": 3206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698356583221861e-06,
+      "loss": 1.6298,
+      "step": 3207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698261035734761e-06,
+      "loss": 1.6081,
+      "step": 3208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698165488247659e-06,
+      "loss": 1.5412,
+      "step": 3209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.698069940760558e-06,
+      "loss": 1.5243,
+      "step": 3210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697974393273457e-06,
+      "loss": 1.7212,
+      "step": 3211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697878845786356e-06,
+      "loss": 1.4801,
+      "step": 3212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6977832982992545e-06,
+      "loss": 2.0148,
+      "step": 3213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697687750812154e-06,
+      "loss": 1.8097,
+      "step": 3214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6975922033250535e-06,
+      "loss": 1.7223,
+      "step": 3215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697496655837952e-06,
+      "loss": 1.6954,
+      "step": 3216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697401108350851e-06,
+      "loss": 1.6246,
+      "step": 3217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.69730556086375e-06,
+      "loss": 1.9579,
+      "step": 3218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697210013376649e-06,
+      "loss": 1.6462,
+      "step": 3219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697114465889547e-06,
+      "loss": 1.2862,
+      "step": 3220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.697018918402446e-06,
+      "loss": 1.3031,
+      "step": 3221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696923370915346e-06,
+      "loss": 1.7852,
+      "step": 3222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696827823428244e-06,
+      "loss": 1.4479,
+      "step": 3223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696732275941143e-06,
+      "loss": 0.9887,
+      "step": 3224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696636728454042e-06,
+      "loss": 1.4765,
+      "step": 3225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696541180966941e-06,
+      "loss": 1.908,
+      "step": 3226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696445633479839e-06,
+      "loss": 1.8241,
+      "step": 3227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696350085992738e-06,
+      "loss": 1.4128,
+      "step": 3228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696254538505638e-06,
+      "loss": 1.1212,
+      "step": 3229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6961589910185365e-06,
+      "loss": 1.4198,
+      "step": 3230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6960634435314355e-06,
+      "loss": 1.1648,
+      "step": 3231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695967896044335e-06,
+      "loss": 1.8835,
+      "step": 3232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695872348557234e-06,
+      "loss": 1.506,
+      "step": 3233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695776801070132e-06,
+      "loss": 1.7824,
+      "step": 3234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695681253583031e-06,
+      "loss": 1.4247,
+      "step": 3235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69558570609593e-06,
+      "loss": 1.077,
+      "step": 3236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695490158608829e-06,
+      "loss": 1.071,
+      "step": 3237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695394611121728e-06,
+      "loss": 1.495,
+      "step": 3238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695299063634627e-06,
+      "loss": 1.867,
+      "step": 3239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695203516147526e-06,
+      "loss": 1.1827,
+      "step": 3240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695107968660424e-06,
+      "loss": 1.1223,
+      "step": 3241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695012421173323e-06,
+      "loss": 1.4067,
+      "step": 3242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694916873686222e-06,
+      "loss": 1.5423,
+      "step": 3243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694821326199121e-06,
+      "loss": 1.6503,
+      "step": 3244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69472577871202e-06,
+      "loss": 1.8304,
+      "step": 3245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6946302312249194e-06,
+      "loss": 1.0644,
+      "step": 3246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6945346837378185e-06,
+      "loss": 1.9622,
+      "step": 3247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694439136250717e-06,
+      "loss": 1.5149,
+      "step": 3248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694343588763616e-06,
+      "loss": 1.51,
+      "step": 3249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694248041276515e-06,
+      "loss": 1.4793,
+      "step": 3250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694152493789414e-06,
+      "loss": 1.1855,
+      "step": 3251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694056946302313e-06,
+      "loss": 1.5449,
+      "step": 3252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693961398815212e-06,
+      "loss": 1.8549,
+      "step": 3253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69386585132811e-06,
+      "loss": 1.6476,
+      "step": 3254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693770303841009e-06,
+      "loss": 1.6854,
+      "step": 3255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693674756353908e-06,
+      "loss": 1.7713,
+      "step": 3256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693579208866807e-06,
+      "loss": 1.2006,
+      "step": 3257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693483661379706e-06,
+      "loss": 2.0333,
+      "step": 3258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693388113892605e-06,
+      "loss": 1.7034,
+      "step": 3259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693292566405504e-06,
+      "loss": 1.4843,
+      "step": 3260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6931970189184025e-06,
+      "loss": 1.1612,
+      "step": 3261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6931014714313015e-06,
+      "loss": 1.5096,
+      "step": 3262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6930059239442005e-06,
+      "loss": 1.7518,
+      "step": 3263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6929103764571e-06,
+      "loss": 1.3352,
+      "step": 3264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692814828969999e-06,
+      "loss": 1.6783,
+      "step": 3265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692719281482898e-06,
+      "loss": 1.6975,
+      "step": 3266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692623733995797e-06,
+      "loss": 1.5027,
+      "step": 3267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692528186508695e-06,
+      "loss": 1.5723,
+      "step": 3268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692432639021594e-06,
+      "loss": 1.3488,
+      "step": 3269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692337091534493e-06,
+      "loss": 1.2127,
+      "step": 3270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692241544047392e-06,
+      "loss": 1.6921,
+      "step": 3271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69214599656029e-06,
+      "loss": 1.5995,
+      "step": 3272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69205044907319e-06,
+      "loss": 1.1397,
+      "step": 3273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691954901586089e-06,
+      "loss": 1.3266,
+      "step": 3274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691859354098987e-06,
+      "loss": 1.4131,
+      "step": 3275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691763806611886e-06,
+      "loss": 1.6256,
+      "step": 3276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691668259124785e-06,
+      "loss": 1.3429,
+      "step": 3277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691572711637684e-06,
+      "loss": 1.8701,
+      "step": 3278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691477164150583e-06,
+      "loss": 1.6851,
+      "step": 3279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6913816166634825e-06,
+      "loss": 1.7166,
+      "step": 3280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6912860691763815e-06,
+      "loss": 1.5557,
+      "step": 3281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.69119052168928e-06,
+      "loss": 1.6349,
+      "step": 3282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.691094974202179e-06,
+      "loss": 1.4587,
+      "step": 3283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690999426715078e-06,
+      "loss": 1.6253,
+      "step": 3284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690903879227977e-06,
+      "loss": 1.2967,
+      "step": 3285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690808331740875e-06,
+      "loss": 1.48,
+      "step": 3286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690712784253775e-06,
+      "loss": 1.3237,
+      "step": 3287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690617236766674e-06,
+      "loss": 1.5986,
+      "step": 3288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690521689279572e-06,
+      "loss": 1.9962,
+      "step": 3289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690426141792471e-06,
+      "loss": 1.6847,
+      "step": 3290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.69033059430537e-06,
+      "loss": 1.2609,
+      "step": 3291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690235046818269e-06,
+      "loss": 1.4409,
+      "step": 3292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6901394993311675e-06,
+      "loss": 1.5087,
+      "step": 3293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.690043951844067e-06,
+      "loss": 1.6482,
+      "step": 3294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689948404356966e-06,
+      "loss": 1.9422,
+      "step": 3295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6898528568698646e-06,
+      "loss": 2.0758,
+      "step": 3296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689757309382764e-06,
+      "loss": 1.2696,
+      "step": 3297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689661761895663e-06,
+      "loss": 1.4518,
+      "step": 3298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689566214408562e-06,
+      "loss": 1.404,
+      "step": 3299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68947066692146e-06,
+      "loss": 1.615,
+      "step": 3300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68937511943436e-06,
+      "loss": 2.0244,
+      "step": 3301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689279571947258e-06,
+      "loss": 1.4573,
+      "step": 3302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689184024460157e-06,
+      "loss": 1.1433,
+      "step": 3303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.689088476973056e-06,
+      "loss": 1.4632,
+      "step": 3304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688992929485955e-06,
+      "loss": 1.2758,
+      "step": 3305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688897381998854e-06,
+      "loss": 1.1027,
+      "step": 3306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688801834511752e-06,
+      "loss": 1.4451,
+      "step": 3307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688706287024651e-06,
+      "loss": 1.6328,
+      "step": 3308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68861073953755e-06,
+      "loss": 1.6812,
+      "step": 3309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688515192050449e-06,
+      "loss": 1.6691,
+      "step": 3310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6884196445633485e-06,
+      "loss": 1.6135,
+      "step": 3311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6883240970762475e-06,
+      "loss": 1.8029,
+      "step": 3312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688228549589146e-06,
+      "loss": 1.5889,
+      "step": 3313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688133002102045e-06,
+      "loss": 2.1229,
+      "step": 3314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.688037454614944e-06,
+      "loss": 1.2082,
+      "step": 3315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687941907127843e-06,
+      "loss": 1.3167,
+      "step": 3316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687846359640742e-06,
+      "loss": 1.1611,
+      "step": 3317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687750812153641e-06,
+      "loss": 1.7395,
+      "step": 3318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68765526466654e-06,
+      "loss": 0.9401,
+      "step": 3319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687559717179438e-06,
+      "loss": 1.8095,
+      "step": 3320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687464169692337e-06,
+      "loss": 1.7545,
+      "step": 3321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687368622205236e-06,
+      "loss": 1.4432,
+      "step": 3322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687273074718135e-06,
+      "loss": 1.2105,
+      "step": 3323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687177527231034e-06,
+      "loss": 1.7402,
+      "step": 3324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687081979743933e-06,
+      "loss": 1.7389,
+      "step": 3325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686986432256832e-06,
+      "loss": 1.1864,
+      "step": 3326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6868908847697305e-06,
+      "loss": 1.5546,
+      "step": 3327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6867953372826296e-06,
+      "loss": 1.3419,
+      "step": 3328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686699789795529e-06,
+      "loss": 1.6048,
+      "step": 3329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686604242308428e-06,
+      "loss": 1.1443,
+      "step": 3330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686508694821327e-06,
+      "loss": 1.4613,
+      "step": 3331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686413147334226e-06,
+      "loss": 1.442,
+      "step": 3332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686317599847125e-06,
+      "loss": 1.9301,
+      "step": 3333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686222052360023e-06,
+      "loss": 1.4789,
+      "step": 3334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686126504872922e-06,
+      "loss": 1.6039,
+      "step": 3335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.686030957385821e-06,
+      "loss": 1.4697,
+      "step": 3336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.68593540989872e-06,
+      "loss": 1.4201,
+      "step": 3337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.685839862411619e-06,
+      "loss": 1.8726,
+      "step": 3338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.685744314924518e-06,
+      "loss": 1.1667,
+      "step": 3339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.685648767437417e-06,
+      "loss": 1.4498,
+      "step": 3340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.685553219950315e-06,
+      "loss": 1.3925,
+      "step": 3341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.685457672463214e-06,
+      "loss": 1.4978,
+      "step": 3342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6853621249761135e-06,
+      "loss": 1.3945,
+      "step": 3343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6852665774890125e-06,
+      "loss": 1.2838,
+      "step": 3344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6851710300019115e-06,
+      "loss": 1.2138,
+      "step": 3345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6850754825148106e-06,
+      "loss": 1.7484,
+      "step": 3346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.68497993502771e-06,
+      "loss": 1.8711,
+      "step": 3347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684884387540608e-06,
+      "loss": 1.6029,
+      "step": 3348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684788840053507e-06,
+      "loss": 1.4672,
+      "step": 3349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684693292566406e-06,
+      "loss": 1.4612,
+      "step": 3350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684597745079305e-06,
+      "loss": 1.3598,
+      "step": 3351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684502197592204e-06,
+      "loss": 1.5237,
+      "step": 3352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684406650105103e-06,
+      "loss": 1.1441,
+      "step": 3353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684311102618001e-06,
+      "loss": 1.6378,
+      "step": 3354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6842155551309e-06,
+      "loss": 1.2001,
+      "step": 3355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684120007643799e-06,
+      "loss": 2.0103,
+      "step": 3356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.684024460156698e-06,
+      "loss": 1.3604,
+      "step": 3357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683928912669597e-06,
+      "loss": 1.2757,
+      "step": 3358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683833365182496e-06,
+      "loss": 1.7025,
+      "step": 3359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683737817695395e-06,
+      "loss": 1.356,
+      "step": 3360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683642270208294e-06,
+      "loss": 1.5918,
+      "step": 3361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683546722721193e-06,
+      "loss": 1.1681,
+      "step": 3362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683451175234092e-06,
+      "loss": 2.1509,
+      "step": 3363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683355627746991e-06,
+      "loss": 1.9013,
+      "step": 3364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683260080259889e-06,
+      "loss": 1.5894,
+      "step": 3365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683164532772789e-06,
+      "loss": 2.0335,
+      "step": 3366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.683068985285688e-06,
+      "loss": 1.339,
+      "step": 3367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682973437798586e-06,
+      "loss": 1.4688,
+      "step": 3368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682877890311485e-06,
+      "loss": 1.4428,
+      "step": 3369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682782342824384e-06,
+      "loss": 1.1584,
+      "step": 3370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682686795337283e-06,
+      "loss": 1.5325,
+      "step": 3371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682591247850181e-06,
+      "loss": 1.3918,
+      "step": 3372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682495700363081e-06,
+      "loss": 1.3731,
+      "step": 3373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.68240015287598e-06,
+      "loss": 1.2704,
+      "step": 3374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6823046053888785e-06,
+      "loss": 1.8239,
+      "step": 3375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6822090579017775e-06,
+      "loss": 1.3452,
+      "step": 3376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6821135104146765e-06,
+      "loss": 1.5531,
+      "step": 3377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6820179629275756e-06,
+      "loss": 1.8551,
+      "step": 3378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681922415440474e-06,
+      "loss": 1.5004,
+      "step": 3379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681826867953373e-06,
+      "loss": 1.6468,
+      "step": 3380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681731320466273e-06,
+      "loss": 1.5373,
+      "step": 3381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681635772979171e-06,
+      "loss": 1.3225,
+      "step": 3382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.68154022549207e-06,
+      "loss": 1.4634,
+      "step": 3383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681444678004969e-06,
+      "loss": 1.2402,
+      "step": 3384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681349130517868e-06,
+      "loss": 1.5527,
+      "step": 3385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681253583030766e-06,
+      "loss": 1.287,
+      "step": 3386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681158035543665e-06,
+      "loss": 1.7412,
+      "step": 3387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.681062488056565e-06,
+      "loss": 1.3601,
+      "step": 3388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680966940569463e-06,
+      "loss": 1.5902,
+      "step": 3389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680871393082362e-06,
+      "loss": 1.4273,
+      "step": 3390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680775845595261e-06,
+      "loss": 1.782,
+      "step": 3391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.68068029810816e-06,
+      "loss": 1.4591,
+      "step": 3392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680584750621059e-06,
+      "loss": 1.8313,
+      "step": 3393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680489203133958e-06,
+      "loss": 1.1913,
+      "step": 3394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680393655646857e-06,
+      "loss": 1.308,
+      "step": 3395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680298108159756e-06,
+      "loss": 1.7355,
+      "step": 3396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680202560672655e-06,
+      "loss": 1.5827,
+      "step": 3397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680107013185554e-06,
+      "loss": 1.7767,
+      "step": 3398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.680011465698453e-06,
+      "loss": 1.5882,
+      "step": 3399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679915918211351e-06,
+      "loss": 1.4654,
+      "step": 3400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.67982037072425e-06,
+      "loss": 1.3826,
+      "step": 3401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679724823237149e-06,
+      "loss": 1.295,
+      "step": 3402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679629275750048e-06,
+      "loss": 1.6931,
+      "step": 3403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679533728262947e-06,
+      "loss": 1.0347,
+      "step": 3404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679438180775846e-06,
+      "loss": 1.4508,
+      "step": 3405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.679342633288744e-06,
+      "loss": 1.6375,
+      "step": 3406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6792470858016435e-06,
+      "loss": 1.3561,
+      "step": 3407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6791515383145425e-06,
+      "loss": 1.3158,
+      "step": 3408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6790559908274415e-06,
+      "loss": 1.3592,
+      "step": 3409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6789604433403406e-06,
+      "loss": 1.386,
+      "step": 3410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.67886489585324e-06,
+      "loss": 2.3236,
+      "step": 3411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678769348366139e-06,
+      "loss": 1.5073,
+      "step": 3412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678673800879037e-06,
+      "loss": 1.8402,
+      "step": 3413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678578253391936e-06,
+      "loss": 1.3071,
+      "step": 3414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678482705904835e-06,
+      "loss": 1.1032,
+      "step": 3415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678387158417734e-06,
+      "loss": 1.234,
+      "step": 3416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678291610930633e-06,
+      "loss": 1.364,
+      "step": 3417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678196063443532e-06,
+      "loss": 1.3309,
+      "step": 3418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678100515956431e-06,
+      "loss": 1.3974,
+      "step": 3419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.678004968469329e-06,
+      "loss": 1.5252,
+      "step": 3420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677909420982228e-06,
+      "loss": 1.9145,
+      "step": 3421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677813873495127e-06,
+      "loss": 1.7646,
+      "step": 3422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677718326008026e-06,
+      "loss": 1.3694,
+      "step": 3423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677622778520925e-06,
+      "loss": 1.7949,
+      "step": 3424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6775272310338245e-06,
+      "loss": 1.7233,
+      "step": 3425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6774316835467235e-06,
+      "loss": 1.6687,
+      "step": 3426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677336136059622e-06,
+      "loss": 1.498,
+      "step": 3427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677240588572521e-06,
+      "loss": 1.3427,
+      "step": 3428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.67714504108542e-06,
+      "loss": 1.3887,
+      "step": 3429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.677049493598319e-06,
+      "loss": 1.4053,
+      "step": 3430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676953946111218e-06,
+      "loss": 1.2125,
+      "step": 3431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676858398624117e-06,
+      "loss": 1.5658,
+      "step": 3432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676762851137016e-06,
+      "loss": 1.4216,
+      "step": 3433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676667303649914e-06,
+      "loss": 1.4215,
+      "step": 3434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676571756162813e-06,
+      "loss": 1.1426,
+      "step": 3435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676476208675712e-06,
+      "loss": 1.3941,
+      "step": 3436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676380661188611e-06,
+      "loss": 1.3367,
+      "step": 3437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.67628511370151e-06,
+      "loss": 1.5823,
+      "step": 3438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676189566214409e-06,
+      "loss": 1.5282,
+      "step": 3439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.676094018727308e-06,
+      "loss": 1.5829,
+      "step": 3440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6759984712402065e-06,
+      "loss": 1.09,
+      "step": 3441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6759029237531056e-06,
+      "loss": 1.6899,
+      "step": 3442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675807376266005e-06,
+      "loss": 1.0793,
+      "step": 3443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675711828778904e-06,
+      "loss": 0.7979,
+      "step": 3444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675616281291803e-06,
+      "loss": 1.1733,
+      "step": 3445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675520733804702e-06,
+      "loss": 1.2035,
+      "step": 3446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6754251863176e-06,
+      "loss": 1.5873,
+      "step": 3447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675329638830499e-06,
+      "loss": 1.6164,
+      "step": 3448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675234091343398e-06,
+      "loss": 1.0425,
+      "step": 3449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675138543856297e-06,
+      "loss": 1.3732,
+      "step": 3450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.675042996369196e-06,
+      "loss": 1.1774,
+      "step": 3451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674947448882094e-06,
+      "loss": 1.5269,
+      "step": 3452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674851901394994e-06,
+      "loss": 1.3034,
+      "step": 3453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674756353907892e-06,
+      "loss": 1.4402,
+      "step": 3454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674660806420791e-06,
+      "loss": 2.0329,
+      "step": 3455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.67456525893369e-06,
+      "loss": 1.4545,
+      "step": 3456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6744697114465895e-06,
+      "loss": 1.0642,
+      "step": 3457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674374163959488e-06,
+      "loss": 1.5821,
+      "step": 3458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674278616472387e-06,
+      "loss": 1.6167,
+      "step": 3459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6741830689852866e-06,
+      "loss": 1.4451,
+      "step": 3460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.674087521498185e-06,
+      "loss": 1.8411,
+      "step": 3461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673991974011084e-06,
+      "loss": 1.7095,
+      "step": 3462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673896426523983e-06,
+      "loss": 1.287,
+      "step": 3463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673800879036882e-06,
+      "loss": 1.4348,
+      "step": 3464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.67370533154978e-06,
+      "loss": 1.3827,
+      "step": 3465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673609784062679e-06,
+      "loss": 1.6988,
+      "step": 3466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673514236575579e-06,
+      "loss": 1.7253,
+      "step": 3467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673418689088477e-06,
+      "loss": 1.3904,
+      "step": 3468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673323141601376e-06,
+      "loss": 1.6178,
+      "step": 3469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673227594114275e-06,
+      "loss": 1.5808,
+      "step": 3470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.673132046627174e-06,
+      "loss": 1.0223,
+      "step": 3471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6730364991400725e-06,
+      "loss": 1.1368,
+      "step": 3472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6729409516529715e-06,
+      "loss": 1.4943,
+      "step": 3473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672845404165871e-06,
+      "loss": 1.5477,
+      "step": 3474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.67274985667877e-06,
+      "loss": 1.6341,
+      "step": 3475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672654309191669e-06,
+      "loss": 1.7942,
+      "step": 3476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672558761704568e-06,
+      "loss": 1.7228,
+      "step": 3477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672463214217467e-06,
+      "loss": 1.2873,
+      "step": 3478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672367666730365e-06,
+      "loss": 1.7697,
+      "step": 3479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672272119243264e-06,
+      "loss": 1.6356,
+      "step": 3480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672176571756164e-06,
+      "loss": 1.2868,
+      "step": 3481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672081024269062e-06,
+      "loss": 1.3747,
+      "step": 3482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.671985476781961e-06,
+      "loss": 1.7477,
+      "step": 3483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67188992929486e-06,
+      "loss": 1.0804,
+      "step": 3484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.671794381807759e-06,
+      "loss": 1.492,
+      "step": 3485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.671698834320657e-06,
+      "loss": 1.4942,
+      "step": 3486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.671603286833556e-06,
+      "loss": 1.931,
+      "step": 3487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.671507739346455e-06,
+      "loss": 1.0215,
+      "step": 3488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6714121918593545e-06,
+      "loss": 1.7281,
+      "step": 3489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6713166443722535e-06,
+      "loss": 1.3001,
+      "step": 3490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6712210968851525e-06,
+      "loss": 1.1164,
+      "step": 3491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6711255493980516e-06,
+      "loss": 1.8432,
+      "step": 3492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67103000191095e-06,
+      "loss": 1.9822,
+      "step": 3493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670934454423849e-06,
+      "loss": 1.5281,
+      "step": 3494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670838906936748e-06,
+      "loss": 1.2416,
+      "step": 3495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670743359449647e-06,
+      "loss": 1.6789,
+      "step": 3496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670647811962546e-06,
+      "loss": 1.5517,
+      "step": 3497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670552264475445e-06,
+      "loss": 1.5588,
+      "step": 3498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670456716988344e-06,
+      "loss": 1.6377,
+      "step": 3499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670361169501242e-06,
+      "loss": 1.7011,
+      "step": 3500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670265622014141e-06,
+      "loss": 1.2566,
+      "step": 3501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.67017007452704e-06,
+      "loss": 1.3453,
+      "step": 3502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.670074527039939e-06,
+      "loss": 1.4617,
+      "step": 3503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669978979552838e-06,
+      "loss": 1.7366,
+      "step": 3504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669883432065737e-06,
+      "loss": 1.8009,
+      "step": 3505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6697878845786356e-06,
+      "loss": 1.4132,
+      "step": 3506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669692337091535e-06,
+      "loss": 1.6207,
+      "step": 3507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669596789604434e-06,
+      "loss": 1.4016,
+      "step": 3508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669501242117333e-06,
+      "loss": 0.8832,
+      "step": 3509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669405694630232e-06,
+      "loss": 1.4843,
+      "step": 3510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669310147143131e-06,
+      "loss": 1.529,
+      "step": 3511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.66921459965603e-06,
+      "loss": 1.7599,
+      "step": 3512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669119052168928e-06,
+      "loss": 1.9328,
+      "step": 3513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.669023504681827e-06,
+      "loss": 1.5124,
+      "step": 3514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668927957194726e-06,
+      "loss": 1.3174,
+      "step": 3515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668832409707625e-06,
+      "loss": 1.3711,
+      "step": 3516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668736862220524e-06,
+      "loss": 1.6633,
+      "step": 3517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668641314733423e-06,
+      "loss": 1.741,
+      "step": 3518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668545767246322e-06,
+      "loss": 1.7309,
+      "step": 3519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.66845021975922e-06,
+      "loss": 1.2367,
+      "step": 3520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6683546722721195e-06,
+      "loss": 1.3338,
+      "step": 3521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6682591247850185e-06,
+      "loss": 1.8008,
+      "step": 3522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6681635772979175e-06,
+      "loss": 1.3517,
+      "step": 3523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.668068029810816e-06,
+      "loss": 1.6558,
+      "step": 3524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667972482323716e-06,
+      "loss": 1.4525,
+      "step": 3525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667876934836615e-06,
+      "loss": 1.105,
+      "step": 3526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667781387349513e-06,
+      "loss": 1.2895,
+      "step": 3527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667685839862412e-06,
+      "loss": 1.7612,
+      "step": 3528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667590292375311e-06,
+      "loss": 1.1522,
+      "step": 3529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.66749474488821e-06,
+      "loss": 1.0778,
+      "step": 3530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667399197401108e-06,
+      "loss": 1.6989,
+      "step": 3531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667303649914008e-06,
+      "loss": 1.4162,
+      "step": 3532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667208102426907e-06,
+      "loss": 1.3856,
+      "step": 3533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667112554939805e-06,
+      "loss": 1.3334,
+      "step": 3534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.667017007452704e-06,
+      "loss": 1.3672,
+      "step": 3535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666921459965603e-06,
+      "loss": 1.5996,
+      "step": 3536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666825912478502e-06,
+      "loss": 1.2453,
+      "step": 3537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6667303649914006e-06,
+      "loss": 0.9655,
+      "step": 3538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6666348175043005e-06,
+      "loss": 1.173,
+      "step": 3539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6665392700171995e-06,
+      "loss": 1.7071,
+      "step": 3540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666443722530098e-06,
+      "loss": 1.2085,
+      "step": 3541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666348175042997e-06,
+      "loss": 1.5536,
+      "step": 3542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666252627555896e-06,
+      "loss": 1.6199,
+      "step": 3543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666157080068795e-06,
+      "loss": 1.7318,
+      "step": 3544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666061532581693e-06,
+      "loss": 1.1261,
+      "step": 3545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665965985094593e-06,
+      "loss": 1.1269,
+      "step": 3546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665870437607491e-06,
+      "loss": 1.4488,
+      "step": 3547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66577489012039e-06,
+      "loss": 1.2195,
+      "step": 3548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665679342633289e-06,
+      "loss": 1.626,
+      "step": 3549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665583795146188e-06,
+      "loss": 1.5232,
+      "step": 3550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665488247659087e-06,
+      "loss": 1.3712,
+      "step": 3551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665392700171985e-06,
+      "loss": 1.9668,
+      "step": 3552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665297152684885e-06,
+      "loss": 1.5865,
+      "step": 3553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6652016051977835e-06,
+      "loss": 1.3662,
+      "step": 3554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6651060577106825e-06,
+      "loss": 1.6455,
+      "step": 3555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6650105102235816e-06,
+      "loss": 1.2837,
+      "step": 3556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664914962736481e-06,
+      "loss": 1.4874,
+      "step": 3557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664819415249379e-06,
+      "loss": 1.3794,
+      "step": 3558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664723867762278e-06,
+      "loss": 1.6251,
+      "step": 3559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664628320275177e-06,
+      "loss": 1.5095,
+      "step": 3560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664532772788076e-06,
+      "loss": 1.6035,
+      "step": 3561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664437225300975e-06,
+      "loss": 1.4319,
+      "step": 3562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664341677813874e-06,
+      "loss": 1.7036,
+      "step": 3563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664246130326773e-06,
+      "loss": 1.675,
+      "step": 3564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664150582839671e-06,
+      "loss": 1.1633,
+      "step": 3565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66405503535257e-06,
+      "loss": 1.7577,
+      "step": 3566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663959487865469e-06,
+      "loss": 1.2972,
+      "step": 3567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663863940378368e-06,
+      "loss": 1.7507,
+      "step": 3568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663768392891267e-06,
+      "loss": 2.0615,
+      "step": 3569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663672845404166e-06,
+      "loss": 1.7347,
+      "step": 3570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6635772979170655e-06,
+      "loss": 1.411,
+      "step": 3571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663481750429964e-06,
+      "loss": 2.0313,
+      "step": 3572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663386202942863e-06,
+      "loss": 1.44,
+      "step": 3573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663290655455762e-06,
+      "loss": 1.3123,
+      "step": 3574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663195107968661e-06,
+      "loss": 1.3795,
+      "step": 3575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66309956048156e-06,
+      "loss": 1.3917,
+      "step": 3576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663004012994459e-06,
+      "loss": 1.0922,
+      "step": 3577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662908465507358e-06,
+      "loss": 1.5118,
+      "step": 3578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662812918020256e-06,
+      "loss": 1.4696,
+      "step": 3579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662717370533155e-06,
+      "loss": 1.464,
+      "step": 3580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662621823046054e-06,
+      "loss": 1.2861,
+      "step": 3581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662526275558953e-06,
+      "loss": 1.794,
+      "step": 3582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662430728071852e-06,
+      "loss": 1.2307,
+      "step": 3583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662335180584751e-06,
+      "loss": 1.5388,
+      "step": 3584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66223963309765e-06,
+      "loss": 1.1138,
+      "step": 3585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6621440856105485e-06,
+      "loss": 1.3962,
+      "step": 3586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6620485381234475e-06,
+      "loss": 1.7225,
+      "step": 3587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6619529906363466e-06,
+      "loss": 1.2707,
+      "step": 3588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661857443149246e-06,
+      "loss": 1.2754,
+      "step": 3589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661761895662145e-06,
+      "loss": 1.6496,
+      "step": 3590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661666348175044e-06,
+      "loss": 1.602,
+      "step": 3591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661570800687943e-06,
+      "loss": 1.3101,
+      "step": 3592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661475253200841e-06,
+      "loss": 1.3386,
+      "step": 3593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66137970571374e-06,
+      "loss": 1.4187,
+      "step": 3594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661284158226639e-06,
+      "loss": 1.4527,
+      "step": 3595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661188610739538e-06,
+      "loss": 1.3886,
+      "step": 3596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661093063252437e-06,
+      "loss": 1.8912,
+      "step": 3597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660997515765336e-06,
+      "loss": 1.062,
+      "step": 3598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660901968278234e-06,
+      "loss": 1.7739,
+      "step": 3599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660806420791133e-06,
+      "loss": 1.4267,
+      "step": 3600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660710873304032e-06,
+      "loss": 1.7933,
+      "step": 3601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660615325816931e-06,
+      "loss": 0.9909,
+      "step": 3602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6605197783298305e-06,
+      "loss": 1.5754,
+      "step": 3603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6604242308427295e-06,
+      "loss": 1.7882,
+      "step": 3604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6603286833556285e-06,
+      "loss": 1.704,
+      "step": 3605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660233135868527e-06,
+      "loss": 1.4263,
+      "step": 3606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660137588381426e-06,
+      "loss": 1.5647,
+      "step": 3607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660042040894325e-06,
+      "loss": 1.6036,
+      "step": 3608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659946493407224e-06,
+      "loss": 1.0767,
+      "step": 3609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659850945920122e-06,
+      "loss": 1.5751,
+      "step": 3610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659755398433022e-06,
+      "loss": 1.6004,
+      "step": 3611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659659850945921e-06,
+      "loss": 1.1211,
+      "step": 3612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659564303458819e-06,
+      "loss": 1.4182,
+      "step": 3613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659468755971718e-06,
+      "loss": 1.4418,
+      "step": 3614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659373208484617e-06,
+      "loss": 1.4432,
+      "step": 3615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659277660997516e-06,
+      "loss": 1.1462,
+      "step": 3616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6591821135104145e-06,
+      "loss": 1.2298,
+      "step": 3617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659086566023314e-06,
+      "loss": 0.8461,
+      "step": 3618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658991018536213e-06,
+      "loss": 1.1776,
+      "step": 3619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6588954710491116e-06,
+      "loss": 1.3915,
+      "step": 3620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658799923562011e-06,
+      "loss": 0.9457,
+      "step": 3621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65870437607491e-06,
+      "loss": 1.2765,
+      "step": 3622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658608828587809e-06,
+      "loss": 1.4407,
+      "step": 3623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658513281100707e-06,
+      "loss": 1.0337,
+      "step": 3624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658417733613607e-06,
+      "loss": 1.9361,
+      "step": 3625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658322186126506e-06,
+      "loss": 1.2009,
+      "step": 3626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658226638639404e-06,
+      "loss": 1.532,
+      "step": 3627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658131091152303e-06,
+      "loss": 1.7011,
+      "step": 3628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658035543665202e-06,
+      "loss": 1.8257,
+      "step": 3629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657939996178101e-06,
+      "loss": 1.5172,
+      "step": 3630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657844448690999e-06,
+      "loss": 1.4444,
+      "step": 3631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657748901203898e-06,
+      "loss": 1.6057,
+      "step": 3632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657653353716798e-06,
+      "loss": 1.5106,
+      "step": 3633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657557806229696e-06,
+      "loss": 1.9023,
+      "step": 3634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6574622587425955e-06,
+      "loss": 1.3918,
+      "step": 3635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6573667112554945e-06,
+      "loss": 1.0918,
+      "step": 3636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6572711637683935e-06,
+      "loss": 1.375,
+      "step": 3637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657175616281292e-06,
+      "loss": 1.3521,
+      "step": 3638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657080068794191e-06,
+      "loss": 1.6019,
+      "step": 3639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65698452130709e-06,
+      "loss": 1.3194,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656888973819989e-06,
+      "loss": 1.6679,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656793426332888e-06,
+      "loss": 1.2925,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656697878845787e-06,
+      "loss": 1.4718,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656602331358686e-06,
+      "loss": 0.9301,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656506783871584e-06,
+      "loss": 1.6113,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656411236384483e-06,
+      "loss": 1.6385,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656315688897382e-06,
+      "loss": 1.1541,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656220141410281e-06,
+      "loss": 1.3653,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65612459392318e-06,
+      "loss": 1.2332,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.656029046436079e-06,
+      "loss": 1.2617,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6559334989489775e-06,
+      "loss": 0.9898,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6558379514618766e-06,
+      "loss": 1.8458,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655742403974776e-06,
+      "loss": 1.2012,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655646856487675e-06,
+      "loss": 1.4314,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655551309000574e-06,
+      "loss": 1.6005,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655455761513473e-06,
+      "loss": 1.2867,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655360214026372e-06,
+      "loss": 1.1919,
+      "step": 3657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65526466653927e-06,
+      "loss": 1.4716,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655169119052169e-06,
+      "loss": 1.3411,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.655073571565068e-06,
+      "loss": 1.5632,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654978024077967e-06,
+      "loss": 1.7453,
+      "step": 3661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654882476590866e-06,
+      "loss": 1.7733,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654786929103765e-06,
+      "loss": 1.848,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654691381616664e-06,
+      "loss": 1.3932,
+      "step": 3664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654595834129562e-06,
+      "loss": 1.1889,
+      "step": 3665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654500286642461e-06,
+      "loss": 1.0898,
+      "step": 3666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6544047391553605e-06,
+      "loss": 1.4594,
+      "step": 3667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6543091916682595e-06,
+      "loss": 1.131,
+      "step": 3668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6542136441811585e-06,
+      "loss": 1.3508,
+      "step": 3669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6541180966940576e-06,
+      "loss": 1.4132,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.654022549206957e-06,
+      "loss": 1.384,
+      "step": 3671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653927001719855e-06,
+      "loss": 1.8032,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653831454232754e-06,
+      "loss": 1.4158,
+      "step": 3673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653735906745653e-06,
+      "loss": 1.1331,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653640359258552e-06,
+      "loss": 1.5373,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653544811771451e-06,
+      "loss": 1.4702,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65344926428435e-06,
+      "loss": 1.8261,
+      "step": 3677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653353716797249e-06,
+      "loss": 1.1628,
+      "step": 3678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653258169310147e-06,
+      "loss": 1.3152,
+      "step": 3679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653162621823046e-06,
+      "loss": 1.447,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.653067074335945e-06,
+      "loss": 2.006,
+      "step": 3681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652971526848844e-06,
+      "loss": 1.7317,
+      "step": 3682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652875979361743e-06,
+      "loss": 1.93,
+      "step": 3683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652780431874642e-06,
+      "loss": 0.8605,
+      "step": 3684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6526848843875415e-06,
+      "loss": 2.0038,
+      "step": 3685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.65258933690044e-06,
+      "loss": 1.3416,
+      "step": 3686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652493789413339e-06,
+      "loss": 1.3394,
+      "step": 3687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652398241926238e-06,
+      "loss": 1.6372,
+      "step": 3688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652302694439137e-06,
+      "loss": 1.2225,
+      "step": 3689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652207146952036e-06,
+      "loss": 1.5276,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652111599464935e-06,
+      "loss": 1.3383,
+      "step": 3691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652016051977834e-06,
+      "loss": 1.2545,
+      "step": 3692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.651920504490732e-06,
+      "loss": 1.1524,
+      "step": 3693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.651824957003631e-06,
+      "loss": 1.2591,
+      "step": 3694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65172940951653e-06,
+      "loss": 1.4479,
+      "step": 3695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.651633862029429e-06,
+      "loss": 1.402,
+      "step": 3696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.651538314542328e-06,
+      "loss": 1.1795,
+      "step": 3697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.651442767055227e-06,
+      "loss": 1.2114,
+      "step": 3698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6513472195681255e-06,
+      "loss": 1.7023,
+      "step": 3699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6512516720810245e-06,
+      "loss": 1.7985,
+      "step": 3700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6511561245939235e-06,
+      "loss": 1.2779,
+      "step": 3701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6510605771068226e-06,
+      "loss": 1.4661,
+      "step": 3702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650965029619722e-06,
+      "loss": 1.5354,
+      "step": 3703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65086948213262e-06,
+      "loss": 1.5998,
+      "step": 3704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65077393464552e-06,
+      "loss": 1.8784,
+      "step": 3705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650678387158418e-06,
+      "loss": 1.7546,
+      "step": 3706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650582839671317e-06,
+      "loss": 1.0107,
+      "step": 3707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650487292184216e-06,
+      "loss": 1.4485,
+      "step": 3708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650391744697115e-06,
+      "loss": 1.5905,
+      "step": 3709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650296197210013e-06,
+      "loss": 1.7432,
+      "step": 3710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650200649722912e-06,
+      "loss": 1.5521,
+      "step": 3711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.650105102235812e-06,
+      "loss": 1.7298,
+      "step": 3712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.65000955474871e-06,
+      "loss": 1.6731,
+      "step": 3713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649914007261609e-06,
+      "loss": 1.7455,
+      "step": 3714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649818459774508e-06,
+      "loss": 1.713,
+      "step": 3715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649722912287407e-06,
+      "loss": 1.1351,
+      "step": 3716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649627364800306e-06,
+      "loss": 1.2779,
+      "step": 3717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649531817313205e-06,
+      "loss": 0.9072,
+      "step": 3718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6494362698261045e-06,
+      "loss": 1.3923,
+      "step": 3719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649340722339003e-06,
+      "loss": 1.4143,
+      "step": 3720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649245174851902e-06,
+      "loss": 1.5659,
+      "step": 3721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649149627364801e-06,
+      "loss": 1.1484,
+      "step": 3722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6490540798777e-06,
+      "loss": 1.5617,
+      "step": 3723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648958532390598e-06,
+      "loss": 1.497,
+      "step": 3724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648862984903497e-06,
+      "loss": 1.4847,
+      "step": 3725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648767437416397e-06,
+      "loss": 1.3176,
+      "step": 3726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648671889929295e-06,
+      "loss": 1.4801,
+      "step": 3727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648576342442194e-06,
+      "loss": 1.5648,
+      "step": 3728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648480794955093e-06,
+      "loss": 1.4461,
+      "step": 3729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648385247467992e-06,
+      "loss": 1.1399,
+      "step": 3730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6482896999808905e-06,
+      "loss": 1.2595,
+      "step": 3731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6481941524937895e-06,
+      "loss": 1.6707,
+      "step": 3732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.648098605006689e-06,
+      "loss": 1.5525,
+      "step": 3733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6480030575195876e-06,
+      "loss": 1.4966,
+      "step": 3734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647907510032487e-06,
+      "loss": 1.4181,
+      "step": 3735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647811962545386e-06,
+      "loss": 1.1579,
+      "step": 3736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647716415058285e-06,
+      "loss": 1.5311,
+      "step": 3737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647620867571183e-06,
+      "loss": 1.3841,
+      "step": 3738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647525320084082e-06,
+      "loss": 1.868,
+      "step": 3739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647429772596981e-06,
+      "loss": 1.5603,
+      "step": 3740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.64733422510988e-06,
+      "loss": 1.3726,
+      "step": 3741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647238677622779e-06,
+      "loss": 1.3411,
+      "step": 3742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647143130135678e-06,
+      "loss": 1.2703,
+      "step": 3743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647047582648577e-06,
+      "loss": 1.6622,
+      "step": 3744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646952035161475e-06,
+      "loss": 1.6168,
+      "step": 3745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646856487674374e-06,
+      "loss": 1.7766,
+      "step": 3746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646760940187273e-06,
+      "loss": 1.1591,
+      "step": 3747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646665392700172e-06,
+      "loss": 1.1874,
+      "step": 3748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6465698452130715e-06,
+      "loss": 1.3684,
+      "step": 3749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6464742977259705e-06,
+      "loss": 1.3723,
+      "step": 3750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646378750238869e-06,
+      "loss": 1.6877,
+      "step": 3751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646283202751768e-06,
+      "loss": 1.2569,
+      "step": 3752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646187655264667e-06,
+      "loss": 1.2353,
+      "step": 3753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.646092107777566e-06,
+      "loss": 1.2935,
+      "step": 3754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645996560290465e-06,
+      "loss": 1.7192,
+      "step": 3755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645901012803364e-06,
+      "loss": 1.2103,
+      "step": 3756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645805465316263e-06,
+      "loss": 1.2637,
+      "step": 3757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645709917829161e-06,
+      "loss": 1.728,
+      "step": 3758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64561437034206e-06,
+      "loss": 1.6274,
+      "step": 3759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645518822854959e-06,
+      "loss": 1.5522,
+      "step": 3760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645423275367858e-06,
+      "loss": 1.4517,
+      "step": 3761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645327727880757e-06,
+      "loss": 1.1673,
+      "step": 3762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645232180393656e-06,
+      "loss": 1.9181,
+      "step": 3763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.645136632906555e-06,
+      "loss": 1.5211,
+      "step": 3764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6450410854194535e-06,
+      "loss": 1.8667,
+      "step": 3765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6449455379323526e-06,
+      "loss": 1.5584,
+      "step": 3766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644849990445252e-06,
+      "loss": 1.5768,
+      "step": 3767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644754442958151e-06,
+      "loss": 1.3092,
+      "step": 3768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64465889547105e-06,
+      "loss": 1.0548,
+      "step": 3769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644563347983949e-06,
+      "loss": 1.1796,
+      "step": 3770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644467800496848e-06,
+      "loss": 0.9743,
+      "step": 3771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644372253009746e-06,
+      "loss": 1.2997,
+      "step": 3772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644276705522645e-06,
+      "loss": 1.3456,
+      "step": 3773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644181158035544e-06,
+      "loss": 1.7084,
+      "step": 3774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.644085610548443e-06,
+      "loss": 1.2421,
+      "step": 3775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643990063061341e-06,
+      "loss": 1.758,
+      "step": 3776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643894515574241e-06,
+      "loss": 1.5297,
+      "step": 3777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64379896808714e-06,
+      "loss": 1.4094,
+      "step": 3778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643703420600038e-06,
+      "loss": 1.1398,
+      "step": 3779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643607873112937e-06,
+      "loss": 1.4365,
+      "step": 3780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6435123256258365e-06,
+      "loss": 1.8753,
+      "step": 3781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6434167781387355e-06,
+      "loss": 1.8761,
+      "step": 3782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643321230651634e-06,
+      "loss": 1.7163,
+      "step": 3783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6432256831645336e-06,
+      "loss": 1.4392,
+      "step": 3784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643130135677433e-06,
+      "loss": 1.2314,
+      "step": 3785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.643034588190331e-06,
+      "loss": 1.2789,
+      "step": 3786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64293904070323e-06,
+      "loss": 1.462,
+      "step": 3787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642843493216129e-06,
+      "loss": 1.4761,
+      "step": 3788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642747945729028e-06,
+      "loss": 1.8316,
+      "step": 3789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642652398241926e-06,
+      "loss": 0.8718,
+      "step": 3790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642556850754826e-06,
+      "loss": 1.4976,
+      "step": 3791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642461303267724e-06,
+      "loss": 1.5397,
+      "step": 3792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642365755780623e-06,
+      "loss": 1.4551,
+      "step": 3793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642270208293522e-06,
+      "loss": 1.465,
+      "step": 3794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642174660806421e-06,
+      "loss": 1.6582,
+      "step": 3795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.64207911331932e-06,
+      "loss": 1.4096,
+      "step": 3796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6419835658322185e-06,
+      "loss": 1.2114,
+      "step": 3797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641888018345118e-06,
+      "loss": 1.2991,
+      "step": 3798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641792470858017e-06,
+      "loss": 1.4192,
+      "step": 3799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641696923370916e-06,
+      "loss": 1.7311,
+      "step": 3800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641601375883815e-06,
+      "loss": 1.1037,
+      "step": 3801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641505828396714e-06,
+      "loss": 1.4724,
+      "step": 3802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641410280909612e-06,
+      "loss": 1.6513,
+      "step": 3803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641314733422511e-06,
+      "loss": 0.6429,
+      "step": 3804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641219185935411e-06,
+      "loss": 1.4946,
+      "step": 3805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641123638448309e-06,
+      "loss": 1.3759,
+      "step": 3806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.641028090961208e-06,
+      "loss": 1.0034,
+      "step": 3807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640932543474107e-06,
+      "loss": 1.1451,
+      "step": 3808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640836995987006e-06,
+      "loss": 1.7188,
+      "step": 3809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640741448499904e-06,
+      "loss": 1.1554,
+      "step": 3810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640645901012803e-06,
+      "loss": 1.5096,
+      "step": 3811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640550353525702e-06,
+      "loss": 1.543,
+      "step": 3812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6404548060386015e-06,
+      "loss": 0.9353,
+      "step": 3813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6403592585515005e-06,
+      "loss": 1.0113,
+      "step": 3814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6402637110643995e-06,
+      "loss": 1.0886,
+      "step": 3815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6401681635772986e-06,
+      "loss": 1.5557,
+      "step": 3816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.640072616090197e-06,
+      "loss": 1.4597,
+      "step": 3817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639977068603096e-06,
+      "loss": 1.1573,
+      "step": 3818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639881521115995e-06,
+      "loss": 1.3987,
+      "step": 3819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639785973628894e-06,
+      "loss": 2.0058,
+      "step": 3820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639690426141793e-06,
+      "loss": 1.49,
+      "step": 3821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639594878654692e-06,
+      "loss": 1.3867,
+      "step": 3822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639499331167591e-06,
+      "loss": 1.329,
+      "step": 3823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639403783680489e-06,
+      "loss": 1.6064,
+      "step": 3824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639308236193388e-06,
+      "loss": 1.4778,
+      "step": 3825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639212688706287e-06,
+      "loss": 1.224,
+      "step": 3826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639117141219186e-06,
+      "loss": 1.3494,
+      "step": 3827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.639021593732085e-06,
+      "loss": 1.413,
+      "step": 3828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638926046244984e-06,
+      "loss": 1.2956,
+      "step": 3829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638830498757883e-06,
+      "loss": 2.2293,
+      "step": 3830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638734951270782e-06,
+      "loss": 1.5599,
+      "step": 3831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638639403783681e-06,
+      "loss": 1.5589,
+      "step": 3832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.63854385629658e-06,
+      "loss": 1.4369,
+      "step": 3833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638448308809479e-06,
+      "loss": 1.489,
+      "step": 3834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638352761322378e-06,
+      "loss": 1.203,
+      "step": 3835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638257213835277e-06,
+      "loss": 1.8351,
+      "step": 3836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638161666348176e-06,
+      "loss": 1.718,
+      "step": 3837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.638066118861074e-06,
+      "loss": 1.2714,
+      "step": 3838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637970571373973e-06,
+      "loss": 1.5611,
+      "step": 3839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637875023886872e-06,
+      "loss": 1.3681,
+      "step": 3840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637779476399771e-06,
+      "loss": 1.5458,
+      "step": 3841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.63768392891267e-06,
+      "loss": 1.6431,
+      "step": 3842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637588381425569e-06,
+      "loss": 1.8056,
+      "step": 3843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637492833938467e-06,
+      "loss": 1.2268,
+      "step": 3844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6373972864513664e-06,
+      "loss": 1.6469,
+      "step": 3845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6373017389642655e-06,
+      "loss": 1.4981,
+      "step": 3846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6372061914771645e-06,
+      "loss": 1.574,
+      "step": 3847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6371106439900636e-06,
+      "loss": 1.9307,
+      "step": 3848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637015096502963e-06,
+      "loss": 1.2024,
+      "step": 3849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636919549015862e-06,
+      "loss": 0.9917,
+      "step": 3850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63682400152876e-06,
+      "loss": 1.3923,
+      "step": 3851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636728454041659e-06,
+      "loss": 1.5766,
+      "step": 3852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636632906554558e-06,
+      "loss": 1.4027,
+      "step": 3853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636537359067457e-06,
+      "loss": 1.517,
+      "step": 3854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636441811580355e-06,
+      "loss": 0.9775,
+      "step": 3855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636346264093255e-06,
+      "loss": 1.5524,
+      "step": 3856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636250716606154e-06,
+      "loss": 1.8942,
+      "step": 3857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636155169119052e-06,
+      "loss": 1.3366,
+      "step": 3858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.636059621631951e-06,
+      "loss": 1.2385,
+      "step": 3859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63596407414485e-06,
+      "loss": 1.2479,
+      "step": 3860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.635868526657749e-06,
+      "loss": 1.1647,
+      "step": 3861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6357729791706476e-06,
+      "loss": 1.7944,
+      "step": 3862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6356774316835475e-06,
+      "loss": 1.4861,
+      "step": 3863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6355818841964465e-06,
+      "loss": 1.5896,
+      "step": 3864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.635486336709345e-06,
+      "loss": 1.3712,
+      "step": 3865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.635390789222244e-06,
+      "loss": 1.2266,
+      "step": 3866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.635295241735143e-06,
+      "loss": 1.5194,
+      "step": 3867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.635199694248042e-06,
+      "loss": 1.7831,
+      "step": 3868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63510414676094e-06,
+      "loss": 1.49,
+      "step": 3869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63500859927384e-06,
+      "loss": 1.8316,
+      "step": 3870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634913051786739e-06,
+      "loss": 1.1372,
+      "step": 3871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634817504299637e-06,
+      "loss": 0.9083,
+      "step": 3872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634721956812536e-06,
+      "loss": 1.6157,
+      "step": 3873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634626409325435e-06,
+      "loss": 1.0687,
+      "step": 3874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634530861838334e-06,
+      "loss": 1.8865,
+      "step": 3875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634435314351232e-06,
+      "loss": 1.783,
+      "step": 3876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634339766864132e-06,
+      "loss": 1.7018,
+      "step": 3877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.634244219377031e-06,
+      "loss": 1.3626,
+      "step": 3878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6341486718899295e-06,
+      "loss": 1.6859,
+      "step": 3879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6340531244028286e-06,
+      "loss": 0.846,
+      "step": 3880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633957576915728e-06,
+      "loss": 1.2663,
+      "step": 3881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633862029428627e-06,
+      "loss": 1.4597,
+      "step": 3882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633766481941525e-06,
+      "loss": 1.5743,
+      "step": 3883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633670934454424e-06,
+      "loss": 1.3204,
+      "step": 3884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633575386967324e-06,
+      "loss": 1.5813,
+      "step": 3885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633479839480222e-06,
+      "loss": 1.558,
+      "step": 3886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633384291993121e-06,
+      "loss": 1.232,
+      "step": 3887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63328874450602e-06,
+      "loss": 1.3514,
+      "step": 3888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633193197018919e-06,
+      "loss": 1.2961,
+      "step": 3889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633097649531817e-06,
+      "loss": 1.3861,
+      "step": 3890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.633002102044716e-06,
+      "loss": 1.3619,
+      "step": 3891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632906554557615e-06,
+      "loss": 1.5167,
+      "step": 3892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632811007070514e-06,
+      "loss": 1.3139,
+      "step": 3893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632715459583413e-06,
+      "loss": 1.2631,
+      "step": 3894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6326199120963124e-06,
+      "loss": 1.3646,
+      "step": 3895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6325243646092115e-06,
+      "loss": 1.3218,
+      "step": 3896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.63242881712211e-06,
+      "loss": 1.6196,
+      "step": 3897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632333269635009e-06,
+      "loss": 1.3081,
+      "step": 3898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632237722147908e-06,
+      "loss": 1.1184,
+      "step": 3899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632142174660807e-06,
+      "loss": 1.4122,
+      "step": 3900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.632046627173706e-06,
+      "loss": 1.8144,
+      "step": 3901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.631951079686605e-06,
+      "loss": 1.1184,
+      "step": 3902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631855532199503e-06,
+      "loss": 1.6082,
+      "step": 3903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631759984712402e-06,
+      "loss": 1.1605,
+      "step": 3904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631664437225301e-06,
+      "loss": 1.1838,
+      "step": 3905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6315688897382e-06,
+      "loss": 1.4883,
+      "step": 3906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631473342251099e-06,
+      "loss": 1.6095,
+      "step": 3907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631377794763998e-06,
+      "loss": 1.4475,
+      "step": 3908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.631282247276897e-06,
+      "loss": 1.2805,
+      "step": 3909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6311866997897955e-06,
+      "loss": 0.9792,
+      "step": 3910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6310911523026945e-06,
+      "loss": 1.3832,
+      "step": 3911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6309956048155936e-06,
+      "loss": 1.5891,
+      "step": 3912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630900057328493e-06,
+      "loss": 0.9795,
+      "step": 3913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630804509841392e-06,
+      "loss": 1.6709,
+      "step": 3914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630708962354291e-06,
+      "loss": 1.1788,
+      "step": 3915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.63061341486719e-06,
+      "loss": 0.8825,
+      "step": 3916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630517867380088e-06,
+      "loss": 1.5559,
+      "step": 3917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630422319892987e-06,
+      "loss": 1.4631,
+      "step": 3918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630326772405886e-06,
+      "loss": 1.5087,
+      "step": 3919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630231224918785e-06,
+      "loss": 1.3263,
+      "step": 3920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630135677431684e-06,
+      "loss": 1.4945,
+      "step": 3921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.630040129944583e-06,
+      "loss": 1.0895,
+      "step": 3922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629944582457482e-06,
+      "loss": 1.4082,
+      "step": 3923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.62984903497038e-06,
+      "loss": 1.5548,
+      "step": 3924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629753487483279e-06,
+      "loss": 1.855,
+      "step": 3925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629657939996178e-06,
+      "loss": 0.8523,
+      "step": 3926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6295623925090774e-06,
+      "loss": 1.3128,
+      "step": 3927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6294668450219765e-06,
+      "loss": 1.4791,
+      "step": 3928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6293712975348755e-06,
+      "loss": 1.0961,
+      "step": 3929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6292757500477746e-06,
+      "loss": 1.4316,
+      "step": 3930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629180202560673e-06,
+      "loss": 1.3035,
+      "step": 3931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.629084655073572e-06,
+      "loss": 1.1121,
+      "step": 3932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628989107586471e-06,
+      "loss": 1.7224,
+      "step": 3933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.62889356009937e-06,
+      "loss": 1.3481,
+      "step": 3934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628798012612269e-06,
+      "loss": 1.3868,
+      "step": 3935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628702465125168e-06,
+      "loss": 1.65,
+      "step": 3936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628606917638067e-06,
+      "loss": 1.572,
+      "step": 3937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628511370150965e-06,
+      "loss": 1.2077,
+      "step": 3938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628415822663864e-06,
+      "loss": 1.4132,
+      "step": 3939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628320275176763e-06,
+      "loss": 1.6745,
+      "step": 3940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628224727689662e-06,
+      "loss": 1.2518,
+      "step": 3941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.628129180202561e-06,
+      "loss": 1.3421,
+      "step": 3942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.62803363271546e-06,
+      "loss": 1.3538,
+      "step": 3943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6279380852283586e-06,
+      "loss": 1.2639,
+      "step": 3944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627842537741258e-06,
+      "loss": 0.8553,
+      "step": 3945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627746990254157e-06,
+      "loss": 1.2691,
+      "step": 3946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627651442767056e-06,
+      "loss": 1.4975,
+      "step": 3947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627555895279955e-06,
+      "loss": 1.47,
+      "step": 3948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627460347792854e-06,
+      "loss": 1.6067,
+      "step": 3949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627364800305753e-06,
+      "loss": 1.5658,
+      "step": 3950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627269252818651e-06,
+      "loss": 1.2404,
+      "step": 3951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.62717370533155e-06,
+      "loss": 1.1707,
+      "step": 3952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627078157844449e-06,
+      "loss": 1.0964,
+      "step": 3953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.626982610357348e-06,
+      "loss": 1.3482,
+      "step": 3954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626887062870246e-06,
+      "loss": 1.1835,
+      "step": 3955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626791515383145e-06,
+      "loss": 1.4391,
+      "step": 3956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626695967896045e-06,
+      "loss": 1.3217,
+      "step": 3957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626600420408943e-06,
+      "loss": 1.639,
+      "step": 3958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6265048729218424e-06,
+      "loss": 1.0703,
+      "step": 3959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6264093254347415e-06,
+      "loss": 1.659,
+      "step": 3960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6263137779476405e-06,
+      "loss": 1.4004,
+      "step": 3961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626218230460539e-06,
+      "loss": 2.1693,
+      "step": 3962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626122682973438e-06,
+      "loss": 1.4161,
+      "step": 3963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.626027135486338e-06,
+      "loss": 1.631,
+      "step": 3964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625931587999236e-06,
+      "loss": 1.4453,
+      "step": 3965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625836040512135e-06,
+      "loss": 1.3876,
+      "step": 3966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625740493025034e-06,
+      "loss": 1.5119,
+      "step": 3967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625644945537933e-06,
+      "loss": 1.5341,
+      "step": 3968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625549398050831e-06,
+      "loss": 1.5744,
+      "step": 3969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62545385056373e-06,
+      "loss": 1.5078,
+      "step": 3970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62535830307663e-06,
+      "loss": 1.6349,
+      "step": 3971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625262755589528e-06,
+      "loss": 0.9871,
+      "step": 3972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625167208102427e-06,
+      "loss": 1.757,
+      "step": 3973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.625071660615326e-06,
+      "loss": 1.8107,
+      "step": 3974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624976113128225e-06,
+      "loss": 1.5609,
+      "step": 3975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6248805656411236e-06,
+      "loss": 0.9532,
+      "step": 3976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624785018154023e-06,
+      "loss": 1.9597,
+      "step": 3977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6246894706669225e-06,
+      "loss": 1.1976,
+      "step": 3978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624593923179821e-06,
+      "loss": 1.3786,
+      "step": 3979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62449837569272e-06,
+      "loss": 1.5643,
+      "step": 3980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624402828205619e-06,
+      "loss": 1.12,
+      "step": 3981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624307280718518e-06,
+      "loss": 1.2022,
+      "step": 3982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624211733231416e-06,
+      "loss": 1.5467,
+      "step": 3983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624116185744315e-06,
+      "loss": 1.373,
+      "step": 3984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.624020638257214e-06,
+      "loss": 1.3293,
+      "step": 3985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.623925090770113e-06,
+      "loss": 1.5235,
+      "step": 3986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.623829543283012e-06,
+      "loss": 1.0675,
+      "step": 3987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.623733995795911e-06,
+      "loss": 1.2914,
+      "step": 3988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62363844830881e-06,
+      "loss": 1.2254,
+      "step": 3989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.623542900821708e-06,
+      "loss": 1.2583,
+      "step": 3990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6234473533346074e-06,
+      "loss": 1.7509,
+      "step": 3991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6233518058475065e-06,
+      "loss": 1.2829,
+      "step": 3992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6232562583604055e-06,
+      "loss": 1.4732,
+      "step": 3993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6231607108733046e-06,
+      "loss": 1.3499,
+      "step": 3994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.623065163386204e-06,
+      "loss": 1.4395,
+      "step": 3995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622969615899102e-06,
+      "loss": 0.9306,
+      "step": 3996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622874068412001e-06,
+      "loss": 1.5769,
+      "step": 3997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6227785209249e-06,
+      "loss": 1.4733,
+      "step": 3998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622682973437799e-06,
+      "loss": 1.0559,
+      "step": 3999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622587425950698e-06,
+      "loss": 1.4539,
+      "step": 4000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622491878463597e-06,
+      "loss": 1.2282,
+      "step": 4001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622396330976496e-06,
+      "loss": 1.3745,
+      "step": 4002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622300783489394e-06,
+      "loss": 1.4813,
+      "step": 4003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622205236002293e-06,
+      "loss": 1.3508,
+      "step": 4004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622109688515192e-06,
+      "loss": 1.479,
+      "step": 4005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622014141028091e-06,
+      "loss": 1.5049,
+      "step": 4006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.62191859354099e-06,
+      "loss": 1.5264,
+      "step": 4007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621823046053889e-06,
+      "loss": 1.1672,
+      "step": 4008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6217274985667884e-06,
+      "loss": 1.489,
+      "step": 4009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621631951079687e-06,
+      "loss": 1.2344,
+      "step": 4010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621536403592586e-06,
+      "loss": 1.6593,
+      "step": 4011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621440856105485e-06,
+      "loss": 1.8316,
+      "step": 4012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621345308618384e-06,
+      "loss": 0.9408,
+      "step": 4013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621249761131283e-06,
+      "loss": 1.6465,
+      "step": 4014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621154213644182e-06,
+      "loss": 1.1782,
+      "step": 4015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621058666157081e-06,
+      "loss": 1.522,
+      "step": 4016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620963118669979e-06,
+      "loss": 1.679,
+      "step": 4017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620867571182878e-06,
+      "loss": 1.5256,
+      "step": 4018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620772023695777e-06,
+      "loss": 1.3904,
+      "step": 4019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620676476208676e-06,
+      "loss": 1.4179,
+      "step": 4020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620580928721575e-06,
+      "loss": 1.2869,
+      "step": 4021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620485381234474e-06,
+      "loss": 1.3127,
+      "step": 4022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620389833747373e-06,
+      "loss": 1.3432,
+      "step": 4023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6202942862602715e-06,
+      "loss": 1.2684,
+      "step": 4024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6201987387731705e-06,
+      "loss": 1.2994,
+      "step": 4025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6201031912860696e-06,
+      "loss": 1.495,
+      "step": 4026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620007643798969e-06,
+      "loss": 1.6689,
+      "step": 4027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619912096311867e-06,
+      "loss": 1.3413,
+      "step": 4028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619816548824767e-06,
+      "loss": 1.3125,
+      "step": 4029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619721001337666e-06,
+      "loss": 1.3217,
+      "step": 4030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619625453850564e-06,
+      "loss": 1.2548,
+      "step": 4031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619529906363463e-06,
+      "loss": 1.3943,
+      "step": 4032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619434358876362e-06,
+      "loss": 1.534,
+      "step": 4033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619338811389261e-06,
+      "loss": 1.5901,
+      "step": 4034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619243263902159e-06,
+      "loss": 1.7725,
+      "step": 4035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619147716415059e-06,
+      "loss": 1.4415,
+      "step": 4036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.619052168927957e-06,
+      "loss": 1.5652,
+      "step": 4037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618956621440856e-06,
+      "loss": 1.5125,
+      "step": 4038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618861073953755e-06,
+      "loss": 1.2455,
+      "step": 4039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618765526466654e-06,
+      "loss": 1.3946,
+      "step": 4040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6186699789795534e-06,
+      "loss": 1.7105,
+      "step": 4041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618574431492452e-06,
+      "loss": 1.8706,
+      "step": 4042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6184788840053515e-06,
+      "loss": 1.2779,
+      "step": 4043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.61838333651825e-06,
+      "loss": 1.4057,
+      "step": 4044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618287789031149e-06,
+      "loss": 1.3465,
+      "step": 4045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618192241544048e-06,
+      "loss": 1.6424,
+      "step": 4046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618096694056947e-06,
+      "loss": 1.3921,
+      "step": 4047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618001146569845e-06,
+      "loss": 1.5597,
+      "step": 4048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617905599082744e-06,
+      "loss": 1.4997,
+      "step": 4049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617810051595644e-06,
+      "loss": 1.3904,
+      "step": 4050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617714504108542e-06,
+      "loss": 1.1619,
+      "step": 4051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617618956621441e-06,
+      "loss": 1.7098,
+      "step": 4052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.61752340913434e-06,
+      "loss": 1.4442,
+      "step": 4053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617427861647239e-06,
+      "loss": 1.3337,
+      "step": 4054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6173323141601374e-06,
+      "loss": 1.1948,
+      "step": 4055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6172367666730365e-06,
+      "loss": 1.5865,
+      "step": 4056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617141219185936e-06,
+      "loss": 1.2897,
+      "step": 4057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6170456716988346e-06,
+      "loss": 1.5865,
+      "step": 4058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616950124211734e-06,
+      "loss": 1.7056,
+      "step": 4059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616854576724633e-06,
+      "loss": 1.3738,
+      "step": 4060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616759029237532e-06,
+      "loss": 1.3463,
+      "step": 4061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.61666348175043e-06,
+      "loss": 1.5376,
+      "step": 4062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616567934263329e-06,
+      "loss": 1.0765,
+      "step": 4063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616472386776228e-06,
+      "loss": 1.2851,
+      "step": 4064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616376839289127e-06,
+      "loss": 1.3443,
+      "step": 4065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616281291802026e-06,
+      "loss": 1.6321,
+      "step": 4066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616185744314925e-06,
+      "loss": 1.2893,
+      "step": 4067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.616090196827824e-06,
+      "loss": 1.8069,
+      "step": 4068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615994649340722e-06,
+      "loss": 1.4739,
+      "step": 4069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615899101853621e-06,
+      "loss": 0.983,
+      "step": 4070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.61580355436652e-06,
+      "loss": 1.4088,
+      "step": 4071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615708006879419e-06,
+      "loss": 1.0632,
+      "step": 4072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6156124593923184e-06,
+      "loss": 1.1524,
+      "step": 4073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6155169119052175e-06,
+      "loss": 1.5771,
+      "step": 4074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6154213644181165e-06,
+      "loss": 1.6267,
+      "step": 4075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615325816931015e-06,
+      "loss": 1.1601,
+      "step": 4076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615230269443914e-06,
+      "loss": 1.2916,
+      "step": 4077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615134721956813e-06,
+      "loss": 1.4143,
+      "step": 4078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615039174469712e-06,
+      "loss": 1.4639,
+      "step": 4079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614943626982611e-06,
+      "loss": 0.7243,
+      "step": 4080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.61484807949551e-06,
+      "loss": 1.4096,
+      "step": 4081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614752532008409e-06,
+      "loss": 1.4613,
+      "step": 4082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614656984521307e-06,
+      "loss": 1.4285,
+      "step": 4083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614561437034206e-06,
+      "loss": 1.4897,
+      "step": 4084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614465889547105e-06,
+      "loss": 1.3651,
+      "step": 4085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614370342060004e-06,
+      "loss": 1.0998,
+      "step": 4086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614274794572903e-06,
+      "loss": 1.8553,
+      "step": 4087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614179247085802e-06,
+      "loss": 1.7243,
+      "step": 4088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.614083699598701e-06,
+      "loss": 1.062,
+      "step": 4089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6139881521115996e-06,
+      "loss": 1.4689,
+      "step": 4090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613892604624499e-06,
+      "loss": 1.3794,
+      "step": 4091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613797057137398e-06,
+      "loss": 1.2291,
+      "step": 4092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613701509650297e-06,
+      "loss": 1.4462,
+      "step": 4093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613605962163196e-06,
+      "loss": 1.3557,
+      "step": 4094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613510414676095e-06,
+      "loss": 1.4569,
+      "step": 4095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613414867188993e-06,
+      "loss": 1.5476,
+      "step": 4096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613319319701892e-06,
+      "loss": 1.2161,
+      "step": 4097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613223772214791e-06,
+      "loss": 1.6965,
+      "step": 4098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.61312822472769e-06,
+      "loss": 1.2532,
+      "step": 4099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.613032677240588e-06,
+      "loss": 1.2213,
+      "step": 4100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612937129753488e-06,
+      "loss": 1.5634,
+      "step": 4101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612841582266387e-06,
+      "loss": 1.7474,
+      "step": 4102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612746034779285e-06,
+      "loss": 1.2505,
+      "step": 4103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612650487292184e-06,
+      "loss": 1.602,
+      "step": 4104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6125549398050834e-06,
+      "loss": 1.7607,
+      "step": 4105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6124593923179825e-06,
+      "loss": 1.0875,
+      "step": 4106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612363844830881e-06,
+      "loss": 1.1116,
+      "step": 4107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6122682973437806e-06,
+      "loss": 1.6557,
+      "step": 4108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.61217274985668e-06,
+      "loss": 1.3199,
+      "step": 4109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.612077202369578e-06,
+      "loss": 1.5665,
+      "step": 4110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.611981654882477e-06,
+      "loss": 1.2457,
+      "step": 4111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611886107395376e-06,
+      "loss": 1.1341,
+      "step": 4112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611790559908275e-06,
+      "loss": 1.3073,
+      "step": 4113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611695012421173e-06,
+      "loss": 1.1227,
+      "step": 4114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611599464934073e-06,
+      "loss": 1.6645,
+      "step": 4115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611503917446972e-06,
+      "loss": 1.2531,
+      "step": 4116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61140836995987e-06,
+      "loss": 1.2712,
+      "step": 4117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611312822472769e-06,
+      "loss": 1.1483,
+      "step": 4118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611217274985668e-06,
+      "loss": 1.5592,
+      "step": 4119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611121727498567e-06,
+      "loss": 1.5969,
+      "step": 4120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6110261800114655e-06,
+      "loss": 1.3289,
+      "step": 4121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610930632524365e-06,
+      "loss": 1.5491,
+      "step": 4122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6108350850372644e-06,
+      "loss": 1.642,
+      "step": 4123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610739537550163e-06,
+      "loss": 1.4775,
+      "step": 4124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610643990063062e-06,
+      "loss": 0.9683,
+      "step": 4125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610548442575961e-06,
+      "loss": 1.5303,
+      "step": 4126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.61045289508886e-06,
+      "loss": 1.4103,
+      "step": 4127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610357347601758e-06,
+      "loss": 1.7998,
+      "step": 4128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610261800114658e-06,
+      "loss": 1.9077,
+      "step": 4129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610166252627557e-06,
+      "loss": 1.1382,
+      "step": 4130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.610070705140455e-06,
+      "loss": 1.2569,
+      "step": 4131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609975157653354e-06,
+      "loss": 1.3366,
+      "step": 4132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609879610166253e-06,
+      "loss": 1.7537,
+      "step": 4133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609784062679152e-06,
+      "loss": 1.4822,
+      "step": 4134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60968851519205e-06,
+      "loss": 0.8235,
+      "step": 4135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609592967704949e-06,
+      "loss": 1.4202,
+      "step": 4136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6094974202178484e-06,
+      "loss": 1.293,
+      "step": 4137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6094018727307475e-06,
+      "loss": 1.6264,
+      "step": 4138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6093063252436465e-06,
+      "loss": 1.528,
+      "step": 4139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6092107777565456e-06,
+      "loss": 1.2393,
+      "step": 4140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609115230269445e-06,
+      "loss": 1.4688,
+      "step": 4141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.609019682782343e-06,
+      "loss": 1.9878,
+      "step": 4142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608924135295242e-06,
+      "loss": 1.3704,
+      "step": 4143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608828587808141e-06,
+      "loss": 1.344,
+      "step": 4144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60873304032104e-06,
+      "loss": 1.3531,
+      "step": 4145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608637492833939e-06,
+      "loss": 1.4661,
+      "step": 4146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608541945346838e-06,
+      "loss": 1.5742,
+      "step": 4147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608446397859736e-06,
+      "loss": 1.1657,
+      "step": 4148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608350850372635e-06,
+      "loss": 1.0801,
+      "step": 4149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608255302885534e-06,
+      "loss": 1.6544,
+      "step": 4150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608159755398433e-06,
+      "loss": 1.5545,
+      "step": 4151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608064207911332e-06,
+      "loss": 1.0203,
+      "step": 4152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607968660424231e-06,
+      "loss": 1.5496,
+      "step": 4153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60787311293713e-06,
+      "loss": 1.1784,
+      "step": 4154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607777565450029e-06,
+      "loss": 1.5323,
+      "step": 4155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607682017962928e-06,
+      "loss": 1.9794,
+      "step": 4156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607586470475827e-06,
+      "loss": 1.1136,
+      "step": 4157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607490922988726e-06,
+      "loss": 1.5246,
+      "step": 4158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607395375501625e-06,
+      "loss": 2.0534,
+      "step": 4159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607299828014524e-06,
+      "loss": 1.2959,
+      "step": 4160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607204280527423e-06,
+      "loss": 1.4729,
+      "step": 4161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607108733040321e-06,
+      "loss": 1.1937,
+      "step": 4162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60701318555322e-06,
+      "loss": 1.326,
+      "step": 4163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.606917638066119e-06,
+      "loss": 1.4137,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.606822090579018e-06,
+      "loss": 1.7163,
+      "step": 4165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.606726543091917e-06,
+      "loss": 1.4759,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.606630995604816e-06,
+      "loss": 1.6288,
+      "step": 4167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.606535448117715e-06,
+      "loss": 1.3001,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6064399006306134e-06,
+      "loss": 1.4018,
+      "step": 4169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6063443531435125e-06,
+      "loss": 1.3222,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6062488056564115e-06,
+      "loss": 1.2471,
+      "step": 4171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6061532581693106e-06,
+      "loss": 1.4147,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.60605771068221e-06,
+      "loss": 1.3579,
+      "step": 4173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605962163195109e-06,
+      "loss": 1.5401,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605866615708008e-06,
+      "loss": 1.3648,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605771068220906e-06,
+      "loss": 1.1585,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605675520733805e-06,
+      "loss": 1.9428,
+      "step": 4177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605579973246704e-06,
+      "loss": 1.1008,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605484425759603e-06,
+      "loss": 1.233,
+      "step": 4179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605388878272502e-06,
+      "loss": 1.7121,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605293330785401e-06,
+      "loss": 0.8821,
+      "step": 4181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6051977832983e-06,
+      "loss": 1.609,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605102235811198e-06,
+      "loss": 1.2996,
+      "step": 4183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.605006688324097e-06,
+      "loss": 1.9172,
+      "step": 4184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604911140836996e-06,
+      "loss": 1.3014,
+      "step": 4185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604815593349895e-06,
+      "loss": 1.5187,
+      "step": 4186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6047200458627944e-06,
+      "loss": 1.8041,
+      "step": 4187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6046244983756935e-06,
+      "loss": 1.3252,
+      "step": 4188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604528950888592e-06,
+      "loss": 1.92,
+      "step": 4189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604433403401491e-06,
+      "loss": 1.8365,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.60433785591439e-06,
+      "loss": 1.2228,
+      "step": 4191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604242308427289e-06,
+      "loss": 1.2918,
+      "step": 4192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604146760940188e-06,
+      "loss": 0.992,
+      "step": 4193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.604051213453087e-06,
+      "loss": 1.3953,
+      "step": 4194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603955665965986e-06,
+      "loss": 1.4006,
+      "step": 4195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603860118478884e-06,
+      "loss": 1.3036,
+      "step": 4196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603764570991783e-06,
+      "loss": 1.1841,
+      "step": 4197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603669023504682e-06,
+      "loss": 1.5987,
+      "step": 4198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603573476017581e-06,
+      "loss": 1.6429,
+      "step": 4199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603477928530479e-06,
+      "loss": 2.0342,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603382381043379e-06,
+      "loss": 1.6251,
+      "step": 4201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603286833556278e-06,
+      "loss": 1.1898,
+      "step": 4202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6031912860691765e-06,
+      "loss": 1.5222,
+      "step": 4203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6030957385820756e-06,
+      "loss": 1.6289,
+      "step": 4204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.603000191094975e-06,
+      "loss": 1.3387,
+      "step": 4205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602904643607874e-06,
+      "loss": 1.5145,
+      "step": 4206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602809096120772e-06,
+      "loss": 1.165,
+      "step": 4207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602713548633671e-06,
+      "loss": 1.1419,
+      "step": 4208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602618001146571e-06,
+      "loss": 1.1871,
+      "step": 4209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602522453659469e-06,
+      "loss": 1.1855,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602426906172368e-06,
+      "loss": 1.7557,
+      "step": 4211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602331358685267e-06,
+      "loss": 1.3976,
+      "step": 4212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602235811198166e-06,
+      "loss": 1.3656,
+      "step": 4213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602140263711064e-06,
+      "loss": 1.1786,
+      "step": 4214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.602044716223963e-06,
+      "loss": 1.3859,
+      "step": 4215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.601949168736863e-06,
+      "loss": 1.1155,
+      "step": 4216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601853621249761e-06,
+      "loss": 1.3346,
+      "step": 4217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.60175807376266e-06,
+      "loss": 1.1636,
+      "step": 4218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6016625262755594e-06,
+      "loss": 1.5077,
+      "step": 4219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6015669787884585e-06,
+      "loss": 1.2647,
+      "step": 4220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601471431301357e-06,
+      "loss": 1.4884,
+      "step": 4221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601375883814256e-06,
+      "loss": 1.8422,
+      "step": 4222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601280336327156e-06,
+      "loss": 1.2596,
+      "step": 4223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601184788840054e-06,
+      "loss": 1.0659,
+      "step": 4224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.601089241352953e-06,
+      "loss": 1.5777,
+      "step": 4225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600993693865852e-06,
+      "loss": 1.3504,
+      "step": 4226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600898146378751e-06,
+      "loss": 1.4506,
+      "step": 4227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600802598891649e-06,
+      "loss": 1.4921,
+      "step": 4228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600707051404548e-06,
+      "loss": 1.534,
+      "step": 4229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600611503917447e-06,
+      "loss": 1.3365,
+      "step": 4230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600515956430346e-06,
+      "loss": 1.5942,
+      "step": 4231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600420408943245e-06,
+      "loss": 1.4372,
+      "step": 4232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600324861456144e-06,
+      "loss": 1.2209,
+      "step": 4233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.600229313969043e-06,
+      "loss": 1.1753,
+      "step": 4234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6001337664819415e-06,
+      "loss": 1.3094,
+      "step": 4235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6000382189948406e-06,
+      "loss": 1.2583,
+      "step": 4236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59994267150774e-06,
+      "loss": 1.6891,
+      "step": 4237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599847124020639e-06,
+      "loss": 1.2995,
+      "step": 4238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599751576533538e-06,
+      "loss": 1.0904,
+      "step": 4239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599656029046437e-06,
+      "loss": 0.9362,
+      "step": 4240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599560481559335e-06,
+      "loss": 1.5135,
+      "step": 4241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599464934072234e-06,
+      "loss": 1.2183,
+      "step": 4242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599369386585133e-06,
+      "loss": 1.5641,
+      "step": 4243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599273839098032e-06,
+      "loss": 1.1366,
+      "step": 4244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.599178291610931e-06,
+      "loss": 2.0562,
+      "step": 4245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59908274412383e-06,
+      "loss": 1.6295,
+      "step": 4246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598987196636729e-06,
+      "loss": 1.229,
+      "step": 4247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598891649149627e-06,
+      "loss": 1.4855,
+      "step": 4248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598796101662526e-06,
+      "loss": 1.1833,
+      "step": 4249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598700554175425e-06,
+      "loss": 1.2654,
+      "step": 4250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5986050066883244e-06,
+      "loss": 1.2931,
+      "step": 4251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5985094592012235e-06,
+      "loss": 1.6879,
+      "step": 4252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5984139117141225e-06,
+      "loss": 1.3758,
+      "step": 4253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5983183642270216e-06,
+      "loss": 1.6588,
+      "step": 4254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59822281673992e-06,
+      "loss": 1.9622,
+      "step": 4255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598127269252819e-06,
+      "loss": 1.3988,
+      "step": 4256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.598031721765718e-06,
+      "loss": 1.6728,
+      "step": 4257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597936174278617e-06,
+      "loss": 1.3713,
+      "step": 4258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597840626791516e-06,
+      "loss": 1.5722,
+      "step": 4259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597745079304415e-06,
+      "loss": 1.3125,
+      "step": 4260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597649531817314e-06,
+      "loss": 1.1834,
+      "step": 4261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597553984330212e-06,
+      "loss": 1.2915,
+      "step": 4262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597458436843111e-06,
+      "loss": 2.1887,
+      "step": 4263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59736288935601e-06,
+      "loss": 1.8042,
+      "step": 4264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597267341868909e-06,
+      "loss": 1.1958,
+      "step": 4265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597171794381808e-06,
+      "loss": 1.4073,
+      "step": 4266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597076246894707e-06,
+      "loss": 1.2392,
+      "step": 4267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.596980699407606e-06,
+      "loss": 1.5106,
+      "step": 4268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596885151920505e-06,
+      "loss": 1.7528,
+      "step": 4269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596789604433404e-06,
+      "loss": 0.9854,
+      "step": 4270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596694056946303e-06,
+      "loss": 1.1222,
+      "step": 4271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596598509459202e-06,
+      "loss": 1.4898,
+      "step": 4272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596502961972101e-06,
+      "loss": 1.3238,
+      "step": 4273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596407414485e-06,
+      "loss": 1.0348,
+      "step": 4274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596311866997899e-06,
+      "loss": 1.58,
+      "step": 4275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596216319510797e-06,
+      "loss": 1.1275,
+      "step": 4276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596120772023696e-06,
+      "loss": 1.0596,
+      "step": 4277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.596025224536595e-06,
+      "loss": 1.4823,
+      "step": 4278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595929677049494e-06,
+      "loss": 1.3037,
+      "step": 4279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595834129562392e-06,
+      "loss": 1.6302,
+      "step": 4280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595738582075292e-06,
+      "loss": 1.4217,
+      "step": 4281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595643034588191e-06,
+      "loss": 1.2185,
+      "step": 4282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5955474871010894e-06,
+      "loss": 1.4473,
+      "step": 4283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5954519396139885e-06,
+      "loss": 1.6864,
+      "step": 4284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5953563921268875e-06,
+      "loss": 1.4116,
+      "step": 4285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5952608446397866e-06,
+      "loss": 1.3381,
+      "step": 4286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595165297152685e-06,
+      "loss": 1.394,
+      "step": 4287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.595069749665585e-06,
+      "loss": 1.3248,
+      "step": 4288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594974202178483e-06,
+      "loss": 1.2971,
+      "step": 4289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594878654691382e-06,
+      "loss": 1.9978,
+      "step": 4290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594783107204281e-06,
+      "loss": 1.5122,
+      "step": 4291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.59468755971718e-06,
+      "loss": 1.4326,
+      "step": 4292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594592012230079e-06,
+      "loss": 1.2064,
+      "step": 4293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594496464742977e-06,
+      "loss": 1.0558,
+      "step": 4294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594400917255877e-06,
+      "loss": 1.0527,
+      "step": 4295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594305369768775e-06,
+      "loss": 0.8289,
+      "step": 4296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594209822281674e-06,
+      "loss": 1.1485,
+      "step": 4297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594114274794573e-06,
+      "loss": 1.2236,
+      "step": 4298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.594018727307472e-06,
+      "loss": 1.458,
+      "step": 4299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5939231798203706e-06,
+      "loss": 1.2758,
+      "step": 4300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.59382763233327e-06,
+      "loss": 1.6724,
+      "step": 4301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5937320848461695e-06,
+      "loss": 1.5771,
+      "step": 4302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593636537359068e-06,
+      "loss": 1.2569,
+      "step": 4303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593540989871967e-06,
+      "loss": 1.1143,
+      "step": 4304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593445442384866e-06,
+      "loss": 1.2242,
+      "step": 4305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593349894897765e-06,
+      "loss": 1.1283,
+      "step": 4306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593254347410663e-06,
+      "loss": 1.4637,
+      "step": 4307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593158799923562e-06,
+      "loss": 1.2335,
+      "step": 4308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.593063252436462e-06,
+      "loss": 1.9019,
+      "step": 4309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.59296770494936e-06,
+      "loss": 0.9817,
+      "step": 4310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592872157462259e-06,
+      "loss": 1.2628,
+      "step": 4311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592776609975158e-06,
+      "loss": 1.7295,
+      "step": 4312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592681062488057e-06,
+      "loss": 1.4763,
+      "step": 4313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592585515000955e-06,
+      "loss": 1.2576,
+      "step": 4314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5924899675138544e-06,
+      "loss": 1.3751,
+      "step": 4315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5923944200267535e-06,
+      "loss": 1.5093,
+      "step": 4316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5922988725396525e-06,
+      "loss": 1.6841,
+      "step": 4317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5922033250525516e-06,
+      "loss": 1.4322,
+      "step": 4318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592107777565451e-06,
+      "loss": 1.436,
+      "step": 4319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.59201223007835e-06,
+      "loss": 1.4133,
+      "step": 4320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.591916682591248e-06,
+      "loss": 1.311,
+      "step": 4321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591821135104147e-06,
+      "loss": 1.5067,
+      "step": 4322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591725587617046e-06,
+      "loss": 1.5631,
+      "step": 4323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591630040129945e-06,
+      "loss": 1.1515,
+      "step": 4324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591534492642844e-06,
+      "loss": 1.2538,
+      "step": 4325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591438945155743e-06,
+      "loss": 1.4599,
+      "step": 4326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591343397668642e-06,
+      "loss": 1.9677,
+      "step": 4327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.59124785018154e-06,
+      "loss": 1.1594,
+      "step": 4328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591152302694439e-06,
+      "loss": 1.7001,
+      "step": 4329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.591056755207338e-06,
+      "loss": 1.2874,
+      "step": 4330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590961207720237e-06,
+      "loss": 1.3748,
+      "step": 4331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590865660233136e-06,
+      "loss": 1.382,
+      "step": 4332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5907701127460354e-06,
+      "loss": 1.501,
+      "step": 4333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5906745652589345e-06,
+      "loss": 1.3788,
+      "step": 4334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590579017771833e-06,
+      "loss": 1.7644,
+      "step": 4335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590483470284732e-06,
+      "loss": 0.7666,
+      "step": 4336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590387922797631e-06,
+      "loss": 1.7036,
+      "step": 4337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.59029237531053e-06,
+      "loss": 0.8577,
+      "step": 4338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590196827823429e-06,
+      "loss": 1.1619,
+      "step": 4339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590101280336328e-06,
+      "loss": 1.4556,
+      "step": 4340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.590005732849226e-06,
+      "loss": 1.3198,
+      "step": 4341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589910185362125e-06,
+      "loss": 1.1038,
+      "step": 4342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589814637875024e-06,
+      "loss": 1.4046,
+      "step": 4343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589719090387923e-06,
+      "loss": 1.2592,
+      "step": 4344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589623542900822e-06,
+      "loss": 1.5143,
+      "step": 4345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589527995413721e-06,
+      "loss": 1.3308,
+      "step": 4346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.58943244792662e-06,
+      "loss": 1.2553,
+      "step": 4347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5893369004395185e-06,
+      "loss": 1.8042,
+      "step": 4348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5892413529524175e-06,
+      "loss": 1.9041,
+      "step": 4349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5891458054653166e-06,
+      "loss": 1.7111,
+      "step": 4350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.589050257978216e-06,
+      "loss": 1.2387,
+      "step": 4351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588954710491114e-06,
+      "loss": 1.1415,
+      "step": 4352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588859163004014e-06,
+      "loss": 1.526,
+      "step": 4353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588763615516913e-06,
+      "loss": 1.7792,
+      "step": 4354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588668068029811e-06,
+      "loss": 1.6938,
+      "step": 4355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.58857252054271e-06,
+      "loss": 1.8145,
+      "step": 4356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588476973055609e-06,
+      "loss": 1.2456,
+      "step": 4357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588381425568508e-06,
+      "loss": 1.2359,
+      "step": 4358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588285878081406e-06,
+      "loss": 1.1938,
+      "step": 4359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588190330594306e-06,
+      "loss": 1.2005,
+      "step": 4360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.588094783107205e-06,
+      "loss": 1.4371,
+      "step": 4361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587999235620103e-06,
+      "loss": 1.1802,
+      "step": 4362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587903688133002e-06,
+      "loss": 1.1139,
+      "step": 4363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587808140645901e-06,
+      "loss": 1.5769,
+      "step": 4364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5877125931588004e-06,
+      "loss": 1.3109,
+      "step": 4365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587617045671699e-06,
+      "loss": 1.7062,
+      "step": 4366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5875214981845985e-06,
+      "loss": 1.263,
+      "step": 4367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5874259506974976e-06,
+      "loss": 1.3702,
+      "step": 4368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587330403210396e-06,
+      "loss": 1.2298,
+      "step": 4369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587234855723295e-06,
+      "loss": 1.7634,
+      "step": 4370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587139308236194e-06,
+      "loss": 1.4618,
+      "step": 4371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587043760749093e-06,
+      "loss": 1.4724,
+      "step": 4372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.586948213261991e-06,
+      "loss": 0.9634,
+      "step": 4373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586852665774891e-06,
+      "loss": 0.8418,
+      "step": 4374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.58675711828779e-06,
+      "loss": 1.0008,
+      "step": 4375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586661570800688e-06,
+      "loss": 1.3992,
+      "step": 4376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586566023313587e-06,
+      "loss": 1.4871,
+      "step": 4377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586470475826486e-06,
+      "loss": 1.7624,
+      "step": 4378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586374928339385e-06,
+      "loss": 1.2957,
+      "step": 4379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5862793808522835e-06,
+      "loss": 1.2159,
+      "step": 4380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.586183833365183e-06,
+      "loss": 1.4068,
+      "step": 4381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5860882858780816e-06,
+      "loss": 1.0862,
+      "step": 4382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585992738390981e-06,
+      "loss": 0.9546,
+      "step": 4383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.58589719090388e-06,
+      "loss": 2.1265,
+      "step": 4384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585801643416779e-06,
+      "loss": 1.9058,
+      "step": 4385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585706095929678e-06,
+      "loss": 1.3206,
+      "step": 4386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585610548442576e-06,
+      "loss": 1.3854,
+      "step": 4387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585515000955475e-06,
+      "loss": 1.1734,
+      "step": 4388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585419453468374e-06,
+      "loss": 1.1318,
+      "step": 4389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585323905981273e-06,
+      "loss": 1.2762,
+      "step": 4390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585228358494172e-06,
+      "loss": 1.4187,
+      "step": 4391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585132811007071e-06,
+      "loss": 1.0917,
+      "step": 4392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.585037263519969e-06,
+      "loss": 0.9019,
+      "step": 4393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584941716032868e-06,
+      "loss": 1.1121,
+      "step": 4394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584846168545767e-06,
+      "loss": 1.6106,
+      "step": 4395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584750621058666e-06,
+      "loss": 1.131,
+      "step": 4396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5846550735715654e-06,
+      "loss": 1.2541,
+      "step": 4397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5845595260844645e-06,
+      "loss": 1.3403,
+      "step": 4398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5844639785973635e-06,
+      "loss": 1.458,
+      "step": 4399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584368431110262e-06,
+      "loss": 1.2575,
+      "step": 4400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584272883623161e-06,
+      "loss": 1.3233,
+      "step": 4401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.58417733613606e-06,
+      "loss": 1.229,
+      "step": 4402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.584081788648959e-06,
+      "loss": 1.402,
+      "step": 4403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583986241161858e-06,
+      "loss": 1.0037,
+      "step": 4404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583890693674757e-06,
+      "loss": 1.2148,
+      "step": 4405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583795146187656e-06,
+      "loss": 1.5367,
+      "step": 4406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583699598700554e-06,
+      "loss": 1.2047,
+      "step": 4407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583604051213453e-06,
+      "loss": 1.2755,
+      "step": 4408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583508503726352e-06,
+      "loss": 1.2176,
+      "step": 4409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583412956239251e-06,
+      "loss": 1.4852,
+      "step": 4410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.58331740875215e-06,
+      "loss": 1.2616,
+      "step": 4411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583221861265049e-06,
+      "loss": 1.2181,
+      "step": 4412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.583126313777948e-06,
+      "loss": 1.074,
+      "step": 4413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5830307662908466e-06,
+      "loss": 1.4219,
+      "step": 4414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582935218803746e-06,
+      "loss": 1.5473,
+      "step": 4415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582839671316645e-06,
+      "loss": 1.0871,
+      "step": 4416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582744123829544e-06,
+      "loss": 1.539,
+      "step": 4417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582648576342443e-06,
+      "loss": 1.2966,
+      "step": 4418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582553028855342e-06,
+      "loss": 1.4065,
+      "step": 4419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582457481368241e-06,
+      "loss": 1.5192,
+      "step": 4420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582361933881139e-06,
+      "loss": 1.1587,
+      "step": 4421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582266386394038e-06,
+      "loss": 1.1369,
+      "step": 4422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582170838906937e-06,
+      "loss": 1.5959,
+      "step": 4423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582075291419836e-06,
+      "loss": 1.4685,
+      "step": 4424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.581979743932735e-06,
+      "loss": 1.23,
+      "step": 4425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.581884196445634e-06,
+      "loss": 1.4391,
+      "step": 4426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.581788648958533e-06,
+      "loss": 1.2215,
+      "step": 4427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.581693101471431e-06,
+      "loss": 1.8983,
+      "step": 4428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5815975539843304e-06,
+      "loss": 1.5069,
+      "step": 4429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5815020064972295e-06,
+      "loss": 1.7229,
+      "step": 4430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5814064590101285e-06,
+      "loss": 1.8699,
+      "step": 4431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5813109115230276e-06,
+      "loss": 1.3946,
+      "step": 4432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.581215364035927e-06,
+      "loss": 1.2241,
+      "step": 4433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.581119816548825e-06,
+      "loss": 1.5012,
+      "step": 4434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.581024269061724e-06,
+      "loss": 1.5788,
+      "step": 4435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580928721574623e-06,
+      "loss": 1.376,
+      "step": 4436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580833174087522e-06,
+      "loss": 1.2536,
+      "step": 4437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580737626600421e-06,
+      "loss": 1.5936,
+      "step": 4438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.58064207911332e-06,
+      "loss": 1.3731,
+      "step": 4439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580546531626219e-06,
+      "loss": 1.4809,
+      "step": 4440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580450984139117e-06,
+      "loss": 1.4824,
+      "step": 4441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580355436652016e-06,
+      "loss": 1.3469,
+      "step": 4442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580259889164915e-06,
+      "loss": 1.0402,
+      "step": 4443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.580164341677814e-06,
+      "loss": 1.0951,
+      "step": 4444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5800687941907125e-06,
+      "loss": 1.3066,
+      "step": 4445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579973246703612e-06,
+      "loss": 1.6404,
+      "step": 4446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5798776992165114e-06,
+      "loss": 1.4227,
+      "step": 4447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.57978215172941e-06,
+      "loss": 1.0807,
+      "step": 4448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579686604242309e-06,
+      "loss": 1.5822,
+      "step": 4449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579591056755208e-06,
+      "loss": 1.2306,
+      "step": 4450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579495509268107e-06,
+      "loss": 1.2982,
+      "step": 4451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579399961781005e-06,
+      "loss": 1.0273,
+      "step": 4452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579304414293905e-06,
+      "loss": 1.2409,
+      "step": 4453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579208866806804e-06,
+      "loss": 1.0317,
+      "step": 4454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579113319319702e-06,
+      "loss": 1.165,
+      "step": 4455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.579017771832601e-06,
+      "loss": 1.5695,
+      "step": 4456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5789222243455e-06,
+      "loss": 1.4927,
+      "step": 4457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.578826676858399e-06,
+      "loss": 1.1537,
+      "step": 4458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.578731129371297e-06,
+      "loss": 1.3722,
+      "step": 4459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.578635581884196e-06,
+      "loss": 1.3742,
+      "step": 4460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.578540034397096e-06,
+      "loss": 1.0229,
+      "step": 4461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5784444869099945e-06,
+      "loss": 1.3614,
+      "step": 4462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5783489394228935e-06,
+      "loss": 1.3113,
+      "step": 4463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5782533919357926e-06,
+      "loss": 1.4086,
+      "step": 4464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.578157844448692e-06,
+      "loss": 0.9231,
+      "step": 4465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.57806229696159e-06,
+      "loss": 1.2775,
+      "step": 4466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577966749474489e-06,
+      "loss": 1.9307,
+      "step": 4467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577871201987389e-06,
+      "loss": 1.2548,
+      "step": 4468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577775654500287e-06,
+      "loss": 1.121,
+      "step": 4469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577680107013186e-06,
+      "loss": 1.8292,
+      "step": 4470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577584559526085e-06,
+      "loss": 0.9315,
+      "step": 4471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577489012038984e-06,
+      "loss": 1.2722,
+      "step": 4472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577393464551882e-06,
+      "loss": 1.4194,
+      "step": 4473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577297917064781e-06,
+      "loss": 1.5664,
+      "step": 4474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577202369577681e-06,
+      "loss": 1.5223,
+      "step": 4475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577106822090579e-06,
+      "loss": 1.533,
+      "step": 4476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577011274603478e-06,
+      "loss": 1.0724,
+      "step": 4477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.576915727116377e-06,
+      "loss": 1.4213,
+      "step": 4478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5768201796292764e-06,
+      "loss": 1.4569,
+      "step": 4479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576724632142175e-06,
+      "loss": 1.4093,
+      "step": 4480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576629084655074e-06,
+      "loss": 1.1807,
+      "step": 4481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576533537167973e-06,
+      "loss": 1.4633,
+      "step": 4482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576437989680872e-06,
+      "loss": 1.7858,
+      "step": 4483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576342442193771e-06,
+      "loss": 1.7299,
+      "step": 4484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57624689470667e-06,
+      "loss": 1.5556,
+      "step": 4485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576151347219569e-06,
+      "loss": 1.2484,
+      "step": 4486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.576055799732467e-06,
+      "loss": 1.2771,
+      "step": 4487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575960252245366e-06,
+      "loss": 1.4932,
+      "step": 4488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575864704758265e-06,
+      "loss": 0.9732,
+      "step": 4489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575769157271164e-06,
+      "loss": 1.7199,
+      "step": 4490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575673609784063e-06,
+      "loss": 1.7336,
+      "step": 4491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575578062296962e-06,
+      "loss": 1.4528,
+      "step": 4492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5754825148098604e-06,
+      "loss": 0.9709,
+      "step": 4493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5753869673227595e-06,
+      "loss": 1.4625,
+      "step": 4494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5752914198356585e-06,
+      "loss": 1.467,
+      "step": 4495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5751958723485576e-06,
+      "loss": 1.5138,
+      "step": 4496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575100324861457e-06,
+      "loss": 1.3578,
+      "step": 4497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.575004777374356e-06,
+      "loss": 0.9374,
+      "step": 4498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574909229887255e-06,
+      "loss": 1.7021,
+      "step": 4499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574813682400153e-06,
+      "loss": 1.2199,
+      "step": 4500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574718134913052e-06,
+      "loss": 1.3648,
+      "step": 4501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574622587425951e-06,
+      "loss": 1.4106,
+      "step": 4502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57452703993885e-06,
+      "loss": 1.543,
+      "step": 4503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574431492451749e-06,
+      "loss": 1.3642,
+      "step": 4504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574335944964648e-06,
+      "loss": 1.4311,
+      "step": 4505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574240397477547e-06,
+      "loss": 1.8555,
+      "step": 4506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574144849990445e-06,
+      "loss": 1.2159,
+      "step": 4507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574049302503344e-06,
+      "loss": 1.1994,
+      "step": 4508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573953755016243e-06,
+      "loss": 2.0527,
+      "step": 4509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573858207529142e-06,
+      "loss": 0.9074,
+      "step": 4510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5737626600420414e-06,
+      "loss": 1.5174,
+      "step": 4511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5736671125549405e-06,
+      "loss": 1.2155,
+      "step": 4512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5735715650678395e-06,
+      "loss": 1.788,
+      "step": 4513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573476017580738e-06,
+      "loss": 1.1126,
+      "step": 4514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573380470093637e-06,
+      "loss": 1.7378,
+      "step": 4515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573284922606536e-06,
+      "loss": 1.2496,
+      "step": 4516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573189375119435e-06,
+      "loss": 1.2345,
+      "step": 4517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.573093827632334e-06,
+      "loss": 1.7803,
+      "step": 4518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572998280145233e-06,
+      "loss": 1.4866,
+      "step": 4519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572902732658132e-06,
+      "loss": 1.1659,
+      "step": 4520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57280718517103e-06,
+      "loss": 1.2339,
+      "step": 4521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572711637683929e-06,
+      "loss": 1.714,
+      "step": 4522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572616090196828e-06,
+      "loss": 1.4659,
+      "step": 4523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572520542709727e-06,
+      "loss": 1.5633,
+      "step": 4524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572424995222626e-06,
+      "loss": 1.5847,
+      "step": 4525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572329447735525e-06,
+      "loss": 1.189,
+      "step": 4526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572233900248424e-06,
+      "loss": 0.9794,
+      "step": 4527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5721383527613226e-06,
+      "loss": 1.1692,
+      "step": 4528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.572042805274222e-06,
+      "loss": 1.5476,
+      "step": 4529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.571947257787121e-06,
+      "loss": 1.2322,
+      "step": 4530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57185171030002e-06,
+      "loss": 1.3856,
+      "step": 4531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571756162812918e-06,
+      "loss": 1.2099,
+      "step": 4532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571660615325818e-06,
+      "loss": 1.8071,
+      "step": 4533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571565067838716e-06,
+      "loss": 1.4742,
+      "step": 4534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571469520351615e-06,
+      "loss": 1.6934,
+      "step": 4535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571373972864514e-06,
+      "loss": 1.8158,
+      "step": 4536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571278425377413e-06,
+      "loss": 1.166,
+      "step": 4537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.571182877890312e-06,
+      "loss": 1.5712,
+      "step": 4538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57108733040321e-06,
+      "loss": 1.3497,
+      "step": 4539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57099178291611e-06,
+      "loss": 1.7693,
+      "step": 4540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570896235429008e-06,
+      "loss": 1.295,
+      "step": 4541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570800687941907e-06,
+      "loss": 1.1965,
+      "step": 4542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5707051404548064e-06,
+      "loss": 1.3835,
+      "step": 4543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5706095929677055e-06,
+      "loss": 1.4856,
+      "step": 4544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570514045480604e-06,
+      "loss": 1.5205,
+      "step": 4545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570418497993503e-06,
+      "loss": 1.5367,
+      "step": 4546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570322950506403e-06,
+      "loss": 1.4929,
+      "step": 4547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570227403019301e-06,
+      "loss": 1.4514,
+      "step": 4548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5701318555322e-06,
+      "loss": 1.5813,
+      "step": 4549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.570036308045099e-06,
+      "loss": 1.2237,
+      "step": 4550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569940760557998e-06,
+      "loss": 1.5876,
+      "step": 4551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569845213070896e-06,
+      "loss": 1.809,
+      "step": 4552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569749665583795e-06,
+      "loss": 1.2994,
+      "step": 4553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569654118096695e-06,
+      "loss": 1.4274,
+      "step": 4554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569558570609593e-06,
+      "loss": 1.395,
+      "step": 4555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569463023122492e-06,
+      "loss": 1.1715,
+      "step": 4556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569367475635391e-06,
+      "loss": 1.3844,
+      "step": 4557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56927192814829e-06,
+      "loss": 1.7559,
+      "step": 4558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5691763806611885e-06,
+      "loss": 1.09,
+      "step": 4559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5690808331740876e-06,
+      "loss": 1.2888,
+      "step": 4560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5689852856869874e-06,
+      "loss": 1.3723,
+      "step": 4561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568889738199886e-06,
+      "loss": 1.4475,
+      "step": 4562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568794190712785e-06,
+      "loss": 1.367,
+      "step": 4563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568698643225684e-06,
+      "loss": 1.2626,
+      "step": 4564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568603095738583e-06,
+      "loss": 1.2655,
+      "step": 4565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568507548251481e-06,
+      "loss": 1.5174,
+      "step": 4566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56841200076438e-06,
+      "loss": 1.3445,
+      "step": 4567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568316453277279e-06,
+      "loss": 1.0859,
+      "step": 4568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568220905790178e-06,
+      "loss": 1.5906,
+      "step": 4569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568125358303077e-06,
+      "loss": 1.5287,
+      "step": 4570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568029810815976e-06,
+      "loss": 1.1246,
+      "step": 4571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567934263328875e-06,
+      "loss": 1.3267,
+      "step": 4572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567838715841773e-06,
+      "loss": 1.1788,
+      "step": 4573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567743168354672e-06,
+      "loss": 1.0712,
+      "step": 4574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5676476208675714e-06,
+      "loss": 1.4457,
+      "step": 4575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5675520733804705e-06,
+      "loss": 1.2906,
+      "step": 4576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5674565258933695e-06,
+      "loss": 1.1384,
+      "step": 4577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5673609784062686e-06,
+      "loss": 1.1149,
+      "step": 4578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567265430919168e-06,
+      "loss": 1.3399,
+      "step": 4579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567169883432066e-06,
+      "loss": 1.3198,
+      "step": 4580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567074335944965e-06,
+      "loss": 1.6088,
+      "step": 4581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566978788457864e-06,
+      "loss": 1.1146,
+      "step": 4582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566883240970763e-06,
+      "loss": 1.196,
+      "step": 4583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566787693483662e-06,
+      "loss": 1.397,
+      "step": 4584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566692145996561e-06,
+      "loss": 1.2983,
+      "step": 4585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566596598509459e-06,
+      "loss": 1.2155,
+      "step": 4586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566501051022358e-06,
+      "loss": 1.2013,
+      "step": 4587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566405503535257e-06,
+      "loss": 1.3876,
+      "step": 4588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566309956048156e-06,
+      "loss": 1.493,
+      "step": 4589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566214408561055e-06,
+      "loss": 1.7862,
+      "step": 4590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566118861073954e-06,
+      "loss": 1.3112,
+      "step": 4591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.566023313586853e-06,
+      "loss": 1.4581,
+      "step": 4592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565927766099752e-06,
+      "loss": 1.185,
+      "step": 4593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565832218612651e-06,
+      "loss": 0.7684,
+      "step": 4594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.56573667112555e-06,
+      "loss": 1.3701,
+      "step": 4595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565641123638449e-06,
+      "loss": 1.27,
+      "step": 4596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565545576151348e-06,
+      "loss": 1.4,
+      "step": 4597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565450028664247e-06,
+      "loss": 1.4366,
+      "step": 4598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565354481177146e-06,
+      "loss": 1.8024,
+      "step": 4599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565258933690044e-06,
+      "loss": 1.3053,
+      "step": 4600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565163386202943e-06,
+      "loss": 1.1576,
+      "step": 4601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.565067838715842e-06,
+      "loss": 1.2121,
+      "step": 4602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564972291228741e-06,
+      "loss": 1.3789,
+      "step": 4603
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564876743741639e-06,
+      "loss": 1.4697,
+      "step": 4604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564781196254539e-06,
+      "loss": 1.0539,
+      "step": 4605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564685648767438e-06,
+      "loss": 1.8278,
+      "step": 4606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5645901012803364e-06,
+      "loss": 1.3188,
+      "step": 4607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5644945537932355e-06,
+      "loss": 1.1387,
+      "step": 4608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5643990063061345e-06,
+      "loss": 1.3656,
+      "step": 4609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5643034588190336e-06,
+      "loss": 1.1492,
+      "step": 4610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564207911331932e-06,
+      "loss": 1.1809,
+      "step": 4611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564112363844832e-06,
+      "loss": 1.5609,
+      "step": 4612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.564016816357731e-06,
+      "loss": 1.3432,
+      "step": 4613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563921268870629e-06,
+      "loss": 1.2725,
+      "step": 4614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563825721383528e-06,
+      "loss": 1.3137,
+      "step": 4615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563730173896427e-06,
+      "loss": 1.4464,
+      "step": 4616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563634626409326e-06,
+      "loss": 1.1616,
+      "step": 4617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563539078922224e-06,
+      "loss": 1.6368,
+      "step": 4618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563443531435124e-06,
+      "loss": 1.1847,
+      "step": 4619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563347983948023e-06,
+      "loss": 1.8563,
+      "step": 4620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563252436460921e-06,
+      "loss": 1.7018,
+      "step": 4621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.56315688897382e-06,
+      "loss": 1.1947,
+      "step": 4622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.563061341486719e-06,
+      "loss": 0.8459,
+      "step": 4623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562965793999618e-06,
+      "loss": 1.8536,
+      "step": 4624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562870246512517e-06,
+      "loss": 1.3243,
+      "step": 4625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5627746990254165e-06,
+      "loss": 1.1823,
+      "step": 4626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562679151538315e-06,
+      "loss": 1.8071,
+      "step": 4627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562583604051214e-06,
+      "loss": 1.3346,
+      "step": 4628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562488056564113e-06,
+      "loss": 1.2436,
+      "step": 4629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562392509077012e-06,
+      "loss": 1.557,
+      "step": 4630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562296961589911e-06,
+      "loss": 1.1846,
+      "step": 4631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562201414102809e-06,
+      "loss": 1.1548,
+      "step": 4632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562105866615709e-06,
+      "loss": 1.9535,
+      "step": 4633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562010319128607e-06,
+      "loss": 1.6123,
+      "step": 4634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.561914771641506e-06,
+      "loss": 1.4083,
+      "step": 4635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561819224154405e-06,
+      "loss": 1.45,
+      "step": 4636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561723676667304e-06,
+      "loss": 1.4148,
+      "step": 4637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561628129180202e-06,
+      "loss": 1.0861,
+      "step": 4638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5615325816931014e-06,
+      "loss": 1.2948,
+      "step": 4639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5614370342060005e-06,
+      "loss": 1.3828,
+      "step": 4640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5613414867188995e-06,
+      "loss": 1.8328,
+      "step": 4641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5612459392317985e-06,
+      "loss": 1.0731,
+      "step": 4642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561150391744698e-06,
+      "loss": 1.1245,
+      "step": 4643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.561054844257597e-06,
+      "loss": 1.3771,
+      "step": 4644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560959296770495e-06,
+      "loss": 1.0422,
+      "step": 4645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560863749283394e-06,
+      "loss": 1.0142,
+      "step": 4646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560768201796293e-06,
+      "loss": 1.2616,
+      "step": 4647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560672654309192e-06,
+      "loss": 1.6476,
+      "step": 4648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560577106822091e-06,
+      "loss": 1.7386,
+      "step": 4649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.56048155933499e-06,
+      "loss": 1.4378,
+      "step": 4650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560386011847889e-06,
+      "loss": 1.1786,
+      "step": 4651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560290464360787e-06,
+      "loss": 1.9143,
+      "step": 4652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560194916873686e-06,
+      "loss": 1.3065,
+      "step": 4653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560099369386585e-06,
+      "loss": 1.4219,
+      "step": 4654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.560003821899484e-06,
+      "loss": 0.9514,
+      "step": 4655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559908274412383e-06,
+      "loss": 1.6374,
+      "step": 4656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5598127269252824e-06,
+      "loss": 1.6428,
+      "step": 4657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5597171794381815e-06,
+      "loss": 1.3303,
+      "step": 4658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55962163195108e-06,
+      "loss": 1.2831,
+      "step": 4659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559526084463979e-06,
+      "loss": 1.3548,
+      "step": 4660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559430536976878e-06,
+      "loss": 0.9938,
+      "step": 4661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559334989489777e-06,
+      "loss": 1.5892,
+      "step": 4662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559239442002676e-06,
+      "loss": 1.6772,
+      "step": 4663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559143894515575e-06,
+      "loss": 1.6309,
+      "step": 4664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.559048347028474e-06,
+      "loss": 1.2737,
+      "step": 4665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558952799541372e-06,
+      "loss": 1.3814,
+      "step": 4666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558857252054271e-06,
+      "loss": 1.4617,
+      "step": 4667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55876170456717e-06,
+      "loss": 1.4131,
+      "step": 4668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558666157080069e-06,
+      "loss": 1.422,
+      "step": 4669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558570609592968e-06,
+      "loss": 1.5994,
+      "step": 4670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558475062105867e-06,
+      "loss": 1.6734,
+      "step": 4671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558379514618766e-06,
+      "loss": 1.1615,
+      "step": 4672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5582839671316645e-06,
+      "loss": 1.4552,
+      "step": 4673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5581884196445635e-06,
+      "loss": 1.3254,
+      "step": 4674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.558092872157463e-06,
+      "loss": 1.0693,
+      "step": 4675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557997324670362e-06,
+      "loss": 1.0225,
+      "step": 4676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557901777183261e-06,
+      "loss": 1.4198,
+      "step": 4677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55780622969616e-06,
+      "loss": 1.0059,
+      "step": 4678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557710682209059e-06,
+      "loss": 1.3489,
+      "step": 4679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557615134721957e-06,
+      "loss": 1.3906,
+      "step": 4680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557519587234856e-06,
+      "loss": 1.1904,
+      "step": 4681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557424039747755e-06,
+      "loss": 0.8597,
+      "step": 4682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557328492260654e-06,
+      "loss": 1.3814,
+      "step": 4683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557232944773553e-06,
+      "loss": 1.0455,
+      "step": 4684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.557137397286452e-06,
+      "loss": 1.1554,
+      "step": 4685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55704184979935e-06,
+      "loss": 1.6416,
+      "step": 4686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556946302312249e-06,
+      "loss": 1.2052,
+      "step": 4687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556850754825148e-06,
+      "loss": 1.434,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5567552073380474e-06,
+      "loss": 1.8569,
+      "step": 4689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5566596598509465e-06,
+      "loss": 1.1539,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5565641123638455e-06,
+      "loss": 1.3652,
+      "step": 4691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5564685648767445e-06,
+      "loss": 1.6077,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.556373017389643e-06,
+      "loss": 1.1913,
+      "step": 4693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.556277469902542e-06,
+      "loss": 1.0462,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.556181922415441e-06,
+      "loss": 1.4759,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.55608637492834e-06,
+      "loss": 1.3702,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555990827441238e-06,
+      "loss": 0.8573,
+      "step": 4697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555895279954138e-06,
+      "loss": 1.0031,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555799732467037e-06,
+      "loss": 1.3887,
+      "step": 4699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555704184979935e-06,
+      "loss": 1.7294,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555608637492834e-06,
+      "loss": 1.382,
+      "step": 4701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555513090005733e-06,
+      "loss": 1.2544,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555417542518632e-06,
+      "loss": 1.612,
+      "step": 4703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5553219950315305e-06,
+      "loss": 1.5226,
+      "step": 4704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.55522644754443e-06,
+      "loss": 1.43,
+      "step": 4705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555130900057329e-06,
+      "loss": 1.2471,
+      "step": 4706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.555035352570228e-06,
+      "loss": 1.1813,
+      "step": 4707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554939805083127e-06,
+      "loss": 1.6612,
+      "step": 4708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554844257596026e-06,
+      "loss": 1.4157,
+      "step": 4709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554748710108925e-06,
+      "loss": 1.4404,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554653162621823e-06,
+      "loss": 1.4635,
+      "step": 4711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554557615134722e-06,
+      "loss": 1.3924,
+      "step": 4712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554462067647622e-06,
+      "loss": 1.779,
+      "step": 4713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.55436652016052e-06,
+      "loss": 1.2405,
+      "step": 4714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554270972673419e-06,
+      "loss": 1.2072,
+      "step": 4715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554175425186318e-06,
+      "loss": 0.9826,
+      "step": 4716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554079877699217e-06,
+      "loss": 1.1649,
+      "step": 4717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553984330212115e-06,
+      "loss": 1.2789,
+      "step": 4718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553888782725014e-06,
+      "loss": 1.3235,
+      "step": 4719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553793235237914e-06,
+      "loss": 1.3376,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5536976877508124e-06,
+      "loss": 0.9224,
+      "step": 4721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5536021402637115e-06,
+      "loss": 1.1815,
+      "step": 4722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5535065927766105e-06,
+      "loss": 1.4493,
+      "step": 4723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5534110452895095e-06,
+      "loss": 1.5605,
+      "step": 4724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553315497802408e-06,
+      "loss": 1.1793,
+      "step": 4725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553219950315307e-06,
+      "loss": 1.1722,
+      "step": 4726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553124402828206e-06,
+      "loss": 1.0153,
+      "step": 4727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.553028855341105e-06,
+      "loss": 1.4672,
+      "step": 4728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552933307854004e-06,
+      "loss": 1.5647,
+      "step": 4729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552837760366903e-06,
+      "loss": 1.1512,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552742212879802e-06,
+      "loss": 1.7483,
+      "step": 4731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5526466653927e-06,
+      "loss": 1.3961,
+      "step": 4732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552551117905599e-06,
+      "loss": 1.2794,
+      "step": 4733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552455570418498e-06,
+      "loss": 1.0853,
+      "step": 4734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552360022931397e-06,
+      "loss": 1.4221,
+      "step": 4735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552264475444296e-06,
+      "loss": 1.0598,
+      "step": 4736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552168927957195e-06,
+      "loss": 1.3631,
+      "step": 4737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5520733804700935e-06,
+      "loss": 1.5032,
+      "step": 4738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.551977832982993e-06,
+      "loss": 1.3788,
+      "step": 4739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.551882285495892e-06,
+      "loss": 1.5533,
+      "step": 4740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551786738008791e-06,
+      "loss": 1.5644,
+      "step": 4741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55169119052169e-06,
+      "loss": 1.4784,
+      "step": 4742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551595643034589e-06,
+      "loss": 1.6092,
+      "step": 4743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551500095547488e-06,
+      "loss": 1.3781,
+      "step": 4744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551404548060386e-06,
+      "loss": 1.6979,
+      "step": 4745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551309000573285e-06,
+      "loss": 1.2883,
+      "step": 4746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551213453086184e-06,
+      "loss": 1.3507,
+      "step": 4747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551117905599083e-06,
+      "loss": 1.216,
+      "step": 4748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.551022358111982e-06,
+      "loss": 0.8727,
+      "step": 4749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550926810624881e-06,
+      "loss": 1.3535,
+      "step": 4750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55083126313778e-06,
+      "loss": 1.131,
+      "step": 4751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550735715650678e-06,
+      "loss": 1.462,
+      "step": 4752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5506401681635774e-06,
+      "loss": 1.6436,
+      "step": 4753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5505446206764765e-06,
+      "loss": 1.7759,
+      "step": 4754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5504490731893755e-06,
+      "loss": 1.057,
+      "step": 4755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5503535257022745e-06,
+      "loss": 1.4083,
+      "step": 4756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550257978215174e-06,
+      "loss": 1.6124,
+      "step": 4757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550162430728073e-06,
+      "loss": 1.2948,
+      "step": 4758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550066883240971e-06,
+      "loss": 1.0019,
+      "step": 4759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54997133575387e-06,
+      "loss": 1.0754,
+      "step": 4760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549875788266769e-06,
+      "loss": 1.4557,
+      "step": 4761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549780240779668e-06,
+      "loss": 1.0488,
+      "step": 4762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549684693292567e-06,
+      "loss": 1.2444,
+      "step": 4763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549589145805466e-06,
+      "loss": 1.606,
+      "step": 4764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549493598318365e-06,
+      "loss": 0.9346,
+      "step": 4765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549398050831263e-06,
+      "loss": 1.4438,
+      "step": 4766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549302503344162e-06,
+      "loss": 1.4641,
+      "step": 4767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549206955857061e-06,
+      "loss": 1.0249,
+      "step": 4768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54911140836996e-06,
+      "loss": 1.304,
+      "step": 4769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549015860882859e-06,
+      "loss": 1.6964,
+      "step": 4770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5489203133957584e-06,
+      "loss": 1.474,
+      "step": 4771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5488247659086575e-06,
+      "loss": 1.258,
+      "step": 4772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548729218421556e-06,
+      "loss": 1.3058,
+      "step": 4773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548633670934455e-06,
+      "loss": 1.1276,
+      "step": 4774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548538123447354e-06,
+      "loss": 1.5592,
+      "step": 4775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548442575960253e-06,
+      "loss": 1.5787,
+      "step": 4776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548347028473152e-06,
+      "loss": 1.31,
+      "step": 4777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548251480986051e-06,
+      "loss": 1.3446,
+      "step": 4778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548155933498949e-06,
+      "loss": 1.3226,
+      "step": 4779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548060386011848e-06,
+      "loss": 1.022,
+      "step": 4780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547964838524747e-06,
+      "loss": 1.5717,
+      "step": 4781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547869291037646e-06,
+      "loss": 1.2644,
+      "step": 4782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547773743550545e-06,
+      "loss": 1.2764,
+      "step": 4783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547678196063443e-06,
+      "loss": 1.2102,
+      "step": 4784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547582648576343e-06,
+      "loss": 1.2104,
+      "step": 4785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5474871010892415e-06,
+      "loss": 1.5525,
+      "step": 4786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5473915536021405e-06,
+      "loss": 1.5685,
+      "step": 4787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5472960061150395e-06,
+      "loss": 1.4194,
+      "step": 4788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547200458627939e-06,
+      "loss": 1.4801,
+      "step": 4789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547104911140837e-06,
+      "loss": 2.0466,
+      "step": 4790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.547009363653736e-06,
+      "loss": 1.3242,
+      "step": 4791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546913816166636e-06,
+      "loss": 1.2682,
+      "step": 4792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546818268679534e-06,
+      "loss": 1.0424,
+      "step": 4793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546722721192433e-06,
+      "loss": 1.0867,
+      "step": 4794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546627173705332e-06,
+      "loss": 0.8685,
+      "step": 4795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546531626218231e-06,
+      "loss": 1.3417,
+      "step": 4796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546436078731129e-06,
+      "loss": 0.833,
+      "step": 4797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546340531244028e-06,
+      "loss": 1.9166,
+      "step": 4798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546244983756928e-06,
+      "loss": 1.1726,
+      "step": 4799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546149436269826e-06,
+      "loss": 1.641,
+      "step": 4800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.546053888782725e-06,
+      "loss": 1.0073,
+      "step": 4801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545958341295624e-06,
+      "loss": 1.6434,
+      "step": 4802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5458627938085234e-06,
+      "loss": 1.2814,
+      "step": 4803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545767246321422e-06,
+      "loss": 1.723,
+      "step": 4804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545671698834321e-06,
+      "loss": 1.5541,
+      "step": 4805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5455761513472205e-06,
+      "loss": 1.2656,
+      "step": 4806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545480603860119e-06,
+      "loss": 1.6373,
+      "step": 4807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545385056373018e-06,
+      "loss": 1.079,
+      "step": 4808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545289508885917e-06,
+      "loss": 0.9192,
+      "step": 4809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545193961398816e-06,
+      "loss": 1.7372,
+      "step": 4810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545098413911714e-06,
+      "loss": 1.6721,
+      "step": 4811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.545002866424613e-06,
+      "loss": 1.1693,
+      "step": 4812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544907318937513e-06,
+      "loss": 0.9325,
+      "step": 4813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544811771450411e-06,
+      "loss": 1.263,
+      "step": 4814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.54471622396331e-06,
+      "loss": 1.1971,
+      "step": 4815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544620676476209e-06,
+      "loss": 1.1442,
+      "step": 4816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544525128989108e-06,
+      "loss": 1.7635,
+      "step": 4817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5444295815020065e-06,
+      "loss": 1.3201,
+      "step": 4818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5443340340149055e-06,
+      "loss": 1.3796,
+      "step": 4819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5442384865278045e-06,
+      "loss": 0.9255,
+      "step": 4820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544142939040704e-06,
+      "loss": 1.0435,
+      "step": 4821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.544047391553603e-06,
+      "loss": 0.8565,
+      "step": 4822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543951844066502e-06,
+      "loss": 1.6736,
+      "step": 4823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543856296579401e-06,
+      "loss": 1.5045,
+      "step": 4824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543760749092299e-06,
+      "loss": 1.7178,
+      "step": 4825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543665201605198e-06,
+      "loss": 1.2704,
+      "step": 4826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543569654118097e-06,
+      "loss": 1.6061,
+      "step": 4827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543474106630996e-06,
+      "loss": 1.5823,
+      "step": 4828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543378559143895e-06,
+      "loss": 1.1871,
+      "step": 4829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543283011656794e-06,
+      "loss": 1.0595,
+      "step": 4830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543187464169692e-06,
+      "loss": 1.4827,
+      "step": 4831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.543091916682591e-06,
+      "loss": 1.1364,
+      "step": 4832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.54299636919549e-06,
+      "loss": 1.5883,
+      "step": 4833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542900821708389e-06,
+      "loss": 1.7675,
+      "step": 4834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5428052742212884e-06,
+      "loss": 1.0427,
+      "step": 4835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5427097267341875e-06,
+      "loss": 1.4667,
+      "step": 4836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5426141792470865e-06,
+      "loss": 1.2777,
+      "step": 4837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542518631759985e-06,
+      "loss": 1.0087,
+      "step": 4838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542423084272884e-06,
+      "loss": 1.8522,
+      "step": 4839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542327536785783e-06,
+      "loss": 1.2052,
+      "step": 4840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542231989298682e-06,
+      "loss": 0.6855,
+      "step": 4841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542136441811581e-06,
+      "loss": 2.0302,
+      "step": 4842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.54204089432448e-06,
+      "loss": 1.4417,
+      "step": 4843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.541945346837379e-06,
+      "loss": 1.5752,
+      "step": 4844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.541849799350277e-06,
+      "loss": 1.0563,
+      "step": 4845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541754251863176e-06,
+      "loss": 1.1646,
+      "step": 4846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541658704376075e-06,
+      "loss": 1.3481,
+      "step": 4847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541563156888974e-06,
+      "loss": 1.55,
+      "step": 4848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541467609401873e-06,
+      "loss": 1.5239,
+      "step": 4849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541372061914772e-06,
+      "loss": 0.8391,
+      "step": 4850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541276514427671e-06,
+      "loss": 1.2854,
+      "step": 4851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5411809669405695e-06,
+      "loss": 1.1134,
+      "step": 4852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.541085419453469e-06,
+      "loss": 1.1382,
+      "step": 4853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540989871966368e-06,
+      "loss": 1.4977,
+      "step": 4854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540894324479267e-06,
+      "loss": 1.377,
+      "step": 4855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540798776992165e-06,
+      "loss": 1.499,
+      "step": 4856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540703229505065e-06,
+      "loss": 1.5473,
+      "step": 4857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540607682017964e-06,
+      "loss": 1.3587,
+      "step": 4858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540512134530862e-06,
+      "loss": 1.1742,
+      "step": 4859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540416587043761e-06,
+      "loss": 1.3463,
+      "step": 4860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.54032103955666e-06,
+      "loss": 1.1616,
+      "step": 4861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540225492069559e-06,
+      "loss": 1.5812,
+      "step": 4862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540129944582457e-06,
+      "loss": 1.3159,
+      "step": 4863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.540034397095357e-06,
+      "loss": 1.1817,
+      "step": 4864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.539938849608256e-06,
+      "loss": 1.5985,
+      "step": 4865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.539843302121154e-06,
+      "loss": 1.1566,
+      "step": 4866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5397477546340534e-06,
+      "loss": 1.6631,
+      "step": 4867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5396522071469525e-06,
+      "loss": 1.6667,
+      "step": 4868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5395566596598515e-06,
+      "loss": 1.4705,
+      "step": 4869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.53946111217275e-06,
+      "loss": 1.5132,
+      "step": 4870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.53936556468565e-06,
+      "loss": 1.3544,
+      "step": 4871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.539270017198549e-06,
+      "loss": 1.9697,
+      "step": 4872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.539174469711447e-06,
+      "loss": 0.9591,
+      "step": 4873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.539078922224346e-06,
+      "loss": 1.4822,
+      "step": 4874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538983374737245e-06,
+      "loss": 1.6889,
+      "step": 4875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538887827250144e-06,
+      "loss": 1.2651,
+      "step": 4876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538792279763042e-06,
+      "loss": 1.359,
+      "step": 4877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538696732275942e-06,
+      "loss": 1.5152,
+      "step": 4878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.53860118478884e-06,
+      "loss": 0.9169,
+      "step": 4879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538505637301739e-06,
+      "loss": 1.5604,
+      "step": 4880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538410089814638e-06,
+      "loss": 1.0742,
+      "step": 4881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538314542327537e-06,
+      "loss": 0.9607,
+      "step": 4882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.538218994840436e-06,
+      "loss": 1.5523,
+      "step": 4883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5381234473533345e-06,
+      "loss": 1.4617,
+      "step": 4884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5380278998662344e-06,
+      "loss": 1.4674,
+      "step": 4885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537932352379133e-06,
+      "loss": 0.821,
+      "step": 4886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537836804892032e-06,
+      "loss": 0.9426,
+      "step": 4887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537741257404931e-06,
+      "loss": 1.1332,
+      "step": 4888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.53764570991783e-06,
+      "loss": 1.98,
+      "step": 4889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537550162430728e-06,
+      "loss": 1.6686,
+      "step": 4890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537454614943627e-06,
+      "loss": 1.2281,
+      "step": 4891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537359067456526e-06,
+      "loss": 1.5286,
+      "step": 4892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537263519969425e-06,
+      "loss": 1.4212,
+      "step": 4893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537167972482324e-06,
+      "loss": 1.4337,
+      "step": 4894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.537072424995223e-06,
+      "loss": 1.4829,
+      "step": 4895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.536976877508122e-06,
+      "loss": 1.1929,
+      "step": 4896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.53688133002102e-06,
+      "loss": 1.4135,
+      "step": 4897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.536785782533919e-06,
+      "loss": 1.5401,
+      "step": 4898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5366902350468184e-06,
+      "loss": 1.6563,
+      "step": 4899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5365946875597175e-06,
+      "loss": 1.0199,
+      "step": 4900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5364991400726165e-06,
+      "loss": 1.1615,
+      "step": 4901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5364035925855155e-06,
+      "loss": 1.1211,
+      "step": 4902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.536308045098415e-06,
+      "loss": 1.388,
+      "step": 4903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.536212497611313e-06,
+      "loss": 1.5638,
+      "step": 4904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.536116950124212e-06,
+      "loss": 1.2438,
+      "step": 4905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.536021402637111e-06,
+      "loss": 1.1565,
+      "step": 4906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53592585515001e-06,
+      "loss": 1.5526,
+      "step": 4907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535830307662909e-06,
+      "loss": 1.3921,
+      "step": 4908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535734760175808e-06,
+      "loss": 1.5231,
+      "step": 4909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535639212688707e-06,
+      "loss": 1.1774,
+      "step": 4910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535543665201605e-06,
+      "loss": 1.3029,
+      "step": 4911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535448117714504e-06,
+      "loss": 1.3125,
+      "step": 4912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535352570227403e-06,
+      "loss": 1.4433,
+      "step": 4913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535257022740302e-06,
+      "loss": 1.4339,
+      "step": 4914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.535161475253201e-06,
+      "loss": 1.1425,
+      "step": 4915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5350659277661e-06,
+      "loss": 1.7751,
+      "step": 4916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5349703802789994e-06,
+      "loss": 1.0637,
+      "step": 4917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534874832791898e-06,
+      "loss": 1.2279,
+      "step": 4918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534779285304797e-06,
+      "loss": 1.4729,
+      "step": 4919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534683737817696e-06,
+      "loss": 1.5546,
+      "step": 4920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534588190330595e-06,
+      "loss": 1.0431,
+      "step": 4921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534492642843494e-06,
+      "loss": 1.3839,
+      "step": 4922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534397095356393e-06,
+      "loss": 1.8365,
+      "step": 4923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534301547869292e-06,
+      "loss": 1.4485,
+      "step": 4924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53420600038219e-06,
+      "loss": 1.0762,
+      "step": 4925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534110452895089e-06,
+      "loss": 1.3116,
+      "step": 4926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.534014905407988e-06,
+      "loss": 1.5881,
+      "step": 4927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533919357920887e-06,
+      "loss": 1.5487,
+      "step": 4928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533823810433786e-06,
+      "loss": 1.3629,
+      "step": 4929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533728262946685e-06,
+      "loss": 1.3656,
+      "step": 4930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5336327154595834e-06,
+      "loss": 1.2528,
+      "step": 4931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5335371679724825e-06,
+      "loss": 0.9553,
+      "step": 4932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5334416204853815e-06,
+      "loss": 1.7872,
+      "step": 4933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5333460729982805e-06,
+      "loss": 1.1851,
+      "step": 4934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53325052551118e-06,
+      "loss": 1.8658,
+      "step": 4935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533154978024079e-06,
+      "loss": 1.5723,
+      "step": 4936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.533059430536978e-06,
+      "loss": 1.1355,
+      "step": 4937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532963883049876e-06,
+      "loss": 1.4078,
+      "step": 4938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532868335562775e-06,
+      "loss": 1.9153,
+      "step": 4939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532772788075674e-06,
+      "loss": 1.5234,
+      "step": 4940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532677240588573e-06,
+      "loss": 1.2135,
+      "step": 4941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532581693101471e-06,
+      "loss": 1.3079,
+      "step": 4942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532486145614371e-06,
+      "loss": 1.1871,
+      "step": 4943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.53239059812727e-06,
+      "loss": 1.1099,
+      "step": 4944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532295050640168e-06,
+      "loss": 1.74,
+      "step": 4945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532199503153067e-06,
+      "loss": 1.147,
+      "step": 4946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532103955665966e-06,
+      "loss": 1.496,
+      "step": 4947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532008408178865e-06,
+      "loss": 1.1946,
+      "step": 4948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.531912860691764e-06,
+      "loss": 1.4976,
+      "step": 4949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5318173132046635e-06,
+      "loss": 1.4326,
+      "step": 4950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5317217657175625e-06,
+      "loss": 1.2597,
+      "step": 4951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531626218230461e-06,
+      "loss": 1.6288,
+      "step": 4952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.53153067074336e-06,
+      "loss": 1.2244,
+      "step": 4953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531435123256259e-06,
+      "loss": 1.003,
+      "step": 4954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531339575769158e-06,
+      "loss": 1.6672,
+      "step": 4955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531244028282056e-06,
+      "loss": 1.6301,
+      "step": 4956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531148480794956e-06,
+      "loss": 1.7417,
+      "step": 4957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.531052933307855e-06,
+      "loss": 1.9139,
+      "step": 4958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530957385820753e-06,
+      "loss": 1.8038,
+      "step": 4959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530861838333652e-06,
+      "loss": 1.4024,
+      "step": 4960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530766290846551e-06,
+      "loss": 0.85,
+      "step": 4961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.53067074335945e-06,
+      "loss": 1.4921,
+      "step": 4962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5305751958723484e-06,
+      "loss": 1.5758,
+      "step": 4963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5304796483852475e-06,
+      "loss": 1.2178,
+      "step": 4964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530384100898147e-06,
+      "loss": 1.1123,
+      "step": 4965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5302885534110455e-06,
+      "loss": 0.9552,
+      "step": 4966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530193005923945e-06,
+      "loss": 1.8008,
+      "step": 4967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530097458436844e-06,
+      "loss": 1.1443,
+      "step": 4968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.530001910949743e-06,
+      "loss": 0.9123,
+      "step": 4969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529906363462641e-06,
+      "loss": 1.2662,
+      "step": 4970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52981081597554e-06,
+      "loss": 1.1543,
+      "step": 4971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529715268488439e-06,
+      "loss": 1.5207,
+      "step": 4972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529619721001338e-06,
+      "loss": 1.1062,
+      "step": 4973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529524173514237e-06,
+      "loss": 1.1194,
+      "step": 4974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529428626027136e-06,
+      "loss": 1.3791,
+      "step": 4975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529333078540035e-06,
+      "loss": 1.4086,
+      "step": 4976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529237531052933e-06,
+      "loss": 1.1782,
+      "step": 4977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529141983565832e-06,
+      "loss": 1.4124,
+      "step": 4978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.529046436078731e-06,
+      "loss": 1.5118,
+      "step": 4979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52895088859163e-06,
+      "loss": 1.4661,
+      "step": 4980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5288553411045294e-06,
+      "loss": 1.6639,
+      "step": 4981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5287597936174285e-06,
+      "loss": 1.4234,
+      "step": 4982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528664246130327e-06,
+      "loss": 1.3698,
+      "step": 4983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528568698643226e-06,
+      "loss": 1.7281,
+      "step": 4984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528473151156125e-06,
+      "loss": 1.5125,
+      "step": 4985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528377603669024e-06,
+      "loss": 1.0713,
+      "step": 4986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528282056181923e-06,
+      "loss": 1.6444,
+      "step": 4987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528186508694822e-06,
+      "loss": 1.7873,
+      "step": 4988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.528090961207721e-06,
+      "loss": 1.2952,
+      "step": 4989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527995413720619e-06,
+      "loss": 1.4133,
+      "step": 4990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527899866233518e-06,
+      "loss": 0.9886,
+      "step": 4991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527804318746417e-06,
+      "loss": 1.1801,
+      "step": 4992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527708771259316e-06,
+      "loss": 1.1423,
+      "step": 4993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527613223772215e-06,
+      "loss": 0.995,
+      "step": 4994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527517676285114e-06,
+      "loss": 1.5631,
+      "step": 4995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527422128798013e-06,
+      "loss": 1.1934,
+      "step": 4996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5273265813109115e-06,
+      "loss": 1.1961,
+      "step": 4997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5272310338238105e-06,
+      "loss": 1.2862,
+      "step": 4998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.52713548633671e-06,
+      "loss": 1.208,
+      "step": 4999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.527039938849609e-06,
+      "loss": 1.3505,
+      "step": 5000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.526944391362508e-06,
+      "loss": 1.3626,
+      "step": 5001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.526848843875407e-06,
+      "loss": 1.5035,
+      "step": 5002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526753296388306e-06,
+      "loss": 1.3499,
+      "step": 5003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526657748901204e-06,
+      "loss": 1.1831,
+      "step": 5004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526562201414103e-06,
+      "loss": 1.396,
+      "step": 5005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526466653927002e-06,
+      "loss": 1.4251,
+      "step": 5006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526371106439901e-06,
+      "loss": 1.4313,
+      "step": 5007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5262755589528e-06,
+      "loss": 0.9896,
+      "step": 5008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526180011465699e-06,
+      "loss": 1.5706,
+      "step": 5009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.526084463978598e-06,
+      "loss": 1.1375,
+      "step": 5010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525988916491496e-06,
+      "loss": 1.7987,
+      "step": 5011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525893369004395e-06,
+      "loss": 1.5932,
+      "step": 5012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5257978215172944e-06,
+      "loss": 1.4993,
+      "step": 5013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5257022740301935e-06,
+      "loss": 1.6368,
+      "step": 5014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5256067265430925e-06,
+      "loss": 1.8724,
+      "step": 5015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5255111790559915e-06,
+      "loss": 1.3826,
+      "step": 5016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525415631568891e-06,
+      "loss": 1.2014,
+      "step": 5017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525320084081789e-06,
+      "loss": 1.3019,
+      "step": 5018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525224536594688e-06,
+      "loss": 1.3867,
+      "step": 5019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525128989107587e-06,
+      "loss": 1.6739,
+      "step": 5020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.525033441620486e-06,
+      "loss": 0.9839,
+      "step": 5021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524937894133385e-06,
+      "loss": 1.5947,
+      "step": 5022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524842346646284e-06,
+      "loss": 1.5929,
+      "step": 5023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524746799159182e-06,
+      "loss": 1.6776,
+      "step": 5024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524651251672081e-06,
+      "loss": 0.9788,
+      "step": 5025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52455570418498e-06,
+      "loss": 1.1792,
+      "step": 5026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524460156697879e-06,
+      "loss": 1.2683,
+      "step": 5027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524364609210778e-06,
+      "loss": 1.1846,
+      "step": 5028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524269061723677e-06,
+      "loss": 1.1625,
+      "step": 5029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524173514236576e-06,
+      "loss": 1.299,
+      "step": 5030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.524077966749475e-06,
+      "loss": 1.7401,
+      "step": 5031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523982419262374e-06,
+      "loss": 1.446,
+      "step": 5032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523886871775273e-06,
+      "loss": 0.9906,
+      "step": 5033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523791324288172e-06,
+      "loss": 0.8777,
+      "step": 5034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52369577680107e-06,
+      "loss": 1.6075,
+      "step": 5035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523600229313969e-06,
+      "loss": 1.5452,
+      "step": 5036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523504681826869e-06,
+      "loss": 1.3739,
+      "step": 5037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523409134339767e-06,
+      "loss": 1.2738,
+      "step": 5038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523313586852666e-06,
+      "loss": 1.6388,
+      "step": 5039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523218039365565e-06,
+      "loss": 1.179,
+      "step": 5040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523122491878464e-06,
+      "loss": 0.993,
+      "step": 5041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.523026944391362e-06,
+      "loss": 1.0509,
+      "step": 5042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522931396904261e-06,
+      "loss": 1.321,
+      "step": 5043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522835849417161e-06,
+      "loss": 1.2245,
+      "step": 5044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5227403019300594e-06,
+      "loss": 1.6368,
+      "step": 5045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5226447544429585e-06,
+      "loss": 1.2893,
+      "step": 5046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5225492069558575e-06,
+      "loss": 1.5168,
+      "step": 5047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5224536594687565e-06,
+      "loss": 1.3488,
+      "step": 5048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522358111981655e-06,
+      "loss": 1.4427,
+      "step": 5049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522262564494554e-06,
+      "loss": 1.7588,
+      "step": 5050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522167017007454e-06,
+      "loss": 1.8535,
+      "step": 5051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522071469520352e-06,
+      "loss": 1.5316,
+      "step": 5052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.521975922033251e-06,
+      "loss": 1.1803,
+      "step": 5053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52188037454615e-06,
+      "loss": 1.2936,
+      "step": 5054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521784827059049e-06,
+      "loss": 1.3336,
+      "step": 5055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521689279571947e-06,
+      "loss": 1.2602,
+      "step": 5056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521593732084846e-06,
+      "loss": 1.3819,
+      "step": 5057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521498184597746e-06,
+      "loss": 1.5926,
+      "step": 5058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521402637110644e-06,
+      "loss": 1.0205,
+      "step": 5059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521307089623543e-06,
+      "loss": 1.3613,
+      "step": 5060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521211542136442e-06,
+      "loss": 1.2797,
+      "step": 5061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.521115994649341e-06,
+      "loss": 1.5581,
+      "step": 5062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52102044716224e-06,
+      "loss": 1.4885,
+      "step": 5063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520924899675139e-06,
+      "loss": 1.4921,
+      "step": 5064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5208293521880385e-06,
+      "loss": 1.1505,
+      "step": 5065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520733804700937e-06,
+      "loss": 1.3555,
+      "step": 5066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520638257213836e-06,
+      "loss": 1.5544,
+      "step": 5067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520542709726735e-06,
+      "loss": 1.3008,
+      "step": 5068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520447162239634e-06,
+      "loss": 1.5396,
+      "step": 5069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520351614752532e-06,
+      "loss": 1.6876,
+      "step": 5070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520256067265431e-06,
+      "loss": 1.4099,
+      "step": 5071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.52016051977833e-06,
+      "loss": 1.5058,
+      "step": 5072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.520064972291229e-06,
+      "loss": 1.1967,
+      "step": 5073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519969424804128e-06,
+      "loss": 1.3654,
+      "step": 5074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519873877317027e-06,
+      "loss": 1.5208,
+      "step": 5075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519778329829926e-06,
+      "loss": 1.0904,
+      "step": 5076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5196827823428244e-06,
+      "loss": 1.5651,
+      "step": 5077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5195872348557235e-06,
+      "loss": 1.2274,
+      "step": 5078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5194916873686225e-06,
+      "loss": 1.2749,
+      "step": 5079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5193961398815215e-06,
+      "loss": 1.4137,
+      "step": 5080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519300592394421e-06,
+      "loss": 1.1363,
+      "step": 5081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.51920504490732e-06,
+      "loss": 1.1422,
+      "step": 5082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519109497420218e-06,
+      "loss": 1.3658,
+      "step": 5083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519013949933117e-06,
+      "loss": 1.1106,
+      "step": 5084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518918402446016e-06,
+      "loss": 1.2539,
+      "step": 5085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518822854958915e-06,
+      "loss": 1.1276,
+      "step": 5086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518727307471814e-06,
+      "loss": 1.5076,
+      "step": 5087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518631759984713e-06,
+      "loss": 1.3067,
+      "step": 5088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518536212497612e-06,
+      "loss": 1.3878,
+      "step": 5089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.51844066501051e-06,
+      "loss": 1.556,
+      "step": 5090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518345117523409e-06,
+      "loss": 1.1469,
+      "step": 5091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518249570036308e-06,
+      "loss": 0.8393,
+      "step": 5092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518154022549207e-06,
+      "loss": 1.1769,
+      "step": 5093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.518058475062106e-06,
+      "loss": 1.2431,
+      "step": 5094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5179629275750054e-06,
+      "loss": 1.0372,
+      "step": 5095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5178673800879045e-06,
+      "loss": 1.1689,
+      "step": 5096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517771832600803e-06,
+      "loss": 0.9895,
+      "step": 5097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517676285113702e-06,
+      "loss": 0.9899,
+      "step": 5098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517580737626601e-06,
+      "loss": 1.4417,
+      "step": 5099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5174851901395e-06,
+      "loss": 1.5301,
+      "step": 5100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517389642652399e-06,
+      "loss": 1.1392,
+      "step": 5101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517294095165298e-06,
+      "loss": 1.6246,
+      "step": 5102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517198547678197e-06,
+      "loss": 0.9162,
+      "step": 5103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517103000191095e-06,
+      "loss": 0.9336,
+      "step": 5104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517007452703994e-06,
+      "loss": 1.7651,
+      "step": 5105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.516911905216893e-06,
+      "loss": 0.8612,
+      "step": 5106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.516816357729792e-06,
+      "loss": 1.2875,
+      "step": 5107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51672081024269e-06,
+      "loss": 1.2159,
+      "step": 5108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51662526275559e-06,
+      "loss": 1.2717,
+      "step": 5109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.516529715268489e-06,
+      "loss": 1.5097,
+      "step": 5110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5164341677813875e-06,
+      "loss": 1.268,
+      "step": 5111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5163386202942865e-06,
+      "loss": 1.2717,
+      "step": 5112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.516243072807186e-06,
+      "loss": 1.1529,
+      "step": 5113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.516147525320085e-06,
+      "loss": 1.2831,
+      "step": 5114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.516051977832983e-06,
+      "loss": 1.0055,
+      "step": 5115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515956430345883e-06,
+      "loss": 1.6618,
+      "step": 5116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515860882858782e-06,
+      "loss": 1.1455,
+      "step": 5117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51576533537168e-06,
+      "loss": 1.6061,
+      "step": 5118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515669787884579e-06,
+      "loss": 1.6071,
+      "step": 5119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515574240397478e-06,
+      "loss": 1.0956,
+      "step": 5120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515478692910377e-06,
+      "loss": 1.2444,
+      "step": 5121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515383145423275e-06,
+      "loss": 1.2702,
+      "step": 5122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515287597936175e-06,
+      "loss": 1.0407,
+      "step": 5123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515192050449073e-06,
+      "loss": 1.2071,
+      "step": 5124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515096502961972e-06,
+      "loss": 1.5575,
+      "step": 5125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.515000955474871e-06,
+      "loss": 1.5763,
+      "step": 5126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5149054079877704e-06,
+      "loss": 1.0251,
+      "step": 5127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5148098605006695e-06,
+      "loss": 1.6307,
+      "step": 5128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514714313013568e-06,
+      "loss": 1.2121,
+      "step": 5129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5146187655264675e-06,
+      "loss": 1.8203,
+      "step": 5130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514523218039366e-06,
+      "loss": 1.1165,
+      "step": 5131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514427670552265e-06,
+      "loss": 1.6297,
+      "step": 5132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514332123065164e-06,
+      "loss": 1.276,
+      "step": 5133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514236575578063e-06,
+      "loss": 1.0871,
+      "step": 5134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.514141028090961e-06,
+      "loss": 1.0954,
+      "step": 5135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51404548060386e-06,
+      "loss": 1.0888,
+      "step": 5136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51394993311676e-06,
+      "loss": 1.1658,
+      "step": 5137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513854385629658e-06,
+      "loss": 1.627,
+      "step": 5138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513758838142557e-06,
+      "loss": 0.8786,
+      "step": 5139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513663290655456e-06,
+      "loss": 1.2791,
+      "step": 5140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513567743168355e-06,
+      "loss": 1.6045,
+      "step": 5141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5134721956812535e-06,
+      "loss": 1.6792,
+      "step": 5142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5133766481941525e-06,
+      "loss": 1.2474,
+      "step": 5143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5132811007070515e-06,
+      "loss": 1.4087,
+      "step": 5144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.513185553219951e-06,
+      "loss": 1.0947,
+      "step": 5145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51309000573285e-06,
+      "loss": 1.5859,
+      "step": 5146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512994458245749e-06,
+      "loss": 1.2525,
+      "step": 5147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512898910758648e-06,
+      "loss": 1.1233,
+      "step": 5148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512803363271546e-06,
+      "loss": 1.2232,
+      "step": 5149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512707815784445e-06,
+      "loss": 0.8932,
+      "step": 5150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512612268297344e-06,
+      "loss": 1.1204,
+      "step": 5151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512516720810243e-06,
+      "loss": 1.2211,
+      "step": 5152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512421173323142e-06,
+      "loss": 1.5543,
+      "step": 5153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512325625836041e-06,
+      "loss": 1.1098,
+      "step": 5154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.51223007834894e-06,
+      "loss": 1.1461,
+      "step": 5155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512134530861838e-06,
+      "loss": 1.2417,
+      "step": 5156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512038983374737e-06,
+      "loss": 0.936,
+      "step": 5157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.511943435887636e-06,
+      "loss": 1.1947,
+      "step": 5158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5118478884005354e-06,
+      "loss": 1.346,
+      "step": 5159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5117523409134345e-06,
+      "loss": 1.224,
+      "step": 5160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5116567934263335e-06,
+      "loss": 1.117,
+      "step": 5161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5115612459392325e-06,
+      "loss": 1.0882,
+      "step": 5162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.511465698452131e-06,
+      "loss": 1.2046,
+      "step": 5163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.51137015096503e-06,
+      "loss": 1.3301,
+      "step": 5164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.511274603477929e-06,
+      "loss": 1.1634,
+      "step": 5165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.511179055990828e-06,
+      "loss": 1.2359,
+      "step": 5166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.511083508503727e-06,
+      "loss": 1.7156,
+      "step": 5167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510987961016626e-06,
+      "loss": 1.3007,
+      "step": 5168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510892413529525e-06,
+      "loss": 1.4411,
+      "step": 5169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510796866042423e-06,
+      "loss": 1.2413,
+      "step": 5170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510701318555322e-06,
+      "loss": 1.4633,
+      "step": 5171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510605771068221e-06,
+      "loss": 1.0881,
+      "step": 5172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.51051022358112e-06,
+      "loss": 0.8795,
+      "step": 5173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510414676094019e-06,
+      "loss": 1.2168,
+      "step": 5174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510319128606918e-06,
+      "loss": 1.5457,
+      "step": 5175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5102235811198165e-06,
+      "loss": 0.8939,
+      "step": 5176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510128033632716e-06,
+      "loss": 1.4889,
+      "step": 5177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.510032486145615e-06,
+      "loss": 1.4127,
+      "step": 5178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509936938658514e-06,
+      "loss": 1.1993,
+      "step": 5179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509841391171413e-06,
+      "loss": 1.7558,
+      "step": 5180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509745843684312e-06,
+      "loss": 1.5014,
+      "step": 5181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509650296197211e-06,
+      "loss": 1.3136,
+      "step": 5182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509554748710109e-06,
+      "loss": 1.0085,
+      "step": 5183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509459201223008e-06,
+      "loss": 1.6115,
+      "step": 5184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509363653735907e-06,
+      "loss": 1.6755,
+      "step": 5185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509268106248806e-06,
+      "loss": 1.2194,
+      "step": 5186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509172558761704e-06,
+      "loss": 1.2633,
+      "step": 5187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509077011274604e-06,
+      "loss": 1.4381,
+      "step": 5188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508981463787503e-06,
+      "loss": 1.0557,
+      "step": 5189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508885916300401e-06,
+      "loss": 1.2562,
+      "step": 5190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5087903688133004e-06,
+      "loss": 1.1738,
+      "step": 5191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5086948213261995e-06,
+      "loss": 1.2969,
+      "step": 5192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5085992738390985e-06,
+      "loss": 1.0194,
+      "step": 5193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508503726351997e-06,
+      "loss": 0.9327,
+      "step": 5194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508408178864897e-06,
+      "loss": 1.1342,
+      "step": 5195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508312631377796e-06,
+      "loss": 1.4448,
+      "step": 5196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508217083890694e-06,
+      "loss": 1.1927,
+      "step": 5197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508121536403593e-06,
+      "loss": 1.3028,
+      "step": 5198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.508025988916492e-06,
+      "loss": 1.2877,
+      "step": 5199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507930441429391e-06,
+      "loss": 1.6546,
+      "step": 5200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507834893942289e-06,
+      "loss": 1.6292,
+      "step": 5201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507739346455189e-06,
+      "loss": 1.1988,
+      "step": 5202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507643798968088e-06,
+      "loss": 1.1129,
+      "step": 5203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507548251480986e-06,
+      "loss": 1.5123,
+      "step": 5204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507452703993885e-06,
+      "loss": 1.5338,
+      "step": 5205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507357156506784e-06,
+      "loss": 1.1674,
+      "step": 5206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.507261609019683e-06,
+      "loss": 1.3275,
+      "step": 5207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5071660615325815e-06,
+      "loss": 1.0597,
+      "step": 5208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5070705140454814e-06,
+      "loss": 1.3408,
+      "step": 5209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5069749665583805e-06,
+      "loss": 1.4719,
+      "step": 5210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506879419071279e-06,
+      "loss": 1.3101,
+      "step": 5211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506783871584178e-06,
+      "loss": 1.2465,
+      "step": 5212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506688324097077e-06,
+      "loss": 1.4815,
+      "step": 5213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506592776609976e-06,
+      "loss": 1.7006,
+      "step": 5214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506497229122874e-06,
+      "loss": 0.9972,
+      "step": 5215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506401681635773e-06,
+      "loss": 1.3259,
+      "step": 5216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506306134148673e-06,
+      "loss": 1.2784,
+      "step": 5217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506210586661571e-06,
+      "loss": 1.4036,
+      "step": 5218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50611503917447e-06,
+      "loss": 1.6036,
+      "step": 5219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.506019491687369e-06,
+      "loss": 1.4348,
+      "step": 5220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505923944200268e-06,
+      "loss": 1.365,
+      "step": 5221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505828396713166e-06,
+      "loss": 1.5084,
+      "step": 5222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505732849226065e-06,
+      "loss": 1.144,
+      "step": 5223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5056373017389645e-06,
+      "loss": 1.3215,
+      "step": 5224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5055417542518635e-06,
+      "loss": 1.308,
+      "step": 5225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5054462067647625e-06,
+      "loss": 1.6993,
+      "step": 5226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505350659277662e-06,
+      "loss": 1.7783,
+      "step": 5227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50525511179056e-06,
+      "loss": 1.1393,
+      "step": 5228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505159564303459e-06,
+      "loss": 1.3933,
+      "step": 5229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.505064016816358e-06,
+      "loss": 1.7063,
+      "step": 5230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504968469329257e-06,
+      "loss": 1.0691,
+      "step": 5231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504872921842156e-06,
+      "loss": 1.2492,
+      "step": 5232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504777374355055e-06,
+      "loss": 1.2815,
+      "step": 5233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504681826867954e-06,
+      "loss": 1.0261,
+      "step": 5234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504586279380852e-06,
+      "loss": 1.3864,
+      "step": 5235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504490731893751e-06,
+      "loss": 1.6813,
+      "step": 5236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50439518440665e-06,
+      "loss": 1.5087,
+      "step": 5237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504299636919549e-06,
+      "loss": 1.0779,
+      "step": 5238
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6533422427988449,
+      "eval_loss": 1.2650419473648071,
+      "eval_runtime": 141.4949,
+      "eval_samples_per_second": 296.145,
+      "eval_steps_per_second": 2.318,
+      "step": 5238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504204089432448e-06,
+      "loss": 1.2938,
+      "step": 5239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504108541945347e-06,
+      "loss": 1.5715,
+      "step": 5240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.504012994458246e-06,
+      "loss": 1.2795,
+      "step": 5241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503917446971145e-06,
+      "loss": 1.3005,
+      "step": 5242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503821899484044e-06,
+      "loss": 1.4977,
+      "step": 5243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503726351996943e-06,
+      "loss": 1.4376,
+      "step": 5244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503630804509842e-06,
+      "loss": 1.5956,
+      "step": 5245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503535257022741e-06,
+      "loss": 1.1762,
+      "step": 5246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.50343970953564e-06,
+      "loss": 1.059,
+      "step": 5247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503344162048539e-06,
+      "loss": 1.1735,
+      "step": 5248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503248614561437e-06,
+      "loss": 0.9168,
+      "step": 5249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503153067074336e-06,
+      "loss": 1.3403,
+      "step": 5250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.503057519587235e-06,
+      "loss": 1.2808,
+      "step": 5251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502961972100134e-06,
+      "loss": 1.2262,
+      "step": 5252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502866424613033e-06,
+      "loss": 1.3835,
+      "step": 5253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502770877125932e-06,
+      "loss": 1.1642,
+      "step": 5254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502675329638831e-06,
+      "loss": 0.7428,
+      "step": 5255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5025797821517295e-06,
+      "loss": 1.2921,
+      "step": 5256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5024842346646285e-06,
+      "loss": 1.2305,
+      "step": 5257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5023886871775275e-06,
+      "loss": 1.3729,
+      "step": 5258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5022931396904266e-06,
+      "loss": 0.763,
+      "step": 5259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502197592203326e-06,
+      "loss": 1.1136,
+      "step": 5260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502102044716225e-06,
+      "loss": 1.3653,
+      "step": 5261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.502006497229124e-06,
+      "loss": 1.4587,
+      "step": 5262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501910949742022e-06,
+      "loss": 1.567,
+      "step": 5263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501815402254921e-06,
+      "loss": 1.209,
+      "step": 5264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.50171985476782e-06,
+      "loss": 1.3746,
+      "step": 5265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501624307280719e-06,
+      "loss": 1.3511,
+      "step": 5266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501528759793618e-06,
+      "loss": 1.4817,
+      "step": 5267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501433212306517e-06,
+      "loss": 0.7973,
+      "step": 5268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501337664819416e-06,
+      "loss": 1.2782,
+      "step": 5269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501242117332314e-06,
+      "loss": 1.2795,
+      "step": 5270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501146569845213e-06,
+      "loss": 1.6882,
+      "step": 5271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.501051022358112e-06,
+      "loss": 1.3447,
+      "step": 5272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500955474871011e-06,
+      "loss": 1.1719,
+      "step": 5273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5008599273839105e-06,
+      "loss": 1.1755,
+      "step": 5274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5007643798968095e-06,
+      "loss": 1.1656,
+      "step": 5275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500668832409708e-06,
+      "loss": 1.3219,
+      "step": 5276
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500573284922607e-06,
+      "loss": 1.5585,
+      "step": 5277
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500477737435506e-06,
+      "loss": 1.0531,
+      "step": 5278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500382189948405e-06,
+      "loss": 1.3665,
+      "step": 5279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500286642461304e-06,
+      "loss": 1.5301,
+      "step": 5280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500191094974203e-06,
+      "loss": 0.8735,
+      "step": 5281
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.500095547487102e-06,
+      "loss": 1.9265,
+      "step": 5282
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5e-06,
+      "loss": 1.2209,
+      "step": 5283
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499904452512899e-06,
+      "loss": 1.2847,
+      "step": 5284
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499808905025798e-06,
+      "loss": 1.1962,
+      "step": 5285
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499713357538697e-06,
+      "loss": 1.3672,
+      "step": 5286
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499617810051595e-06,
+      "loss": 1.1954,
+      "step": 5287
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4995222625644945e-06,
+      "loss": 0.981,
+      "step": 5288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499426715077394e-06,
+      "loss": 1.2732,
+      "step": 5289
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4993311675902925e-06,
+      "loss": 1.1057,
+      "step": 5290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4992356201031916e-06,
+      "loss": 1.1452,
+      "step": 5291
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.499140072616091e-06,
+      "loss": 1.51,
+      "step": 5292
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.49904452512899e-06,
+      "loss": 1.3322,
+      "step": 5293
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498948977641888e-06,
+      "loss": 1.5968,
+      "step": 5294
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498853430154787e-06,
+      "loss": 1.057,
+      "step": 5295
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498757882667687e-06,
+      "loss": 0.6513,
+      "step": 5296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498662335180585e-06,
+      "loss": 0.7731,
+      "step": 5297
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498566787693484e-06,
+      "loss": 1.2063,
+      "step": 5298
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498471240206383e-06,
+      "loss": 1.2351,
+      "step": 5299
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498375692719282e-06,
+      "loss": 1.2132,
+      "step": 5300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.49828014523218e-06,
+      "loss": 1.4513,
+      "step": 5301
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498184597745079e-06,
+      "loss": 1.2966,
+      "step": 5302
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.498089050257979e-06,
+      "loss": 1.1462,
+      "step": 5303
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497993502770877e-06,
+      "loss": 1.392,
+      "step": 5304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497897955283776e-06,
+      "loss": 1.7105,
+      "step": 5305
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4978024077966755e-06,
+      "loss": 1.1598,
+      "step": 5306
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4977068603095745e-06,
+      "loss": 1.3159,
+      "step": 5307
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497611312822473e-06,
+      "loss": 1.3809,
+      "step": 5308
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497515765335372e-06,
+      "loss": 0.7175,
+      "step": 5309
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497420217848272e-06,
+      "loss": 1.7866,
+      "step": 5310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.49732467036117e-06,
+      "loss": 1.0796,
+      "step": 5311
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497229122874069e-06,
+      "loss": 1.5885,
+      "step": 5312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497133575386968e-06,
+      "loss": 1.133,
+      "step": 5313
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497038027899867e-06,
+      "loss": 1.2394,
+      "step": 5314
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.496942480412765e-06,
+      "loss": 1.1909,
+      "step": 5315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.496846932925664e-06,
+      "loss": 1.4006,
+      "step": 5316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496751385438563e-06,
+      "loss": 1.6451,
+      "step": 5317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496655837951462e-06,
+      "loss": 1.3334,
+      "step": 5318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496560290464361e-06,
+      "loss": 1.524,
+      "step": 5319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49646474297726e-06,
+      "loss": 1.4025,
+      "step": 5320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496369195490159e-06,
+      "loss": 1.1804,
+      "step": 5321
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4962736480030575e-06,
+      "loss": 1.3003,
+      "step": 5322
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4961781005159566e-06,
+      "loss": 1.145,
+      "step": 5323
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496082553028856e-06,
+      "loss": 1.3368,
+      "step": 5324
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495987005541755e-06,
+      "loss": 1.4675,
+      "step": 5325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495891458054654e-06,
+      "loss": 1.3664,
+      "step": 5326
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495795910567553e-06,
+      "loss": 1.237,
+      "step": 5327
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495700363080451e-06,
+      "loss": 1.0104,
+      "step": 5328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49560481559335e-06,
+      "loss": 1.1208,
+      "step": 5329
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495509268106249e-06,
+      "loss": 1.274,
+      "step": 5330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495413720619148e-06,
+      "loss": 1.3894,
+      "step": 5331
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495318173132047e-06,
+      "loss": 1.563,
+      "step": 5332
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495222625644946e-06,
+      "loss": 1.0148,
+      "step": 5333
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495127078157845e-06,
+      "loss": 1.4393,
+      "step": 5334
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.495031530670743e-06,
+      "loss": 1.1983,
+      "step": 5335
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494935983183642e-06,
+      "loss": 1.664,
+      "step": 5336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494840435696541e-06,
+      "loss": 1.4346,
+      "step": 5337
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4947448882094405e-06,
+      "loss": 1.3632,
+      "step": 5338
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4946493407223395e-06,
+      "loss": 1.6422,
+      "step": 5339
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4945537932352385e-06,
+      "loss": 1.5342,
+      "step": 5340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4944582457481376e-06,
+      "loss": 1.6161,
+      "step": 5341
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494362698261036e-06,
+      "loss": 1.0522,
+      "step": 5342
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494267150773935e-06,
+      "loss": 0.953,
+      "step": 5343
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494171603286834e-06,
+      "loss": 1.4038,
+      "step": 5344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.494076055799733e-06,
+      "loss": 1.234,
+      "step": 5345
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493980508312632e-06,
+      "loss": 1.3275,
+      "step": 5346
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493884960825531e-06,
+      "loss": 1.3107,
+      "step": 5347
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49378941333843e-06,
+      "loss": 0.9773,
+      "step": 5348
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493693865851328e-06,
+      "loss": 1.5859,
+      "step": 5349
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493598318364227e-06,
+      "loss": 1.2226,
+      "step": 5350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493502770877126e-06,
+      "loss": 1.2958,
+      "step": 5351
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493407223390025e-06,
+      "loss": 1.0032,
+      "step": 5352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493311675902924e-06,
+      "loss": 1.4445,
+      "step": 5353
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493216128415823e-06,
+      "loss": 1.1494,
+      "step": 5354
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493120580928722e-06,
+      "loss": 1.485,
+      "step": 5355
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.493025033441621e-06,
+      "loss": 1.1597,
+      "step": 5356
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49292948595452e-06,
+      "loss": 0.9451,
+      "step": 5357
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492833938467419e-06,
+      "loss": 1.1342,
+      "step": 5358
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492738390980318e-06,
+      "loss": 1.4718,
+      "step": 5359
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492642843493216e-06,
+      "loss": 0.7015,
+      "step": 5360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492547296006116e-06,
+      "loss": 0.8797,
+      "step": 5361
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492451748519015e-06,
+      "loss": 1.0051,
+      "step": 5362
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492356201031913e-06,
+      "loss": 1.3355,
+      "step": 5363
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492260653544812e-06,
+      "loss": 0.9274,
+      "step": 5364
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.492165106057711e-06,
+      "loss": 1.2256,
+      "step": 5365
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.49206955857061e-06,
+      "loss": 1.4029,
+      "step": 5366
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.491974011083508e-06,
+      "loss": 1.2238,
+      "step": 5367
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.491878463596408e-06,
+      "loss": 0.9274,
+      "step": 5368
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.491782916109306e-06,
+      "loss": 1.1958,
+      "step": 5369
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4916873686222055e-06,
+      "loss": 1.5379,
+      "step": 5370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4915918211351045e-06,
+      "loss": 1.2574,
+      "step": 5371
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4914962736480035e-06,
+      "loss": 1.1344,
+      "step": 5372
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4914007261609026e-06,
+      "loss": 1.1447,
+      "step": 5373
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.491305178673801e-06,
+      "loss": 1.4325,
+      "step": 5374
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.491209631186701e-06,
+      "loss": 1.2906,
+      "step": 5375
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.491114083699599e-06,
+      "loss": 1.005,
+      "step": 5376
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.491018536212498e-06,
+      "loss": 1.2942,
+      "step": 5377
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490922988725397e-06,
+      "loss": 1.0313,
+      "step": 5378
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490827441238296e-06,
+      "loss": 1.2603,
+      "step": 5379
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490731893751194e-06,
+      "loss": 1.3691,
+      "step": 5380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490636346264093e-06,
+      "loss": 1.1469,
+      "step": 5381
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490540798776993e-06,
+      "loss": 1.176,
+      "step": 5382
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490445251289891e-06,
+      "loss": 0.9912,
+      "step": 5383
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.49034970380279e-06,
+      "loss": 0.8677,
+      "step": 5384
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490254156315689e-06,
+      "loss": 1.1323,
+      "step": 5385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.490158608828588e-06,
+      "loss": 1.3699,
+      "step": 5386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4900630613414866e-06,
+      "loss": 0.9969,
+      "step": 5387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489967513854386e-06,
+      "loss": 1.9922,
+      "step": 5388
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4898719663672855e-06,
+      "loss": 0.6166,
+      "step": 5389
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489776418880184e-06,
+      "loss": 1.379,
+      "step": 5390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489680871393083e-06,
+      "loss": 0.9642,
+      "step": 5391
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489585323905982e-06,
+      "loss": 0.9405,
+      "step": 5392
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489489776418881e-06,
+      "loss": 1.25,
+      "step": 5393
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489394228931779e-06,
+      "loss": 0.9285,
+      "step": 5394
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489298681444678e-06,
+      "loss": 1.5431,
+      "step": 5395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489203133957577e-06,
+      "loss": 1.5388,
+      "step": 5396
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489107586470476e-06,
+      "loss": 1.1214,
+      "step": 5397
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.489012038983375e-06,
+      "loss": 1.1396,
+      "step": 5398
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488916491496274e-06,
+      "loss": 1.0945,
+      "step": 5399
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488820944009173e-06,
+      "loss": 0.8853,
+      "step": 5400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488725396522071e-06,
+      "loss": 1.5186,
+      "step": 5401
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4886298490349705e-06,
+      "loss": 1.2477,
+      "step": 5402
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4885343015478695e-06,
+      "loss": 0.7067,
+      "step": 5403
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4884387540607685e-06,
+      "loss": 1.5049,
+      "step": 5404
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4883432065736676e-06,
+      "loss": 1.3226,
+      "step": 5405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488247659086567e-06,
+      "loss": 1.1203,
+      "step": 5406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488152111599466e-06,
+      "loss": 1.2589,
+      "step": 5407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.488056564112364e-06,
+      "loss": 0.8782,
+      "step": 5408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487961016625263e-06,
+      "loss": 0.9771,
+      "step": 5409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487865469138162e-06,
+      "loss": 1.2285,
+      "step": 5410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487769921651061e-06,
+      "loss": 1.3628,
+      "step": 5411
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.48767437416396e-06,
+      "loss": 1.3578,
+      "step": 5412
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487578826676859e-06,
+      "loss": 1.3172,
+      "step": 5413
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487483279189758e-06,
+      "loss": 0.8899,
+      "step": 5414
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487387731702656e-06,
+      "loss": 1.0903,
+      "step": 5415
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487292184215555e-06,
+      "loss": 1.6452,
+      "step": 5416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487196636728454e-06,
+      "loss": 1.518,
+      "step": 5417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487101089241353e-06,
+      "loss": 1.658,
+      "step": 5418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.487005541754252e-06,
+      "loss": 1.5534,
+      "step": 5419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4869099942671515e-06,
+      "loss": 1.4781,
+      "step": 5420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.48681444678005e-06,
+      "loss": 1.1938,
+      "step": 5421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486718899292949e-06,
+      "loss": 1.435,
+      "step": 5422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486623351805848e-06,
+      "loss": 1.5786,
+      "step": 5423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486527804318747e-06,
+      "loss": 1.0706,
+      "step": 5424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486432256831646e-06,
+      "loss": 1.5216,
+      "step": 5425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486336709344545e-06,
+      "loss": 1.0278,
+      "step": 5426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486241161857444e-06,
+      "loss": 1.1854,
+      "step": 5427
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486145614370342e-06,
+      "loss": 1.1523,
+      "step": 5428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.486050066883241e-06,
+      "loss": 1.2408,
+      "step": 5429
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48595451939614e-06,
+      "loss": 0.7041,
+      "step": 5430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.485858971909039e-06,
+      "loss": 1.0084,
+      "step": 5431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.485763424421937e-06,
+      "loss": 0.7976,
+      "step": 5432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.485667876934837e-06,
+      "loss": 1.142,
+      "step": 5433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.485572329447736e-06,
+      "loss": 0.9537,
+      "step": 5434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4854767819606345e-06,
+      "loss": 1.053,
+      "step": 5435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4853812344735335e-06,
+      "loss": 1.3981,
+      "step": 5436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4852856869864326e-06,
+      "loss": 1.1179,
+      "step": 5437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.485190139499332e-06,
+      "loss": 0.7224,
+      "step": 5438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48509459201223e-06,
+      "loss": 1.0065,
+      "step": 5439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48499904452513e-06,
+      "loss": 1.2489,
+      "step": 5440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484903497038029e-06,
+      "loss": 1.1708,
+      "step": 5441
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484807949550927e-06,
+      "loss": 1.3967,
+      "step": 5442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484712402063826e-06,
+      "loss": 1.1914,
+      "step": 5443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484616854576725e-06,
+      "loss": 1.5776,
+      "step": 5444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484521307089624e-06,
+      "loss": 1.051,
+      "step": 5445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484425759602522e-06,
+      "loss": 1.3556,
+      "step": 5446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484330212115422e-06,
+      "loss": 1.1447,
+      "step": 5447
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484234664628321e-06,
+      "loss": 1.4856,
+      "step": 5448
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484139117141219e-06,
+      "loss": 1.1647,
+      "step": 5449
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.484043569654118e-06,
+      "loss": 1.1866,
+      "step": 5450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483948022167017e-06,
+      "loss": 1.3013,
+      "step": 5451
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4838524746799165e-06,
+      "loss": 1.0064,
+      "step": 5452
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483756927192815e-06,
+      "loss": 0.9019,
+      "step": 5453
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4836613797057145e-06,
+      "loss": 1.1622,
+      "step": 5454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4835658322186136e-06,
+      "loss": 1.18,
+      "step": 5455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483470284731512e-06,
+      "loss": 1.5131,
+      "step": 5456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483374737244411e-06,
+      "loss": 1.6028,
+      "step": 5457
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.48327918975731e-06,
+      "loss": 1.0394,
+      "step": 5458
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483183642270209e-06,
+      "loss": 1.3955,
+      "step": 5459
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.483088094783107e-06,
+      "loss": 1.464,
+      "step": 5460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482992547296007e-06,
+      "loss": 1.4742,
+      "step": 5461
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482896999808906e-06,
+      "loss": 1.216,
+      "step": 5462
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482801452321804e-06,
+      "loss": 1.2371,
+      "step": 5463
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482705904834703e-06,
+      "loss": 0.958,
+      "step": 5464
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482610357347602e-06,
+      "loss": 0.7646,
+      "step": 5465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482514809860501e-06,
+      "loss": 1.4183,
+      "step": 5466
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4824192623733995e-06,
+      "loss": 1.6377,
+      "step": 5467
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4823237148862985e-06,
+      "loss": 1.4202,
+      "step": 5468
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4822281673991976e-06,
+      "loss": 1.8048,
+      "step": 5469
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482132619912097e-06,
+      "loss": 1.0144,
+      "step": 5470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482037072424996e-06,
+      "loss": 1.0649,
+      "step": 5471
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481941524937895e-06,
+      "loss": 1.1305,
+      "step": 5472
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481845977450794e-06,
+      "loss": 1.5719,
+      "step": 5473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481750429963692e-06,
+      "loss": 1.71,
+      "step": 5474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481654882476591e-06,
+      "loss": 1.5252,
+      "step": 5475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.48155933498949e-06,
+      "loss": 1.358,
+      "step": 5476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481463787502389e-06,
+      "loss": 1.0857,
+      "step": 5477
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481368240015288e-06,
+      "loss": 1.0804,
+      "step": 5478
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481272692528187e-06,
+      "loss": 1.3047,
+      "step": 5479
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481177145041085e-06,
+      "loss": 1.4431,
+      "step": 5480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.481081597553984e-06,
+      "loss": 1.4131,
+      "step": 5481
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480986050066883e-06,
+      "loss": 0.9819,
+      "step": 5482
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480890502579782e-06,
+      "loss": 1.5792,
+      "step": 5483
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4807949550926815e-06,
+      "loss": 0.8433,
+      "step": 5484
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4806994076055805e-06,
+      "loss": 1.4516,
+      "step": 5485
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4806038601184795e-06,
+      "loss": 0.9712,
+      "step": 5486
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480508312631378e-06,
+      "loss": 1.3528,
+      "step": 5487
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480412765144277e-06,
+      "loss": 1.4654,
+      "step": 5488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480317217657176e-06,
+      "loss": 1.6231,
+      "step": 5489
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480221670170075e-06,
+      "loss": 1.3514,
+      "step": 5490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480126122682974e-06,
+      "loss": 1.5513,
+      "step": 5491
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.480030575195873e-06,
+      "loss": 1.2849,
+      "step": 5492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479935027708772e-06,
+      "loss": 1.5998,
+      "step": 5493
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.47983948022167e-06,
+      "loss": 1.4153,
+      "step": 5494
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479743932734569e-06,
+      "loss": 1.2397,
+      "step": 5495
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479648385247468e-06,
+      "loss": 1.3939,
+      "step": 5496
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479552837760367e-06,
+      "loss": 1.1008,
+      "step": 5497
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479457290273266e-06,
+      "loss": 1.3972,
+      "step": 5498
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479361742786165e-06,
+      "loss": 1.106,
+      "step": 5499
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479266195299064e-06,
+      "loss": 1.1682,
+      "step": 5500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4791706478119626e-06,
+      "loss": 1.1539,
+      "step": 5501
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.479075100324862e-06,
+      "loss": 1.0822,
+      "step": 5502
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478979552837761e-06,
+      "loss": 1.2614,
+      "step": 5503
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.47888400535066e-06,
+      "loss": 1.2411,
+      "step": 5504
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478788457863559e-06,
+      "loss": 1.4402,
+      "step": 5505
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478692910376458e-06,
+      "loss": 1.2862,
+      "step": 5506
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478597362889357e-06,
+      "loss": 1.6361,
+      "step": 5507
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478501815402255e-06,
+      "loss": 1.0774,
+      "step": 5508
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478406267915154e-06,
+      "loss": 1.8407,
+      "step": 5509
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478310720428053e-06,
+      "loss": 0.8423,
+      "step": 5510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478215172940952e-06,
+      "loss": 1.3667,
+      "step": 5511
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.478119625453851e-06,
+      "loss": 0.6617,
+      "step": 5512
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.47802407796675e-06,
+      "loss": 1.254,
+      "step": 5513
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477928530479649e-06,
+      "loss": 0.9021,
+      "step": 5514
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477832982992547e-06,
+      "loss": 1.7067,
+      "step": 5515
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4777374355054465e-06,
+      "loss": 1.5973,
+      "step": 5516
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4776418880183455e-06,
+      "loss": 1.642,
+      "step": 5517
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4775463405312445e-06,
+      "loss": 1.1301,
+      "step": 5518
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4774507930441436e-06,
+      "loss": 1.3911,
+      "step": 5519
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477355245557043e-06,
+      "loss": 1.26,
+      "step": 5520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477259698069941e-06,
+      "loss": 1.5793,
+      "step": 5521
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.47716415058284e-06,
+      "loss": 1.3569,
+      "step": 5522
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477068603095739e-06,
+      "loss": 1.2554,
+      "step": 5523
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476973055608638e-06,
+      "loss": 1.7295,
+      "step": 5524
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476877508121537e-06,
+      "loss": 1.1299,
+      "step": 5525
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476781960634436e-06,
+      "loss": 0.7732,
+      "step": 5526
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.476686413147335e-06,
+      "loss": 1.1494,
+      "step": 5527
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.476590865660233e-06,
+      "loss": 1.1754,
+      "step": 5528
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.476495318173132e-06,
+      "loss": 1.2106,
+      "step": 5529
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.476399770686031e-06,
+      "loss": 1.4822,
+      "step": 5530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47630422319893e-06,
+      "loss": 1.4545,
+      "step": 5531
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4762086757118285e-06,
+      "loss": 1.6082,
+      "step": 5532
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.476113128224728e-06,
+      "loss": 0.9215,
+      "step": 5533
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4760175807376275e-06,
+      "loss": 1.352,
+      "step": 5534
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475922033250526e-06,
+      "loss": 1.7591,
+      "step": 5535
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475826485763425e-06,
+      "loss": 0.9511,
+      "step": 5536
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475730938276324e-06,
+      "loss": 1.2303,
+      "step": 5537
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475635390789223e-06,
+      "loss": 1.4897,
+      "step": 5538
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475539843302121e-06,
+      "loss": 1.3352,
+      "step": 5539
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47544429581502e-06,
+      "loss": 1.3586,
+      "step": 5540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47534874832792e-06,
+      "loss": 0.9031,
+      "step": 5541
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475253200840818e-06,
+      "loss": 1.0483,
+      "step": 5542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475157653353717e-06,
+      "loss": 1.5525,
+      "step": 5543
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.475062105866616e-06,
+      "loss": 1.3419,
+      "step": 5544
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474966558379515e-06,
+      "loss": 1.2805,
+      "step": 5545
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474871010892413e-06,
+      "loss": 1.7488,
+      "step": 5546
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474775463405312e-06,
+      "loss": 1.5113,
+      "step": 5547
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474679915918212e-06,
+      "loss": 1.5116,
+      "step": 5548
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4745843684311105e-06,
+      "loss": 1.352,
+      "step": 5549
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4744888209440095e-06,
+      "loss": 1.1418,
+      "step": 5550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4743932734569086e-06,
+      "loss": 1.0765,
+      "step": 5551
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474297725969808e-06,
+      "loss": 1.0691,
+      "step": 5552
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474202178482706e-06,
+      "loss": 1.4288,
+      "step": 5553
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474106630995605e-06,
+      "loss": 1.226,
+      "step": 5554
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.474011083508505e-06,
+      "loss": 1.6986,
+      "step": 5555
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473915536021403e-06,
+      "loss": 1.2222,
+      "step": 5556
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473819988534302e-06,
+      "loss": 1.3538,
+      "step": 5557
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473724441047201e-06,
+      "loss": 0.9825,
+      "step": 5558
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4736288935601e-06,
+      "loss": 1.1357,
+      "step": 5559
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473533346072998e-06,
+      "loss": 0.6761,
+      "step": 5560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473437798585897e-06,
+      "loss": 0.808,
+      "step": 5561
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473342251098796e-06,
+      "loss": 1.2169,
+      "step": 5562
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473246703611695e-06,
+      "loss": 1.44,
+      "step": 5563
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473151156124594e-06,
+      "loss": 1.6693,
+      "step": 5564
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.473055608637493e-06,
+      "loss": 1.0386,
+      "step": 5565
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4729600611503925e-06,
+      "loss": 1.5448,
+      "step": 5566
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472864513663291e-06,
+      "loss": 1.2231,
+      "step": 5567
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47276896617619e-06,
+      "loss": 1.0304,
+      "step": 5568
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472673418689089e-06,
+      "loss": 1.3237,
+      "step": 5569
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472577871201988e-06,
+      "loss": 1.3203,
+      "step": 5570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472482323714887e-06,
+      "loss": 1.5454,
+      "step": 5571
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472386776227786e-06,
+      "loss": 1.3765,
+      "step": 5572
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472291228740684e-06,
+      "loss": 1.5113,
+      "step": 5573
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472195681253583e-06,
+      "loss": 1.2137,
+      "step": 5574
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472100133766482e-06,
+      "loss": 0.6084,
+      "step": 5575
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.472004586279381e-06,
+      "loss": 0.995,
+      "step": 5576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.47190903879228e-06,
+      "loss": 2.0319,
+      "step": 5577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.471813491305179e-06,
+      "loss": 1.1084,
+      "step": 5578
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.471717943818078e-06,
+      "loss": 0.9113,
+      "step": 5579
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4716223963309765e-06,
+      "loss": 0.9347,
+      "step": 5580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4715268488438755e-06,
+      "loss": 1.3183,
+      "step": 5581
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4714313013567745e-06,
+      "loss": 1.0268,
+      "step": 5582
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4713357538696736e-06,
+      "loss": 1.1423,
+      "step": 5583
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.471240206382573e-06,
+      "loss": 1.1769,
+      "step": 5584
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.471144658895472e-06,
+      "loss": 1.2254,
+      "step": 5585
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.471049111408371e-06,
+      "loss": 1.5181,
+      "step": 5586
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470953563921269e-06,
+      "loss": 1.5228,
+      "step": 5587
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470858016434168e-06,
+      "loss": 0.9985,
+      "step": 5588
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470762468947067e-06,
+      "loss": 1.7721,
+      "step": 5589
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470666921459966e-06,
+      "loss": 1.0806,
+      "step": 5590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470571373972865e-06,
+      "loss": 1.2984,
+      "step": 5591
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470475826485764e-06,
+      "loss": 0.822,
+      "step": 5592
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470380278998663e-06,
+      "loss": 1.2665,
+      "step": 5593
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470284731511561e-06,
+      "loss": 1.3888,
+      "step": 5594
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.47018918402446e-06,
+      "loss": 1.3139,
+      "step": 5595
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.470093636537359e-06,
+      "loss": 1.0359,
+      "step": 5596
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469998089050258e-06,
+      "loss": 1.5568,
+      "step": 5597
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4699025415631575e-06,
+      "loss": 1.3708,
+      "step": 5598
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4698069940760565e-06,
+      "loss": 1.4653,
+      "step": 5599
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4697114465889555e-06,
+      "loss": 1.4884,
+      "step": 5600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469615899101854e-06,
+      "loss": 1.3127,
+      "step": 5601
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469520351614753e-06,
+      "loss": 1.3989,
+      "step": 5602
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469424804127652e-06,
+      "loss": 1.3364,
+      "step": 5603
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469329256640551e-06,
+      "loss": 1.2359,
+      "step": 5604
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.46923370915345e-06,
+      "loss": 0.98,
+      "step": 5605
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469138161666349e-06,
+      "loss": 1.4426,
+      "step": 5606
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.469042614179248e-06,
+      "loss": 0.9886,
+      "step": 5607
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468947066692146e-06,
+      "loss": 1.6781,
+      "step": 5608
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468851519205045e-06,
+      "loss": 0.9269,
+      "step": 5609
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468755971717944e-06,
+      "loss": 1.7727,
+      "step": 5610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468660424230843e-06,
+      "loss": 1.161,
+      "step": 5611
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4685648767437415e-06,
+      "loss": 1.2151,
+      "step": 5612
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468469329256641e-06,
+      "loss": 1.0007,
+      "step": 5613
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4683737817695395e-06,
+      "loss": 1.3838,
+      "step": 5614
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4682782342824386e-06,
+      "loss": 1.1591,
+      "step": 5615
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468182686795338e-06,
+      "loss": 0.9171,
+      "step": 5616
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.468087139308237e-06,
+      "loss": 1.1329,
+      "step": 5617
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467991591821136e-06,
+      "loss": 1.4181,
+      "step": 5618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467896044334034e-06,
+      "loss": 0.9106,
+      "step": 5619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467800496846934e-06,
+      "loss": 1.4607,
+      "step": 5620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467704949359832e-06,
+      "loss": 1.0217,
+      "step": 5621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467609401872731e-06,
+      "loss": 1.3022,
+      "step": 5622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.46751385438563e-06,
+      "loss": 1.3549,
+      "step": 5623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467418306898529e-06,
+      "loss": 1.3834,
+      "step": 5624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467322759411427e-06,
+      "loss": 0.8822,
+      "step": 5625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467227211924326e-06,
+      "loss": 1.3475,
+      "step": 5626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467131664437226e-06,
+      "loss": 1.269,
+      "step": 5627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467036116950124e-06,
+      "loss": 0.9042,
+      "step": 5628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.466940569463023e-06,
+      "loss": 1.2849,
+      "step": 5629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4668450219759225e-06,
+      "loss": 1.4039,
+      "step": 5630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4667494744888215e-06,
+      "loss": 1.0801,
+      "step": 5631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.46665392700172e-06,
+      "loss": 1.1261,
+      "step": 5632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466558379514619e-06,
+      "loss": 1.278,
+      "step": 5633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466462832027519e-06,
+      "loss": 0.9993,
+      "step": 5634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466367284540417e-06,
+      "loss": 1.7092,
+      "step": 5635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466271737053316e-06,
+      "loss": 1.3189,
+      "step": 5636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466176189566215e-06,
+      "loss": 1.1676,
+      "step": 5637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.466080642079114e-06,
+      "loss": 0.9465,
+      "step": 5638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465985094592012e-06,
+      "loss": 1.3779,
+      "step": 5639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465889547104911e-06,
+      "loss": 1.3229,
+      "step": 5640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465793999617811e-06,
+      "loss": 1.1095,
+      "step": 5641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465698452130709e-06,
+      "loss": 1.2807,
+      "step": 5642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465602904643608e-06,
+      "loss": 1.5098,
+      "step": 5643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465507357156507e-06,
+      "loss": 1.4339,
+      "step": 5644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465411809669406e-06,
+      "loss": 1.4227,
+      "step": 5645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4653162621823045e-06,
+      "loss": 1.2171,
+      "step": 5646
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4652207146952036e-06,
+      "loss": 1.0929,
+      "step": 5647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465125167208103e-06,
+      "loss": 1.6073,
+      "step": 5648
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.465029619721002e-06,
+      "loss": 1.4118,
+      "step": 5649
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464934072233901e-06,
+      "loss": 1.0945,
+      "step": 5650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4648385247468e-06,
+      "loss": 1.1619,
+      "step": 5651
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464742977259699e-06,
+      "loss": 1.0058,
+      "step": 5652
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464647429772597e-06,
+      "loss": 1.5618,
+      "step": 5653
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464551882285496e-06,
+      "loss": 1.0787,
+      "step": 5654
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464456334798395e-06,
+      "loss": 1.0665,
+      "step": 5655
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464360787311294e-06,
+      "loss": 1.1792,
+      "step": 5656
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464265239824193e-06,
+      "loss": 0.9846,
+      "step": 5657
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464169692337092e-06,
+      "loss": 0.8743,
+      "step": 5658
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.464074144849991e-06,
+      "loss": 1.2488,
+      "step": 5659
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463978597362889e-06,
+      "loss": 1.4566,
+      "step": 5660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463883049875788e-06,
+      "loss": 0.7201,
+      "step": 5661
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4637875023886875e-06,
+      "loss": 1.5294,
+      "step": 5662
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4636919549015865e-06,
+      "loss": 1.321,
+      "step": 5663
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4635964074144855e-06,
+      "loss": 0.987,
+      "step": 5664
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4635008599273846e-06,
+      "loss": 0.995,
+      "step": 5665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463405312440284e-06,
+      "loss": 1.3823,
+      "step": 5666
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463309764953182e-06,
+      "loss": 1.2857,
+      "step": 5667
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463214217466081e-06,
+      "loss": 1.2055,
+      "step": 5668
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.46311866997898e-06,
+      "loss": 1.4302,
+      "step": 5669
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.463023122491879e-06,
+      "loss": 1.1333,
+      "step": 5670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462927575004778e-06,
+      "loss": 1.2173,
+      "step": 5671
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462832027517677e-06,
+      "loss": 1.3048,
+      "step": 5672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462736480030575e-06,
+      "loss": 1.8739,
+      "step": 5673
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462640932543474e-06,
+      "loss": 1.4791,
+      "step": 5674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462545385056373e-06,
+      "loss": 1.2765,
+      "step": 5675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462449837569272e-06,
+      "loss": 0.9028,
+      "step": 5676
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462354290082171e-06,
+      "loss": 0.9675,
+      "step": 5677
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.46225874259507e-06,
+      "loss": 1.136,
+      "step": 5678
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462163195107969e-06,
+      "loss": 1.4754,
+      "step": 5679
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462067647620868e-06,
+      "loss": 1.0598,
+      "step": 5680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461972100133767e-06,
+      "loss": 0.9031,
+      "step": 5681
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461876552646666e-06,
+      "loss": 1.1974,
+      "step": 5682
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461781005159565e-06,
+      "loss": 1.5607,
+      "step": 5683
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461685457672463e-06,
+      "loss": 0.9409,
+      "step": 5684
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461589910185363e-06,
+      "loss": 1.4517,
+      "step": 5685
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461494362698262e-06,
+      "loss": 1.5714,
+      "step": 5686
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.46139881521116e-06,
+      "loss": 1.0656,
+      "step": 5687
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461303267724059e-06,
+      "loss": 1.2872,
+      "step": 5688
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461207720236958e-06,
+      "loss": 1.5295,
+      "step": 5689
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461112172749857e-06,
+      "loss": 0.9181,
+      "step": 5690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.461016625262755e-06,
+      "loss": 1.3817,
+      "step": 5691
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460921077775655e-06,
+      "loss": 0.9948,
+      "step": 5692
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460825530288554e-06,
+      "loss": 1.1063,
+      "step": 5693
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4607299828014525e-06,
+      "loss": 1.1017,
+      "step": 5694
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4606344353143515e-06,
+      "loss": 0.9358,
+      "step": 5695
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4605388878272505e-06,
+      "loss": 0.8188,
+      "step": 5696
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4604433403401496e-06,
+      "loss": 1.3457,
+      "step": 5697
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460347792853048e-06,
+      "loss": 0.8007,
+      "step": 5698
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460252245365948e-06,
+      "loss": 1.4266,
+      "step": 5699
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460156697878847e-06,
+      "loss": 1.0754,
+      "step": 5700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.460061150391745e-06,
+      "loss": 1.2157,
+      "step": 5701
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459965602904644e-06,
+      "loss": 0.8513,
+      "step": 5702
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459870055417543e-06,
+      "loss": 1.2919,
+      "step": 5703
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459774507930442e-06,
+      "loss": 1.4163,
+      "step": 5704
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.45967896044334e-06,
+      "loss": 1.4858,
+      "step": 5705
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.45958341295624e-06,
+      "loss": 1.2578,
+      "step": 5706
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459487865469139e-06,
+      "loss": 1.378,
+      "step": 5707
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459392317982037e-06,
+      "loss": 1.579,
+      "step": 5708
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459296770494936e-06,
+      "loss": 1.1595,
+      "step": 5709
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459201223007835e-06,
+      "loss": 0.6231,
+      "step": 5710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459105675520734e-06,
+      "loss": 0.923,
+      "step": 5711
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.459010128033633e-06,
+      "loss": 1.5222,
+      "step": 5712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4589145805465325e-06,
+      "loss": 1.1565,
+      "step": 5713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458819033059431e-06,
+      "loss": 0.8143,
+      "step": 5714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.45872348557233e-06,
+      "loss": 1.3737,
+      "step": 5715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458627938085229e-06,
+      "loss": 1.4401,
+      "step": 5716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458532390598128e-06,
+      "loss": 1.1937,
+      "step": 5717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458436843111027e-06,
+      "loss": 1.908,
+      "step": 5718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458341295623925e-06,
+      "loss": 1.0982,
+      "step": 5719
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458245748136824e-06,
+      "loss": 1.3426,
+      "step": 5720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458150200649723e-06,
+      "loss": 1.7665,
+      "step": 5721
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.458054653162622e-06,
+      "loss": 0.8988,
+      "step": 5722
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457959105675521e-06,
+      "loss": 1.0292,
+      "step": 5723
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.45786355818842e-06,
+      "loss": 1.1359,
+      "step": 5724
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457768010701318e-06,
+      "loss": 1.4925,
+      "step": 5725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4576724632142175e-06,
+      "loss": 1.2339,
+      "step": 5726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4575769157271165e-06,
+      "loss": 1.1859,
+      "step": 5727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4574813682400155e-06,
+      "loss": 1.23,
+      "step": 5728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4573858207529146e-06,
+      "loss": 1.0891,
+      "step": 5729
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457290273265814e-06,
+      "loss": 1.5012,
+      "step": 5730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457194725778713e-06,
+      "loss": 0.9712,
+      "step": 5731
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457099178291611e-06,
+      "loss": 1.2602,
+      "step": 5732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.45700363080451e-06,
+      "loss": 1.4998,
+      "step": 5733
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456908083317409e-06,
+      "loss": 1.184,
+      "step": 5734
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456812535830308e-06,
+      "loss": 1.3851,
+      "step": 5735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456716988343207e-06,
+      "loss": 1.0956,
+      "step": 5736
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456621440856106e-06,
+      "loss": 1.2312,
+      "step": 5737
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456525893369005e-06,
+      "loss": 0.9854,
+      "step": 5738
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456430345881903e-06,
+      "loss": 1.0291,
+      "step": 5739
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456334798394802e-06,
+      "loss": 1.2931,
+      "step": 5740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456239250907701e-06,
+      "loss": 1.6185,
+      "step": 5741
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4561437034206e-06,
+      "loss": 1.3483,
+      "step": 5742
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.456048155933499e-06,
+      "loss": 1.1223,
+      "step": 5743
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4559526084463985e-06,
+      "loss": 1.23,
+      "step": 5744
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4558570609592975e-06,
+      "loss": 1.3528,
+      "step": 5745
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455761513472196e-06,
+      "loss": 1.255,
+      "step": 5746
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455665965985095e-06,
+      "loss": 1.1526,
+      "step": 5747
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455570418497994e-06,
+      "loss": 1.7386,
+      "step": 5748
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455474871010893e-06,
+      "loss": 1.3652,
+      "step": 5749
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455379323523792e-06,
+      "loss": 0.7372,
+      "step": 5750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455283776036691e-06,
+      "loss": 1.1144,
+      "step": 5751
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.45518822854959e-06,
+      "loss": 1.3145,
+      "step": 5752
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.455092681062488e-06,
+      "loss": 1.2955,
+      "step": 5753
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454997133575387e-06,
+      "loss": 1.2072,
+      "step": 5754
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454901586088286e-06,
+      "loss": 1.2008,
+      "step": 5755
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454806038601185e-06,
+      "loss": 1.2412,
+      "step": 5756
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454710491114084e-06,
+      "loss": 1.3048,
+      "step": 5757
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454614943626983e-06,
+      "loss": 1.4487,
+      "step": 5758
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454519396139882e-06,
+      "loss": 1.0542,
+      "step": 5759
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4544238486527805e-06,
+      "loss": 1.3324,
+      "step": 5760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4543283011656796e-06,
+      "loss": 1.1114,
+      "step": 5761
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454232753678579e-06,
+      "loss": 1.1998,
+      "step": 5762
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454137206191478e-06,
+      "loss": 1.2222,
+      "step": 5763
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.454041658704377e-06,
+      "loss": 1.5331,
+      "step": 5764
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453946111217276e-06,
+      "loss": 1.0285,
+      "step": 5765
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453850563730174e-06,
+      "loss": 1.1012,
+      "step": 5766
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453755016243073e-06,
+      "loss": 0.8895,
+      "step": 5767
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453659468755972e-06,
+      "loss": 1.4388,
+      "step": 5768
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453563921268871e-06,
+      "loss": 1.6596,
+      "step": 5769
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.45346837378177e-06,
+      "loss": 1.5747,
+      "step": 5770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453372826294669e-06,
+      "loss": 1.4126,
+      "step": 5771
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453277278807568e-06,
+      "loss": 0.94,
+      "step": 5772
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453181731320466e-06,
+      "loss": 1.1482,
+      "step": 5773
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.453086183833365e-06,
+      "loss": 0.9375,
+      "step": 5774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452990636346264e-06,
+      "loss": 1.0194,
+      "step": 5775
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4528950888591635e-06,
+      "loss": 1.1819,
+      "step": 5776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452799541372062e-06,
+      "loss": 1.4477,
+      "step": 5777
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4527039938849615e-06,
+      "loss": 1.1617,
+      "step": 5778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4526084463978606e-06,
+      "loss": 1.1853,
+      "step": 5779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452512898910759e-06,
+      "loss": 1.6795,
+      "step": 5780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452417351423658e-06,
+      "loss": 1.0167,
+      "step": 5781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452321803936557e-06,
+      "loss": 1.5931,
+      "step": 5782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452226256449456e-06,
+      "loss": 1.0236,
+      "step": 5783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452130708962354e-06,
+      "loss": 0.919,
+      "step": 5784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.452035161475254e-06,
+      "loss": 0.8358,
+      "step": 5785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451939613988153e-06,
+      "loss": 1.5399,
+      "step": 5786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451844066501051e-06,
+      "loss": 1.2457,
+      "step": 5787
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45174851901395e-06,
+      "loss": 1.1608,
+      "step": 5788
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.451652971526849e-06,
+      "loss": 1.1635,
+      "step": 5789
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.451557424039748e-06,
+      "loss": 1.2353,
+      "step": 5790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4514618765526465e-06,
+      "loss": 0.9364,
+      "step": 5791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4513663290655455e-06,
+      "loss": 1.3481,
+      "step": 5792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.451270781578445e-06,
+      "loss": 1.1574,
+      "step": 5793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.451175234091344e-06,
+      "loss": 1.2477,
+      "step": 5794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.451079686604243e-06,
+      "loss": 1.1401,
+      "step": 5795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450984139117142e-06,
+      "loss": 1.1821,
+      "step": 5796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450888591630041e-06,
+      "loss": 1.1876,
+      "step": 5797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450793044142939e-06,
+      "loss": 1.0718,
+      "step": 5798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450697496655838e-06,
+      "loss": 1.5597,
+      "step": 5799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450601949168738e-06,
+      "loss": 1.1859,
+      "step": 5800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450506401681636e-06,
+      "loss": 0.8397,
+      "step": 5801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450410854194535e-06,
+      "loss": 1.5421,
+      "step": 5802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450315306707434e-06,
+      "loss": 0.9835,
+      "step": 5803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450219759220333e-06,
+      "loss": 1.0538,
+      "step": 5804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.450124211733231e-06,
+      "loss": 1.3629,
+      "step": 5805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.45002866424613e-06,
+      "loss": 0.9946,
+      "step": 5806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.44993311675903e-06,
+      "loss": 1.052,
+      "step": 5807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4498375692719285e-06,
+      "loss": 1.526,
+      "step": 5808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4497420217848275e-06,
+      "loss": 1.077,
+      "step": 5809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4496464742977265e-06,
+      "loss": 1.2404,
+      "step": 5810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4495509268106256e-06,
+      "loss": 0.8959,
+      "step": 5811
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.449455379323524e-06,
+      "loss": 1.254,
+      "step": 5812
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.449359831836423e-06,
+      "loss": 0.9332,
+      "step": 5813
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.449264284349322e-06,
+      "loss": 1.3991,
+      "step": 5814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.449168736862221e-06,
+      "loss": 1.3242,
+      "step": 5815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.44907318937512e-06,
+      "loss": 0.8204,
+      "step": 5816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448977641888019e-06,
+      "loss": 1.0692,
+      "step": 5817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448882094400917e-06,
+      "loss": 1.1291,
+      "step": 5818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448786546913816e-06,
+      "loss": 0.9486,
+      "step": 5819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448690999426715e-06,
+      "loss": 1.4553,
+      "step": 5820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448595451939614e-06,
+      "loss": 1.0889,
+      "step": 5821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448499904452513e-06,
+      "loss": 1.1593,
+      "step": 5822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448404356965412e-06,
+      "loss": 1.0508,
+      "step": 5823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448308809478311e-06,
+      "loss": 1.332,
+      "step": 5824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4482132619912096e-06,
+      "loss": 1.5333,
+      "step": 5825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448117714504109e-06,
+      "loss": 1.0132,
+      "step": 5826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.448022167017008e-06,
+      "loss": 1.325,
+      "step": 5827
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447926619529907e-06,
+      "loss": 1.2029,
+      "step": 5828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447831072042806e-06,
+      "loss": 0.9896,
+      "step": 5829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447735524555705e-06,
+      "loss": 1.297,
+      "step": 5830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447639977068604e-06,
+      "loss": 1.32,
+      "step": 5831
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447544429581502e-06,
+      "loss": 1.2675,
+      "step": 5832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447448882094401e-06,
+      "loss": 0.8853,
+      "step": 5833
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4473533346073e-06,
+      "loss": 0.8595,
+      "step": 5834
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447257787120199e-06,
+      "loss": 1.3245,
+      "step": 5835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447162239633098e-06,
+      "loss": 1.0468,
+      "step": 5836
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.447066692145997e-06,
+      "loss": 1.1001,
+      "step": 5837
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.446971144658896e-06,
+      "loss": 1.1073,
+      "step": 5838
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.446875597171794e-06,
+      "loss": 1.424,
+      "step": 5839
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4467800496846934e-06,
+      "loss": 1.5007,
+      "step": 5840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4466845021975925e-06,
+      "loss": 0.6694,
+      "step": 5841
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4465889547104915e-06,
+      "loss": 1.1579,
+      "step": 5842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4464934072233906e-06,
+      "loss": 1.7861,
+      "step": 5843
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44639785973629e-06,
+      "loss": 1.4838,
+      "step": 5844
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.446302312249189e-06,
+      "loss": 1.3881,
+      "step": 5845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.446206764762087e-06,
+      "loss": 1.191,
+      "step": 5846
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.446111217274986e-06,
+      "loss": 1.1774,
+      "step": 5847
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.446015669787885e-06,
+      "loss": 1.1655,
+      "step": 5848
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445920122300784e-06,
+      "loss": 1.1247,
+      "step": 5849
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445824574813683e-06,
+      "loss": 2.1432,
+      "step": 5850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445729027326582e-06,
+      "loss": 1.4564,
+      "step": 5851
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445633479839481e-06,
+      "loss": 1.0899,
+      "step": 5852
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445537932352379e-06,
+      "loss": 1.3353,
+      "step": 5853
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445442384865278e-06,
+      "loss": 1.1858,
+      "step": 5854
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445346837378177e-06,
+      "loss": 1.0452,
+      "step": 5855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445251289891076e-06,
+      "loss": 1.5077,
+      "step": 5856
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.445155742403975e-06,
+      "loss": 1.5286,
+      "step": 5857
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4450601949168744e-06,
+      "loss": 1.4486,
+      "step": 5858
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4449646474297735e-06,
+      "loss": 0.7201,
+      "step": 5859
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444869099942672e-06,
+      "loss": 0.7228,
+      "step": 5860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444773552455571e-06,
+      "loss": 1.4797,
+      "step": 5861
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44467800496847e-06,
+      "loss": 1.3441,
+      "step": 5862
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444582457481369e-06,
+      "loss": 1.5204,
+      "step": 5863
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444486909994267e-06,
+      "loss": 1.4026,
+      "step": 5864
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444391362507167e-06,
+      "loss": 1.4394,
+      "step": 5865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444295815020065e-06,
+      "loss": 1.7461,
+      "step": 5866
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444200267532964e-06,
+      "loss": 0.7961,
+      "step": 5867
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444104720045863e-06,
+      "loss": 1.3148,
+      "step": 5868
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.444009172558762e-06,
+      "loss": 0.9393,
+      "step": 5869
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443913625071661e-06,
+      "loss": 1.7671,
+      "step": 5870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443818077584559e-06,
+      "loss": 1.1539,
+      "step": 5871
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443722530097459e-06,
+      "loss": 1.2638,
+      "step": 5872
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4436269826103575e-06,
+      "loss": 1.4686,
+      "step": 5873
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4435314351232565e-06,
+      "loss": 0.9529,
+      "step": 5874
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4434358876361556e-06,
+      "loss": 1.4091,
+      "step": 5875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443340340149055e-06,
+      "loss": 1.6023,
+      "step": 5876
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443244792661953e-06,
+      "loss": 0.6326,
+      "step": 5877
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443149245174852e-06,
+      "loss": 1.2244,
+      "step": 5878
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.443053697687752e-06,
+      "loss": 0.9914,
+      "step": 5879
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44295815020065e-06,
+      "loss": 1.5537,
+      "step": 5880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442862602713549e-06,
+      "loss": 1.3029,
+      "step": 5881
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442767055226448e-06,
+      "loss": 1.0127,
+      "step": 5882
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442671507739347e-06,
+      "loss": 0.8467,
+      "step": 5883
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442575960252245e-06,
+      "loss": 1.1684,
+      "step": 5884
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442480412765144e-06,
+      "loss": 1.2334,
+      "step": 5885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442384865278044e-06,
+      "loss": 1.2924,
+      "step": 5886
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442289317790942e-06,
+      "loss": 1.5435,
+      "step": 5887
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442193770303841e-06,
+      "loss": 0.9281,
+      "step": 5888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44209822281674e-06,
+      "loss": 0.9968,
+      "step": 5889
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4420026753296394e-06,
+      "loss": 1.0596,
+      "step": 5890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.441907127842538e-06,
+      "loss": 1.4802,
+      "step": 5891
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.441811580355437e-06,
+      "loss": 1.3865,
+      "step": 5892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4417160328683366e-06,
+      "loss": 1.2112,
+      "step": 5893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441620485381235e-06,
+      "loss": 1.6394,
+      "step": 5894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441524937894134e-06,
+      "loss": 1.3216,
+      "step": 5895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441429390407033e-06,
+      "loss": 1.5947,
+      "step": 5896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441333842919932e-06,
+      "loss": 1.8933,
+      "step": 5897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44123829543283e-06,
+      "loss": 1.1982,
+      "step": 5898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441142747945729e-06,
+      "loss": 1.4839,
+      "step": 5899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.441047200458628e-06,
+      "loss": 1.3255,
+      "step": 5900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440951652971527e-06,
+      "loss": 1.6871,
+      "step": 5901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440856105484426e-06,
+      "loss": 1.1494,
+      "step": 5902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440760557997325e-06,
+      "loss": 1.369,
+      "step": 5903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440665010510224e-06,
+      "loss": 0.8458,
+      "step": 5904
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4405694630231225e-06,
+      "loss": 1.5744,
+      "step": 5905
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4404739155360215e-06,
+      "loss": 1.3464,
+      "step": 5906
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4403783680489206e-06,
+      "loss": 1.5549,
+      "step": 5907
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44028282056182e-06,
+      "loss": 1.0122,
+      "step": 5908
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440187273074719e-06,
+      "loss": 1.1539,
+      "step": 5909
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.440091725587618e-06,
+      "loss": 1.2254,
+      "step": 5910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439996178100517e-06,
+      "loss": 1.2056,
+      "step": 5911
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439900630613415e-06,
+      "loss": 1.1701,
+      "step": 5912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439805083126314e-06,
+      "loss": 1.2277,
+      "step": 5913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439709535639213e-06,
+      "loss": 1.0413,
+      "step": 5914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439613988152112e-06,
+      "loss": 1.2213,
+      "step": 5915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439518440665011e-06,
+      "loss": 0.8734,
+      "step": 5916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.43942289317791e-06,
+      "loss": 1.5048,
+      "step": 5917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439327345690808e-06,
+      "loss": 1.1514,
+      "step": 5918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439231798203707e-06,
+      "loss": 0.9269,
+      "step": 5919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439136250716606e-06,
+      "loss": 1.1859,
+      "step": 5920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.439040703229505e-06,
+      "loss": 1.1064,
+      "step": 5921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4389451557424044e-06,
+      "loss": 1.4466,
+      "step": 5922
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4388496082553035e-06,
+      "loss": 1.2302,
+      "step": 5923
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4387540607682025e-06,
+      "loss": 1.0984,
+      "step": 5924
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438658513281101e-06,
+      "loss": 0.7954,
+      "step": 5925
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438562965794e-06,
+      "loss": 1.439,
+      "step": 5926
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438467418306899e-06,
+      "loss": 1.3873,
+      "step": 5927
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438371870819798e-06,
+      "loss": 0.9062,
+      "step": 5928
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438276323332697e-06,
+      "loss": 1.3878,
+      "step": 5929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438180775845596e-06,
+      "loss": 1.4785,
+      "step": 5930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.438085228358495e-06,
+      "loss": 1.0358,
+      "step": 5931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437989680871393e-06,
+      "loss": 0.9096,
+      "step": 5932
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437894133384292e-06,
+      "loss": 0.7618,
+      "step": 5933
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437798585897191e-06,
+      "loss": 1.1074,
+      "step": 5934
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.43770303841009e-06,
+      "loss": 1.8959,
+      "step": 5935
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4376074909229884e-06,
+      "loss": 1.4104,
+      "step": 5936
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437511943435888e-06,
+      "loss": 1.1815,
+      "step": 5937
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437416395948787e-06,
+      "loss": 0.9577,
+      "step": 5938
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4373208484616856e-06,
+      "loss": 1.3509,
+      "step": 5939
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437225300974585e-06,
+      "loss": 1.3203,
+      "step": 5940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437129753487484e-06,
+      "loss": 1.2038,
+      "step": 5941
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437034206000383e-06,
+      "loss": 1.0658,
+      "step": 5942
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.436938658513281e-06,
+      "loss": 1.1109,
+      "step": 5943
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.436843111026181e-06,
+      "loss": 1.3121,
+      "step": 5944
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.43674756353908e-06,
+      "loss": 0.9919,
+      "step": 5945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436652016051978e-06,
+      "loss": 1.2235,
+      "step": 5946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436556468564877e-06,
+      "loss": 1.1875,
+      "step": 5947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436460921077776e-06,
+      "loss": 1.5227,
+      "step": 5948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436365373590675e-06,
+      "loss": 0.7582,
+      "step": 5949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436269826103573e-06,
+      "loss": 1.9681,
+      "step": 5950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436174278616473e-06,
+      "loss": 0.981,
+      "step": 5951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436078731129372e-06,
+      "loss": 0.8837,
+      "step": 5952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43598318364227e-06,
+      "loss": 1.3528,
+      "step": 5953
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4358876361551694e-06,
+      "loss": 0.6898,
+      "step": 5954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4357920886680685e-06,
+      "loss": 1.1359,
+      "step": 5955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4356965411809675e-06,
+      "loss": 1.0266,
+      "step": 5956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435600993693866e-06,
+      "loss": 0.8527,
+      "step": 5957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435505446206766e-06,
+      "loss": 1.0092,
+      "step": 5958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435409898719664e-06,
+      "loss": 1.2691,
+      "step": 5959
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435314351232563e-06,
+      "loss": 1.2088,
+      "step": 5960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435218803745462e-06,
+      "loss": 1.9392,
+      "step": 5961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435123256258361e-06,
+      "loss": 0.9992,
+      "step": 5962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43502770877126e-06,
+      "loss": 1.3368,
+      "step": 5963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434932161284158e-06,
+      "loss": 1.2212,
+      "step": 5964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434836613797058e-06,
+      "loss": 0.9895,
+      "step": 5965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434741066309956e-06,
+      "loss": 1.055,
+      "step": 5966
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434645518822855e-06,
+      "loss": 0.9438,
+      "step": 5967
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434549971335754e-06,
+      "loss": 1.6782,
+      "step": 5968
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434454423848653e-06,
+      "loss": 1.4976,
+      "step": 5969
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4343588763615515e-06,
+      "loss": 1.6812,
+      "step": 5970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4342633288744506e-06,
+      "loss": 1.008,
+      "step": 5971
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43416778138735e-06,
+      "loss": 1.2702,
+      "step": 5972
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434072233900249e-06,
+      "loss": 1.5839,
+      "step": 5973
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433976686413148e-06,
+      "loss": 1.4698,
+      "step": 5974
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433881138926047e-06,
+      "loss": 1.5161,
+      "step": 5975
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433785591438946e-06,
+      "loss": 1.7458,
+      "step": 5976
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433690043951844e-06,
+      "loss": 1.5811,
+      "step": 5977
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433594496464743e-06,
+      "loss": 1.0498,
+      "step": 5978
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433498948977642e-06,
+      "loss": 1.3179,
+      "step": 5979
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433403401490541e-06,
+      "loss": 0.9964,
+      "step": 5980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43330785400344e-06,
+      "loss": 1.3474,
+      "step": 5981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433212306516339e-06,
+      "loss": 1.8644,
+      "step": 5982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433116759029238e-06,
+      "loss": 1.5053,
+      "step": 5983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433021211542136e-06,
+      "loss": 1.55,
+      "step": 5984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432925664055035e-06,
+      "loss": 1.5165,
+      "step": 5985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4328301165679344e-06,
+      "loss": 1.7973,
+      "step": 5986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4327345690808335e-06,
+      "loss": 1.4121,
+      "step": 5987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4326390215937325e-06,
+      "loss": 1.4505,
+      "step": 5988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4325434741066316e-06,
+      "loss": 1.2065,
+      "step": 5989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432447926619531e-06,
+      "loss": 1.5006,
+      "step": 5990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432352379132429e-06,
+      "loss": 1.233,
+      "step": 5991
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432256831645328e-06,
+      "loss": 1.6054,
+      "step": 5992
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432161284158227e-06,
+      "loss": 1.3095,
+      "step": 5993
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432065736671126e-06,
+      "loss": 1.2422,
+      "step": 5994
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431970189184025e-06,
+      "loss": 1.4394,
+      "step": 5995
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431874641696924e-06,
+      "loss": 0.9123,
+      "step": 5996
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431779094209823e-06,
+      "loss": 1.2209,
+      "step": 5997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431683546722721e-06,
+      "loss": 1.2434,
+      "step": 5998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.43158799923562e-06,
+      "loss": 1.0866,
+      "step": 5999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431492451748519e-06,
+      "loss": 1.1384,
+      "step": 6000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431396904261418e-06,
+      "loss": 1.3222,
+      "step": 6001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431301356774317e-06,
+      "loss": 1.2106,
+      "step": 6002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431205809287216e-06,
+      "loss": 1.2443,
+      "step": 6003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4311102618001154e-06,
+      "loss": 1.1706,
+      "step": 6004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.431014714313014e-06,
+      "loss": 1.1987,
+      "step": 6005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430919166825913e-06,
+      "loss": 0.9306,
+      "step": 6006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430823619338812e-06,
+      "loss": 1.3655,
+      "step": 6007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430728071851711e-06,
+      "loss": 1.5962,
+      "step": 6008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.43063252436461e-06,
+      "loss": 1.4392,
+      "step": 6009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430536976877509e-06,
+      "loss": 1.8033,
+      "step": 6010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430441429390407e-06,
+      "loss": 1.193,
+      "step": 6011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430345881903306e-06,
+      "loss": 0.8451,
+      "step": 6012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430250334416205e-06,
+      "loss": 0.8903,
+      "step": 6013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430154786929104e-06,
+      "loss": 1.5342,
+      "step": 6014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.430059239442003e-06,
+      "loss": 0.9866,
+      "step": 6015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429963691954902e-06,
+      "loss": 1.2539,
+      "step": 6016
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429868144467801e-06,
+      "loss": 1.1658,
+      "step": 6017
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4297725969806994e-06,
+      "loss": 1.3175,
+      "step": 6018
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4296770494935985e-06,
+      "loss": 0.8409,
+      "step": 6019
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4295815020064975e-06,
+      "loss": 1.3353,
+      "step": 6020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4294859545193966e-06,
+      "loss": 1.1509,
+      "step": 6021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429390407032295e-06,
+      "loss": 1.3411,
+      "step": 6022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429294859545195e-06,
+      "loss": 1.2306,
+      "step": 6023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429199312058094e-06,
+      "loss": 1.5347,
+      "step": 6024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429103764570992e-06,
+      "loss": 1.2907,
+      "step": 6025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.429008217083891e-06,
+      "loss": 1.3462,
+      "step": 6026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.42891266959679e-06,
+      "loss": 1.0328,
+      "step": 6027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428817122109689e-06,
+      "loss": 1.4284,
+      "step": 6028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428721574622587e-06,
+      "loss": 1.5567,
+      "step": 6029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428626027135487e-06,
+      "loss": 1.1112,
+      "step": 6030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428530479648386e-06,
+      "loss": 1.1208,
+      "step": 6031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428434932161284e-06,
+      "loss": 1.3818,
+      "step": 6032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428339384674183e-06,
+      "loss": 1.4425,
+      "step": 6033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428243837187082e-06,
+      "loss": 1.223,
+      "step": 6034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.428148289699981e-06,
+      "loss": 1.5604,
+      "step": 6035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.42805274221288e-06,
+      "loss": 1.2153,
+      "step": 6036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4279571947257795e-06,
+      "loss": 0.9785,
+      "step": 6037
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4278616472386785e-06,
+      "loss": 0.8422,
+      "step": 6038
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427766099751577e-06,
+      "loss": 1.049,
+      "step": 6039
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427670552264476e-06,
+      "loss": 1.1941,
+      "step": 6040
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427575004777375e-06,
+      "loss": 1.0058,
+      "step": 6041
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427479457290274e-06,
+      "loss": 0.9053,
+      "step": 6042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427383909803172e-06,
+      "loss": 0.9958,
+      "step": 6043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427288362316071e-06,
+      "loss": 1.5324,
+      "step": 6044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427192814828971e-06,
+      "loss": 0.8436,
+      "step": 6045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427097267341869e-06,
+      "loss": 0.9413,
+      "step": 6046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427001719854768e-06,
+      "loss": 1.3831,
+      "step": 6047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.426906172367667e-06,
+      "loss": 1.109,
+      "step": 6048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.426810624880566e-06,
+      "loss": 1.3464,
+      "step": 6049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4267150773934644e-06,
+      "loss": 1.4395,
+      "step": 6050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4266195299063635e-06,
+      "loss": 1.3636,
+      "step": 6051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.426523982419263e-06,
+      "loss": 1.5632,
+      "step": 6052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4264284349321616e-06,
+      "loss": 0.8661,
+      "step": 6053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.426332887445061e-06,
+      "loss": 1.3915,
+      "step": 6054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42623733995796e-06,
+      "loss": 1.1923,
+      "step": 6055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.426141792470859e-06,
+      "loss": 1.2303,
+      "step": 6056
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.426046244983757e-06,
+      "loss": 1.5833,
+      "step": 6057
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425950697496656e-06,
+      "loss": 1.6491,
+      "step": 6058
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425855150009555e-06,
+      "loss": 1.2852,
+      "step": 6059
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425759602522454e-06,
+      "loss": 1.1835,
+      "step": 6060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425664055035353e-06,
+      "loss": 1.1613,
+      "step": 6061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425568507548252e-06,
+      "loss": 1.366,
+      "step": 6062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425472960061151e-06,
+      "loss": 1.6408,
+      "step": 6063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425377412574049e-06,
+      "loss": 1.2175,
+      "step": 6064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425281865086948e-06,
+      "loss": 0.9921,
+      "step": 6065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425186317599847e-06,
+      "loss": 1.1017,
+      "step": 6066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.425090770112746e-06,
+      "loss": 1.1095,
+      "step": 6067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4249952226256454e-06,
+      "loss": 1.1671,
+      "step": 6068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4248996751385445e-06,
+      "loss": 1.1637,
+      "step": 6069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424804127651443e-06,
+      "loss": 1.3725,
+      "step": 6070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424708580164342e-06,
+      "loss": 1.0071,
+      "step": 6071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424613032677241e-06,
+      "loss": 1.222,
+      "step": 6072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42451748519014e-06,
+      "loss": 1.68,
+      "step": 6073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424421937703039e-06,
+      "loss": 1.0083,
+      "step": 6074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424326390215938e-06,
+      "loss": 1.5758,
+      "step": 6075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424230842728837e-06,
+      "loss": 1.0902,
+      "step": 6076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424135295241735e-06,
+      "loss": 0.8474,
+      "step": 6077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.424039747754634e-06,
+      "loss": 1.0913,
+      "step": 6078
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423944200267533e-06,
+      "loss": 1.2134,
+      "step": 6079
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423848652780432e-06,
+      "loss": 1.562,
+      "step": 6080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423753105293331e-06,
+      "loss": 1.1529,
+      "step": 6081
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42365755780623e-06,
+      "loss": 1.0133,
+      "step": 6082
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423562010319129e-06,
+      "loss": 1.1305,
+      "step": 6083
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4234664628320275e-06,
+      "loss": 1.6452,
+      "step": 6084
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4233709153449266e-06,
+      "loss": 1.6192,
+      "step": 6085
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423275367857826e-06,
+      "loss": 1.4306,
+      "step": 6086
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423179820370725e-06,
+      "loss": 1.6009,
+      "step": 6087
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.423084272883624e-06,
+      "loss": 1.213,
+      "step": 6088
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422988725396523e-06,
+      "loss": 1.3097,
+      "step": 6089
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422893177909422e-06,
+      "loss": 0.8217,
+      "step": 6090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.42279763042232e-06,
+      "loss": 0.8372,
+      "step": 6091
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422702082935219e-06,
+      "loss": 1.1976,
+      "step": 6092
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422606535448118e-06,
+      "loss": 1.4185,
+      "step": 6093
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422510987961017e-06,
+      "loss": 1.2016,
+      "step": 6094
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422415440473916e-06,
+      "loss": 1.1198,
+      "step": 6095
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422319892986815e-06,
+      "loss": 1.5833,
+      "step": 6096
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422224345499714e-06,
+      "loss": 1.3849,
+      "step": 6097
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422128798012612e-06,
+      "loss": 1.0797,
+      "step": 6098
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422033250525511e-06,
+      "loss": 0.9312,
+      "step": 6099
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4219377030384104e-06,
+      "loss": 1.2794,
+      "step": 6100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4218421555513095e-06,
+      "loss": 1.2633,
+      "step": 6101
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4217466080642085e-06,
+      "loss": 0.9472,
+      "step": 6102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4216510605771076e-06,
+      "loss": 0.8132,
+      "step": 6103
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421555513090007e-06,
+      "loss": 1.1384,
+      "step": 6104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421459965602905e-06,
+      "loss": 1.0442,
+      "step": 6105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421364418115804e-06,
+      "loss": 0.892,
+      "step": 6106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421268870628703e-06,
+      "loss": 1.7143,
+      "step": 6107
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421173323141602e-06,
+      "loss": 1.6421,
+      "step": 6108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.421077775654501e-06,
+      "loss": 1.046,
+      "step": 6109
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4209822281674e-06,
+      "loss": 1.0932,
+      "step": 6110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420886680680298e-06,
+      "loss": 1.4498,
+      "step": 6111
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420791133193197e-06,
+      "loss": 1.4509,
+      "step": 6112
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420695585706096e-06,
+      "loss": 1.073,
+      "step": 6113
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420600038218995e-06,
+      "loss": 1.4385,
+      "step": 6114
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420504490731894e-06,
+      "loss": 1.0544,
+      "step": 6115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4204089432447925e-06,
+      "loss": 1.0702,
+      "step": 6116
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420313395757692e-06,
+      "loss": 1.093,
+      "step": 6117
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420217848270591e-06,
+      "loss": 1.1647,
+      "step": 6118
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.42012230078349e-06,
+      "loss": 1.4054,
+      "step": 6119
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.420026753296389e-06,
+      "loss": 1.2898,
+      "step": 6120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419931205809288e-06,
+      "loss": 1.2139,
+      "step": 6121
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419835658322186e-06,
+      "loss": 1.5275,
+      "step": 6122
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419740110835085e-06,
+      "loss": 0.9511,
+      "step": 6123
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419644563347985e-06,
+      "loss": 1.0886,
+      "step": 6124
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419549015860883e-06,
+      "loss": 1.0321,
+      "step": 6125
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419453468373782e-06,
+      "loss": 0.9558,
+      "step": 6126
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419357920886681e-06,
+      "loss": 1.12,
+      "step": 6127
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.41926237339958e-06,
+      "loss": 1.4095,
+      "step": 6128
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419166825912478e-06,
+      "loss": 1.639,
+      "step": 6129
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.419071278425377e-06,
+      "loss": 1.1268,
+      "step": 6130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.418975730938277e-06,
+      "loss": 1.3803,
+      "step": 6131
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4188801834511754e-06,
+      "loss": 1.1109,
+      "step": 6132
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4187846359640745e-06,
+      "loss": 1.7959,
+      "step": 6133
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4186890884769735e-06,
+      "loss": 1.0434,
+      "step": 6134
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4185935409898726e-06,
+      "loss": 1.0446,
+      "step": 6135
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.418497993502771e-06,
+      "loss": 1.4517,
+      "step": 6136
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.41840244601567e-06,
+      "loss": 1.3794,
+      "step": 6137
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.41830689852857e-06,
+      "loss": 1.4625,
+      "step": 6138
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.418211351041468e-06,
+      "loss": 1.4783,
+      "step": 6139
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.418115803554367e-06,
+      "loss": 1.393,
+      "step": 6140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.418020256067266e-06,
+      "loss": 0.8973,
+      "step": 6141
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417924708580165e-06,
+      "loss": 1.8438,
+      "step": 6142
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417829161093063e-06,
+      "loss": 1.0394,
+      "step": 6143
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417733613605962e-06,
+      "loss": 1.2128,
+      "step": 6144
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417638066118862e-06,
+      "loss": 1.2247,
+      "step": 6145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.41754251863176e-06,
+      "loss": 0.6479,
+      "step": 6146
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417446971144659e-06,
+      "loss": 1.089,
+      "step": 6147
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417351423657558e-06,
+      "loss": 0.9091,
+      "step": 6148
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417255876170457e-06,
+      "loss": 1.6268,
+      "step": 6149
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417160328683356e-06,
+      "loss": 0.8457,
+      "step": 6150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417064781196255e-06,
+      "loss": 0.8088,
+      "step": 6151
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416969233709154e-06,
+      "loss": 1.0584,
+      "step": 6152
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416873686222053e-06,
+      "loss": 1.2678,
+      "step": 6153
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416778138734952e-06,
+      "loss": 1.1347,
+      "step": 6154
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416682591247851e-06,
+      "loss": 1.6762,
+      "step": 6155
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41658704376075e-06,
+      "loss": 1.3512,
+      "step": 6156
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416491496273648e-06,
+      "loss": 1.526,
+      "step": 6157
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416395948786547e-06,
+      "loss": 1.172,
+      "step": 6158
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416300401299446e-06,
+      "loss": 1.9687,
+      "step": 6159
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416204853812345e-06,
+      "loss": 1.134,
+      "step": 6160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416109306325244e-06,
+      "loss": 0.7895,
+      "step": 6161
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.416013758838143e-06,
+      "loss": 1.1674,
+      "step": 6162
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415918211351041e-06,
+      "loss": 0.9632,
+      "step": 6163
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4158226638639404e-06,
+      "loss": 1.3055,
+      "step": 6164
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4157271163768395e-06,
+      "loss": 1.0792,
+      "step": 6165
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4156315688897385e-06,
+      "loss": 1.1586,
+      "step": 6166
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4155360214026376e-06,
+      "loss": 1.0993,
+      "step": 6167
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415440473915537e-06,
+      "loss": 1.3346,
+      "step": 6168
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415344926428436e-06,
+      "loss": 1.5589,
+      "step": 6169
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415249378941334e-06,
+      "loss": 1.1705,
+      "step": 6170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415153831454233e-06,
+      "loss": 0.669,
+      "step": 6171
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.415058283967132e-06,
+      "loss": 1.42,
+      "step": 6172
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414962736480031e-06,
+      "loss": 1.5981,
+      "step": 6173
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41486718899293e-06,
+      "loss": 1.3334,
+      "step": 6174
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414771641505829e-06,
+      "loss": 1.3145,
+      "step": 6175
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414676094018728e-06,
+      "loss": 1.2979,
+      "step": 6176
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414580546531626e-06,
+      "loss": 1.3614,
+      "step": 6177
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414484999044525e-06,
+      "loss": 1.4199,
+      "step": 6178
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414389451557424e-06,
+      "loss": 1.2188,
+      "step": 6179
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414293904070323e-06,
+      "loss": 0.9978,
+      "step": 6180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.414198356583222e-06,
+      "loss": 1.3084,
+      "step": 6181
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4141028090961214e-06,
+      "loss": 1.4341,
+      "step": 6182
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4140072616090205e-06,
+      "loss": 1.287,
+      "step": 6183
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413911714121919e-06,
+      "loss": 1.3233,
+      "step": 6184
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413816166634818e-06,
+      "loss": 0.872,
+      "step": 6185
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413720619147717e-06,
+      "loss": 1.2106,
+      "step": 6186
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413625071660616e-06,
+      "loss": 1.2087,
+      "step": 6187
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413529524173514e-06,
+      "loss": 1.0404,
+      "step": 6188
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413433976686414e-06,
+      "loss": 1.7011,
+      "step": 6189
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413338429199313e-06,
+      "loss": 1.4906,
+      "step": 6190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413242881712211e-06,
+      "loss": 0.8669,
+      "step": 6191
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.41314733422511e-06,
+      "loss": 1.6243,
+      "step": 6192
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.413051786738009e-06,
+      "loss": 0.8309,
+      "step": 6193
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412956239250908e-06,
+      "loss": 1.5405,
+      "step": 6194
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412860691763806e-06,
+      "loss": 1.1181,
+      "step": 6195
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412765144276706e-06,
+      "loss": 1.4409,
+      "step": 6196
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412669596789605e-06,
+      "loss": 1.1344,
+      "step": 6197
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4125740493025035e-06,
+      "loss": 1.0331,
+      "step": 6198
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4124785018154026e-06,
+      "loss": 1.3754,
+      "step": 6199
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412382954328302e-06,
+      "loss": 1.5785,
+      "step": 6200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412287406841201e-06,
+      "loss": 0.8296,
+      "step": 6201
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412191859354099e-06,
+      "loss": 1.7286,
+      "step": 6202
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412096311866999e-06,
+      "loss": 1.135,
+      "step": 6203
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.412000764379898e-06,
+      "loss": 0.912,
+      "step": 6204
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.411905216892796e-06,
+      "loss": 1.5043,
+      "step": 6205
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.411809669405695e-06,
+      "loss": 1.3155,
+      "step": 6206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.411714121918594e-06,
+      "loss": 1.2085,
+      "step": 6207
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411618574431493e-06,
+      "loss": 0.9673,
+      "step": 6208
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411523026944391e-06,
+      "loss": 0.6229,
+      "step": 6209
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411427479457291e-06,
+      "loss": 0.8285,
+      "step": 6210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411331931970189e-06,
+      "loss": 0.9811,
+      "step": 6211
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411236384483088e-06,
+      "loss": 0.9587,
+      "step": 6212
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.411140836995987e-06,
+      "loss": 1.1713,
+      "step": 6213
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4110452895088864e-06,
+      "loss": 1.1423,
+      "step": 6214
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410949742021785e-06,
+      "loss": 1.2511,
+      "step": 6215
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410854194534684e-06,
+      "loss": 1.6164,
+      "step": 6216
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4107586470475836e-06,
+      "loss": 1.6459,
+      "step": 6217
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410663099560482e-06,
+      "loss": 1.2191,
+      "step": 6218
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410567552073381e-06,
+      "loss": 1.1804,
+      "step": 6219
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.41047200458628e-06,
+      "loss": 1.4364,
+      "step": 6220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410376457099179e-06,
+      "loss": 0.8616,
+      "step": 6221
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410280909612077e-06,
+      "loss": 1.1719,
+      "step": 6222
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410185362124976e-06,
+      "loss": 1.481,
+      "step": 6223
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.410089814637875e-06,
+      "loss": 1.1691,
+      "step": 6224
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409994267150774e-06,
+      "loss": 0.8454,
+      "step": 6225
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409898719663673e-06,
+      "loss": 1.2854,
+      "step": 6226
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409803172176572e-06,
+      "loss": 0.9112,
+      "step": 6227
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409707624689471e-06,
+      "loss": 1.1607,
+      "step": 6228
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4096120772023695e-06,
+      "loss": 0.8982,
+      "step": 6229
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4095165297152685e-06,
+      "loss": 1.4888,
+      "step": 6230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4094209822281676e-06,
+      "loss": 0.9986,
+      "step": 6231
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409325434741067e-06,
+      "loss": 1.742,
+      "step": 6232
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409229887253966e-06,
+      "loss": 0.9273,
+      "step": 6233
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409134339766865e-06,
+      "loss": 1.4038,
+      "step": 6234
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.409038792279764e-06,
+      "loss": 0.7646,
+      "step": 6235
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408943244792662e-06,
+      "loss": 1.1359,
+      "step": 6236
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408847697305561e-06,
+      "loss": 1.7161,
+      "step": 6237
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.40875214981846e-06,
+      "loss": 0.9862,
+      "step": 6238
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408656602331359e-06,
+      "loss": 1.0086,
+      "step": 6239
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408561054844258e-06,
+      "loss": 1.3208,
+      "step": 6240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408465507357157e-06,
+      "loss": 1.176,
+      "step": 6241
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408369959870056e-06,
+      "loss": 1.278,
+      "step": 6242
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408274412382954e-06,
+      "loss": 0.9858,
+      "step": 6243
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408178864895853e-06,
+      "loss": 0.9467,
+      "step": 6244
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.408083317408752e-06,
+      "loss": 1.1544,
+      "step": 6245
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4079877699216514e-06,
+      "loss": 0.9507,
+      "step": 6246
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4078922224345505e-06,
+      "loss": 0.7383,
+      "step": 6247
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4077966749474495e-06,
+      "loss": 1.3067,
+      "step": 6248
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4077011274603486e-06,
+      "loss": 1.5542,
+      "step": 6249
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407605579973247e-06,
+      "loss": 1.2966,
+      "step": 6250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407510032486146e-06,
+      "loss": 1.2052,
+      "step": 6251
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407414484999045e-06,
+      "loss": 1.437,
+      "step": 6252
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407318937511944e-06,
+      "loss": 2.0748,
+      "step": 6253
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407223390024843e-06,
+      "loss": 1.2867,
+      "step": 6254
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407127842537742e-06,
+      "loss": 1.3119,
+      "step": 6255
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.407032295050641e-06,
+      "loss": 1.0657,
+      "step": 6256
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406936747563539e-06,
+      "loss": 1.0578,
+      "step": 6257
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406841200076438e-06,
+      "loss": 1.1916,
+      "step": 6258
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406745652589337e-06,
+      "loss": 1.1878,
+      "step": 6259
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.406650105102236e-06,
+      "loss": 1.3179,
+      "step": 6260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.406554557615135e-06,
+      "loss": 1.2217,
+      "step": 6261
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.406459010128034e-06,
+      "loss": 1.4086,
+      "step": 6262
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4063634626409326e-06,
+      "loss": 1.2202,
+      "step": 6263
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.406267915153832e-06,
+      "loss": 1.1353,
+      "step": 6264
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.406172367666731e-06,
+      "loss": 1.4048,
+      "step": 6265
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40607682017963e-06,
+      "loss": 1.1511,
+      "step": 6266
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405981272692529e-06,
+      "loss": 1.4369,
+      "step": 6267
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405885725205428e-06,
+      "loss": 0.9574,
+      "step": 6268
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405790177718327e-06,
+      "loss": 1.4781,
+      "step": 6269
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405694630231225e-06,
+      "loss": 1.5953,
+      "step": 6270
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405599082744124e-06,
+      "loss": 1.063,
+      "step": 6271
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405503535257023e-06,
+      "loss": 1.3601,
+      "step": 6272
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405407987769922e-06,
+      "loss": 1.1943,
+      "step": 6273
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40531244028282e-06,
+      "loss": 1.159,
+      "step": 6274
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40521689279572e-06,
+      "loss": 1.609,
+      "step": 6275
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405121345308619e-06,
+      "loss": 1.3156,
+      "step": 6276
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.405025797821517e-06,
+      "loss": 1.5013,
+      "step": 6277
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4049302503344164e-06,
+      "loss": 1.2453,
+      "step": 6278
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4048347028473155e-06,
+      "loss": 1.7498,
+      "step": 6279
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4047391553602145e-06,
+      "loss": 1.5696,
+      "step": 6280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404643607873113e-06,
+      "loss": 1.2806,
+      "step": 6281
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404548060386013e-06,
+      "loss": 1.4676,
+      "step": 6282
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404452512898912e-06,
+      "loss": 1.2454,
+      "step": 6283
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40435696541181e-06,
+      "loss": 1.2216,
+      "step": 6284
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404261417924709e-06,
+      "loss": 1.5912,
+      "step": 6285
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404165870437608e-06,
+      "loss": 0.8647,
+      "step": 6286
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.404070322950507e-06,
+      "loss": 1.2016,
+      "step": 6287
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403974775463405e-06,
+      "loss": 1.0783,
+      "step": 6288
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403879227976305e-06,
+      "loss": 1.4731,
+      "step": 6289
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403783680489204e-06,
+      "loss": 0.7186,
+      "step": 6290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403688133002102e-06,
+      "loss": 0.8185,
+      "step": 6291
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403592585515001e-06,
+      "loss": 0.8661,
+      "step": 6292
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4034970380279e-06,
+      "loss": 1.3728,
+      "step": 6293
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403401490540799e-06,
+      "loss": 1.203,
+      "step": 6294
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4033059430536976e-06,
+      "loss": 1.1259,
+      "step": 6295
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403210395566597e-06,
+      "loss": 1.4332,
+      "step": 6296
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4031148480794965e-06,
+      "loss": 1.622,
+      "step": 6297
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.403019300592395e-06,
+      "loss": 1.2281,
+      "step": 6298
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402923753105294e-06,
+      "loss": 1.6466,
+      "step": 6299
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402828205618193e-06,
+      "loss": 0.7441,
+      "step": 6300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402732658131092e-06,
+      "loss": 1.3866,
+      "step": 6301
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.40263711064399e-06,
+      "loss": 1.1271,
+      "step": 6302
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402541563156889e-06,
+      "loss": 1.1844,
+      "step": 6303
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402446015669788e-06,
+      "loss": 1.6075,
+      "step": 6304
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402350468182687e-06,
+      "loss": 0.9646,
+      "step": 6305
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402254920695586e-06,
+      "loss": 0.9404,
+      "step": 6306
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402159373208485e-06,
+      "loss": 1.0849,
+      "step": 6307
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.402063825721384e-06,
+      "loss": 1.2979,
+      "step": 6308
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.401968278234282e-06,
+      "loss": 0.9909,
+      "step": 6309
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4018727307471814e-06,
+      "loss": 1.2225,
+      "step": 6310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4017771832600805e-06,
+      "loss": 1.4984,
+      "step": 6311
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4016816357729795e-06,
+      "loss": 1.204,
+      "step": 6312
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4015860882858786e-06,
+      "loss": 1.0794,
+      "step": 6313
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401490540798778e-06,
+      "loss": 1.1408,
+      "step": 6314
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401394993311676e-06,
+      "loss": 1.5935,
+      "step": 6315
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401299445824575e-06,
+      "loss": 1.305,
+      "step": 6316
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401203898337474e-06,
+      "loss": 1.6857,
+      "step": 6317
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401108350850373e-06,
+      "loss": 0.5233,
+      "step": 6318
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.401012803363272e-06,
+      "loss": 1.1111,
+      "step": 6319
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400917255876171e-06,
+      "loss": 0.8662,
+      "step": 6320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.40082170838907e-06,
+      "loss": 1.5722,
+      "step": 6321
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400726160901968e-06,
+      "loss": 1.3261,
+      "step": 6322
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400630613414867e-06,
+      "loss": 1.3246,
+      "step": 6323
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400535065927766e-06,
+      "loss": 1.1881,
+      "step": 6324
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400439518440665e-06,
+      "loss": 1.3008,
+      "step": 6325
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400343970953564e-06,
+      "loss": 0.9996,
+      "step": 6326
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400248423466463e-06,
+      "loss": 1.2321,
+      "step": 6327
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4001528759793624e-06,
+      "loss": 1.1336,
+      "step": 6328
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.400057328492261e-06,
+      "loss": 0.8205,
+      "step": 6329
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.39996178100516e-06,
+      "loss": 1.5143,
+      "step": 6330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399866233518059e-06,
+      "loss": 0.8939,
+      "step": 6331
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399770686030958e-06,
+      "loss": 1.0209,
+      "step": 6332
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399675138543857e-06,
+      "loss": 0.9268,
+      "step": 6333
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399579591056756e-06,
+      "loss": 1.1986,
+      "step": 6334
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399484043569655e-06,
+      "loss": 1.0218,
+      "step": 6335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399388496082553e-06,
+      "loss": 1.574,
+      "step": 6336
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399292948595452e-06,
+      "loss": 1.1959,
+      "step": 6337
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399197401108351e-06,
+      "loss": 0.8805,
+      "step": 6338
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.39910185362125e-06,
+      "loss": 1.6746,
+      "step": 6339
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.399006306134149e-06,
+      "loss": 1.1581,
+      "step": 6340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398910758647048e-06,
+      "loss": 1.3142,
+      "step": 6341
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398815211159947e-06,
+      "loss": 1.4947,
+      "step": 6342
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3987196636728455e-06,
+      "loss": 1.3567,
+      "step": 6343
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3986241161857445e-06,
+      "loss": 1.4774,
+      "step": 6344
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3985285686986436e-06,
+      "loss": 1.4715,
+      "step": 6345
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398433021211543e-06,
+      "loss": 1.0435,
+      "step": 6346
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398337473724442e-06,
+      "loss": 0.9372,
+      "step": 6347
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398241926237341e-06,
+      "loss": 1.3351,
+      "step": 6348
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.39814637875024e-06,
+      "loss": 1.1924,
+      "step": 6349
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.398050831263138e-06,
+      "loss": 1.221,
+      "step": 6350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397955283776037e-06,
+      "loss": 1.3655,
+      "step": 6351
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397859736288936e-06,
+      "loss": 1.0285,
+      "step": 6352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397764188801835e-06,
+      "loss": 1.2075,
+      "step": 6353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397668641314734e-06,
+      "loss": 1.2776,
+      "step": 6354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397573093827633e-06,
+      "loss": 1.1284,
+      "step": 6355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397477546340531e-06,
+      "loss": 1.0167,
+      "step": 6356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.39738199885343e-06,
+      "loss": 1.2259,
+      "step": 6357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397286451366329e-06,
+      "loss": 1.5107,
+      "step": 6358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.397190903879228e-06,
+      "loss": 1.5597,
+      "step": 6359
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3970953563921274e-06,
+      "loss": 1.4144,
+      "step": 6360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3969998089050265e-06,
+      "loss": 0.8577,
+      "step": 6361
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3969042614179255e-06,
+      "loss": 1.2262,
+      "step": 6362
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396808713930824e-06,
+      "loss": 0.9026,
+      "step": 6363
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396713166443723e-06,
+      "loss": 1.1933,
+      "step": 6364
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396617618956622e-06,
+      "loss": 1.0914,
+      "step": 6365
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396522071469521e-06,
+      "loss": 0.4539,
+      "step": 6366
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396426523982419e-06,
+      "loss": 0.8097,
+      "step": 6367
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396330976495318e-06,
+      "loss": 1.3739,
+      "step": 6368
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396235429008218e-06,
+      "loss": 0.937,
+      "step": 6369
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396139881521116e-06,
+      "loss": 1.1343,
+      "step": 6370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.396044334034015e-06,
+      "loss": 1.2478,
+      "step": 6371
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395948786546914e-06,
+      "loss": 0.9635,
+      "step": 6372
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395853239059813e-06,
+      "loss": 0.8162,
+      "step": 6373
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3957576915727114e-06,
+      "loss": 1.453,
+      "step": 6374
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3956621440856105e-06,
+      "loss": 1.3901,
+      "step": 6375
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39556659659851e-06,
+      "loss": 1.1516,
+      "step": 6376
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3954710491114086e-06,
+      "loss": 1.4778,
+      "step": 6377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395375501624308e-06,
+      "loss": 0.8954,
+      "step": 6378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395279954137207e-06,
+      "loss": 1.3433,
+      "step": 6379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395184406650106e-06,
+      "loss": 1.1248,
+      "step": 6380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.395088859163004e-06,
+      "loss": 1.0961,
+      "step": 6381
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394993311675903e-06,
+      "loss": 1.2453,
+      "step": 6382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394897764188803e-06,
+      "loss": 1.5653,
+      "step": 6383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394802216701701e-06,
+      "loss": 1.1993,
+      "step": 6384
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3947066692146e-06,
+      "loss": 1.1903,
+      "step": 6385
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394611121727499e-06,
+      "loss": 0.9553,
+      "step": 6386
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394515574240398e-06,
+      "loss": 1.4938,
+      "step": 6387
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394420026753296e-06,
+      "loss": 1.3587,
+      "step": 6388
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394324479266195e-06,
+      "loss": 1.412,
+      "step": 6389
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394228931779095e-06,
+      "loss": 1.211,
+      "step": 6390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.394133384291993e-06,
+      "loss": 1.1566,
+      "step": 6391
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3940378368048924e-06,
+      "loss": 1.227,
+      "step": 6392
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3939422893177915e-06,
+      "loss": 0.8453,
+      "step": 6393
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3938467418306905e-06,
+      "loss": 1.293,
+      "step": 6394
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393751194343589e-06,
+      "loss": 1.0964,
+      "step": 6395
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393655646856488e-06,
+      "loss": 0.9024,
+      "step": 6396
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393560099369388e-06,
+      "loss": 1.5073,
+      "step": 6397
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393464551882286e-06,
+      "loss": 0.9515,
+      "step": 6398
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393369004395185e-06,
+      "loss": 1.1959,
+      "step": 6399
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393273456908084e-06,
+      "loss": 0.7541,
+      "step": 6400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393177909420983e-06,
+      "loss": 0.8672,
+      "step": 6401
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.393082361933881e-06,
+      "loss": 0.9086,
+      "step": 6402
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39298681444678e-06,
+      "loss": 1.121,
+      "step": 6403
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392891266959679e-06,
+      "loss": 1.3726,
+      "step": 6404
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392795719472578e-06,
+      "loss": 1.3584,
+      "step": 6405
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392700171985477e-06,
+      "loss": 0.7575,
+      "step": 6406
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392604624498376e-06,
+      "loss": 1.1954,
+      "step": 6407
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3925090770112745e-06,
+      "loss": 1.2048,
+      "step": 6408
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3924135295241736e-06,
+      "loss": 1.2274,
+      "step": 6409
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392317982037073e-06,
+      "loss": 1.3567,
+      "step": 6410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392222434549972e-06,
+      "loss": 1.2448,
+      "step": 6411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392126887062871e-06,
+      "loss": 1.1552,
+      "step": 6412
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39203133957577e-06,
+      "loss": 0.9756,
+      "step": 6413
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.391935792088669e-06,
+      "loss": 1.2293,
+      "step": 6414
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.391840244601567e-06,
+      "loss": 1.1054,
+      "step": 6415
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.391744697114466e-06,
+      "loss": 1.2964,
+      "step": 6416
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391649149627365e-06,
+      "loss": 1.5509,
+      "step": 6417
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391553602140264e-06,
+      "loss": 1.6729,
+      "step": 6418
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391458054653163e-06,
+      "loss": 1.5099,
+      "step": 6419
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391362507166062e-06,
+      "loss": 1.315,
+      "step": 6420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391266959678961e-06,
+      "loss": 1.0307,
+      "step": 6421
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391171412191859e-06,
+      "loss": 1.1943,
+      "step": 6422
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.391075864704758e-06,
+      "loss": 1.6641,
+      "step": 6423
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3909803172176574e-06,
+      "loss": 1.1813,
+      "step": 6424
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3908847697305565e-06,
+      "loss": 0.7967,
+      "step": 6425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3907892222434555e-06,
+      "loss": 1.1629,
+      "step": 6426
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3906936747563546e-06,
+      "loss": 0.7592,
+      "step": 6427
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390598127269254e-06,
+      "loss": 1.2778,
+      "step": 6428
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390502579782152e-06,
+      "loss": 1.2057,
+      "step": 6429
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390407032295051e-06,
+      "loss": 1.2604,
+      "step": 6430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.39031148480795e-06,
+      "loss": 1.0639,
+      "step": 6431
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390215937320849e-06,
+      "loss": 1.6121,
+      "step": 6432
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390120389833748e-06,
+      "loss": 1.4684,
+      "step": 6433
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.390024842346647e-06,
+      "loss": 1.1926,
+      "step": 6434
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389929294859546e-06,
+      "loss": 1.2299,
+      "step": 6435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389833747372444e-06,
+      "loss": 1.3932,
+      "step": 6436
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389738199885343e-06,
+      "loss": 1.4813,
+      "step": 6437
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389642652398242e-06,
+      "loss": 1.246,
+      "step": 6438
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389547104911141e-06,
+      "loss": 1.4987,
+      "step": 6439
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3894515574240395e-06,
+      "loss": 1.1863,
+      "step": 6440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389356009936939e-06,
+      "loss": 1.3598,
+      "step": 6441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3892604624498384e-06,
+      "loss": 1.5592,
+      "step": 6442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389164914962737e-06,
+      "loss": 0.8933,
+      "step": 6443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.389069367475636e-06,
+      "loss": 1.4369,
+      "step": 6444
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388973819988535e-06,
+      "loss": 1.2,
+      "step": 6445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388878272501434e-06,
+      "loss": 1.1155,
+      "step": 6446
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388782725014332e-06,
+      "loss": 1.2107,
+      "step": 6447
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388687177527232e-06,
+      "loss": 0.864,
+      "step": 6448
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388591630040131e-06,
+      "loss": 1.2932,
+      "step": 6449
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388496082553029e-06,
+      "loss": 1.2387,
+      "step": 6450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388400535065928e-06,
+      "loss": 1.3427,
+      "step": 6451
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388304987578827e-06,
+      "loss": 0.908,
+      "step": 6452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388209440091726e-06,
+      "loss": 1.5214,
+      "step": 6453
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388113892604624e-06,
+      "loss": 1.1719,
+      "step": 6454
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.388018345117524e-06,
+      "loss": 1.1023,
+      "step": 6455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3879227976304224e-06,
+      "loss": 1.3826,
+      "step": 6456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3878272501433215e-06,
+      "loss": 1.2177,
+      "step": 6457
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3877317026562205e-06,
+      "loss": 1.0574,
+      "step": 6458
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3876361551691196e-06,
+      "loss": 1.477,
+      "step": 6459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387540607682019e-06,
+      "loss": 1.3451,
+      "step": 6460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387445060194917e-06,
+      "loss": 1.0142,
+      "step": 6461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387349512707817e-06,
+      "loss": 0.639,
+      "step": 6462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387253965220715e-06,
+      "loss": 1.2676,
+      "step": 6463
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387158417733614e-06,
+      "loss": 1.2944,
+      "step": 6464
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387062870246513e-06,
+      "loss": 1.3714,
+      "step": 6465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.386967322759412e-06,
+      "loss": 1.4464,
+      "step": 6466
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.38687177527231e-06,
+      "loss": 0.905,
+      "step": 6467
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.386776227785209e-06,
+      "loss": 1.4458,
+      "step": 6468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386680680298109e-06,
+      "loss": 0.8918,
+      "step": 6469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386585132811007e-06,
+      "loss": 1.078,
+      "step": 6470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386489585323906e-06,
+      "loss": 1.1211,
+      "step": 6471
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386394037836805e-06,
+      "loss": 0.9174,
+      "step": 6472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386298490349704e-06,
+      "loss": 1.2942,
+      "step": 6473
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386202942862603e-06,
+      "loss": 1.2147,
+      "step": 6474
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386107395375502e-06,
+      "loss": 1.5185,
+      "step": 6475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.386011847888401e-06,
+      "loss": 1.0256,
+      "step": 6476
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3859163004013e-06,
+      "loss": 1.5738,
+      "step": 6477
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385820752914199e-06,
+      "loss": 1.4043,
+      "step": 6478
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385725205427098e-06,
+      "loss": 1.5171,
+      "step": 6479
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385629657939997e-06,
+      "loss": 1.0516,
+      "step": 6480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385534110452895e-06,
+      "loss": 0.8376,
+      "step": 6481
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385438562965794e-06,
+      "loss": 1.222,
+      "step": 6482
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385343015478693e-06,
+      "loss": 1.0773,
+      "step": 6483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385247467991592e-06,
+      "loss": 0.9507,
+      "step": 6484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.385151920504491e-06,
+      "loss": 1.4778,
+      "step": 6485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38505637301739e-06,
+      "loss": 1.6813,
+      "step": 6486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384960825530289e-06,
+      "loss": 1.0161,
+      "step": 6487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3848652780431874e-06,
+      "loss": 1.1854,
+      "step": 6488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3847697305560865e-06,
+      "loss": 1.3697,
+      "step": 6489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3846741830689855e-06,
+      "loss": 1.2492,
+      "step": 6490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3845786355818846e-06,
+      "loss": 1.7707,
+      "step": 6491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384483088094784e-06,
+      "loss": 0.9199,
+      "step": 6492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384387540607683e-06,
+      "loss": 1.3726,
+      "step": 6493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384291993120582e-06,
+      "loss": 1.3345,
+      "step": 6494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38419644563348e-06,
+      "loss": 1.2475,
+      "step": 6495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384100898146379e-06,
+      "loss": 0.9114,
+      "step": 6496
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.384005350659278e-06,
+      "loss": 0.9942,
+      "step": 6497
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383909803172177e-06,
+      "loss": 0.8484,
+      "step": 6498
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383814255685076e-06,
+      "loss": 1.6891,
+      "step": 6499
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383718708197975e-06,
+      "loss": 1.028,
+      "step": 6500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383623160710874e-06,
+      "loss": 1.3538,
+      "step": 6501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383527613223772e-06,
+      "loss": 1.3459,
+      "step": 6502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383432065736671e-06,
+      "loss": 1.4284,
+      "step": 6503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38333651824957e-06,
+      "loss": 1.2245,
+      "step": 6504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.383240970762469e-06,
+      "loss": 0.8983,
+      "step": 6505
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3831454232753684e-06,
+      "loss": 1.2121,
+      "step": 6506
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3830498757882675e-06,
+      "loss": 1.5305,
+      "step": 6507
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382954328301166e-06,
+      "loss": 1.0228,
+      "step": 6508
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382858780814065e-06,
+      "loss": 0.9341,
+      "step": 6509
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382763233326964e-06,
+      "loss": 1.0854,
+      "step": 6510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382667685839863e-06,
+      "loss": 1.109,
+      "step": 6511
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382572138352762e-06,
+      "loss": 1.8334,
+      "step": 6512
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382476590865661e-06,
+      "loss": 1.2535,
+      "step": 6513
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.38238104337856e-06,
+      "loss": 1.5368,
+      "step": 6514
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382285495891458e-06,
+      "loss": 1.3752,
+      "step": 6515
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382189948404357e-06,
+      "loss": 1.0747,
+      "step": 6516
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.382094400917256e-06,
+      "loss": 1.3161,
+      "step": 6517
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381998853430155e-06,
+      "loss": 1.3011,
+      "step": 6518
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381903305943053e-06,
+      "loss": 1.0338,
+      "step": 6519
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381807758455953e-06,
+      "loss": 1.7414,
+      "step": 6520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381712210968852e-06,
+      "loss": 1.3255,
+      "step": 6521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3816166634817505e-06,
+      "loss": 1.6222,
+      "step": 6522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3815211159946496e-06,
+      "loss": 1.3416,
+      "step": 6523
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.381425568507549e-06,
+      "loss": 0.8167,
+      "step": 6524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.381330021020448e-06,
+      "loss": 0.8815,
+      "step": 6525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.381234473533346e-06,
+      "loss": 1.1354,
+      "step": 6526
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.381138926046246e-06,
+      "loss": 1.4331,
+      "step": 6527
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.381043378559145e-06,
+      "loss": 1.5507,
+      "step": 6528
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380947831072043e-06,
+      "loss": 1.0096,
+      "step": 6529
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380852283584942e-06,
+      "loss": 1.1487,
+      "step": 6530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380756736097841e-06,
+      "loss": 1.2703,
+      "step": 6531
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38066118861074e-06,
+      "loss": 1.0909,
+      "step": 6532
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380565641123638e-06,
+      "loss": 1.4583,
+      "step": 6533
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380470093636538e-06,
+      "loss": 1.1691,
+      "step": 6534
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380374546149437e-06,
+      "loss": 1.3656,
+      "step": 6535
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380278998662335e-06,
+      "loss": 1.0535,
+      "step": 6536
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.380183451175234e-06,
+      "loss": 1.6324,
+      "step": 6537
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3800879036881334e-06,
+      "loss": 1.0612,
+      "step": 6538
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3799923562010325e-06,
+      "loss": 1.0681,
+      "step": 6539
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379896808713931e-06,
+      "loss": 1.5282,
+      "step": 6540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3798012612268306e-06,
+      "loss": 1.0544,
+      "step": 6541
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.37970571373973e-06,
+      "loss": 0.8122,
+      "step": 6542
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379610166252628e-06,
+      "loss": 1.631,
+      "step": 6543
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379514618765527e-06,
+      "loss": 1.237,
+      "step": 6544
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379419071278426e-06,
+      "loss": 1.4038,
+      "step": 6545
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379323523791325e-06,
+      "loss": 1.6943,
+      "step": 6546
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379227976304223e-06,
+      "loss": 1.4574,
+      "step": 6547
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379132428817122e-06,
+      "loss": 1.1284,
+      "step": 6548
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.379036881330021e-06,
+      "loss": 1.133,
+      "step": 6549
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.37894133384292e-06,
+      "loss": 1.4073,
+      "step": 6550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378845786355819e-06,
+      "loss": 1.2824,
+      "step": 6551
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378750238868718e-06,
+      "loss": 1.3164,
+      "step": 6552
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378654691381617e-06,
+      "loss": 0.9713,
+      "step": 6553
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3785591438945155e-06,
+      "loss": 1.1961,
+      "step": 6554
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3784635964074146e-06,
+      "loss": 1.5885,
+      "step": 6555
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378368048920314e-06,
+      "loss": 0.954,
+      "step": 6556
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378272501433213e-06,
+      "loss": 1.2095,
+      "step": 6557
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378176953946112e-06,
+      "loss": 0.9006,
+      "step": 6558
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.378081406459011e-06,
+      "loss": 1.3135,
+      "step": 6559
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377985858971909e-06,
+      "loss": 1.1801,
+      "step": 6560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377890311484808e-06,
+      "loss": 1.1623,
+      "step": 6561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377794763997707e-06,
+      "loss": 1.3389,
+      "step": 6562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377699216510606e-06,
+      "loss": 1.1766,
+      "step": 6563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377603669023505e-06,
+      "loss": 1.0697,
+      "step": 6564
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377508121536404e-06,
+      "loss": 1.4207,
+      "step": 6565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377412574049303e-06,
+      "loss": 1.0814,
+      "step": 6566
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377317026562201e-06,
+      "loss": 1.5141,
+      "step": 6567
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3772214790751e-06,
+      "loss": 1.0998,
+      "step": 6568
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377125931587999e-06,
+      "loss": 1.0952,
+      "step": 6569
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3770303841008984e-06,
+      "loss": 1.212,
+      "step": 6570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3769348366137975e-06,
+      "loss": 2.1753,
+      "step": 6571
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3768392891266965e-06,
+      "loss": 0.8939,
+      "step": 6572
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3767437416395956e-06,
+      "loss": 1.2492,
+      "step": 6573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376648194152494e-06,
+      "loss": 1.6339,
+      "step": 6574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376552646665393e-06,
+      "loss": 1.4215,
+      "step": 6575
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376457099178292e-06,
+      "loss": 1.1879,
+      "step": 6576
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376361551691191e-06,
+      "loss": 0.6975,
+      "step": 6577
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37626600420409e-06,
+      "loss": 1.2785,
+      "step": 6578
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376170456716989e-06,
+      "loss": 1.3542,
+      "step": 6579
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376074909229888e-06,
+      "loss": 0.9888,
+      "step": 6580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375979361742786e-06,
+      "loss": 1.3674,
+      "step": 6581
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375883814255685e-06,
+      "loss": 1.802,
+      "step": 6582
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375788266768584e-06,
+      "loss": 1.3189,
+      "step": 6583
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375692719281483e-06,
+      "loss": 1.3346,
+      "step": 6584
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375597171794382e-06,
+      "loss": 1.3336,
+      "step": 6585
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375501624307281e-06,
+      "loss": 1.3768,
+      "step": 6586
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37540607682018e-06,
+      "loss": 1.2083,
+      "step": 6587
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375310529333079e-06,
+      "loss": 1.5252,
+      "step": 6588
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375214981845978e-06,
+      "loss": 1.3163,
+      "step": 6589
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375119434358877e-06,
+      "loss": 1.7385,
+      "step": 6590
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.375023886871776e-06,
+      "loss": 0.9514,
+      "step": 6591
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374928339384675e-06,
+      "loss": 1.5482,
+      "step": 6592
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374832791897574e-06,
+      "loss": 1.3037,
+      "step": 6593
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374737244410473e-06,
+      "loss": 1.2702,
+      "step": 6594
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374641696923371e-06,
+      "loss": 1.0815,
+      "step": 6595
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37454614943627e-06,
+      "loss": 1.1054,
+      "step": 6596
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374450601949169e-06,
+      "loss": 1.2113,
+      "step": 6597
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374355054462068e-06,
+      "loss": 1.4464,
+      "step": 6598
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374259506974967e-06,
+      "loss": 1.2076,
+      "step": 6599
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374163959487866e-06,
+      "loss": 0.8771,
+      "step": 6600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.374068412000764e-06,
+      "loss": 1.3501,
+      "step": 6601
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3739728645136634e-06,
+      "loss": 0.8359,
+      "step": 6602
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3738773170265625e-06,
+      "loss": 1.0274,
+      "step": 6603
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3737817695394615e-06,
+      "loss": 1.1204,
+      "step": 6604
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3736862220523606e-06,
+      "loss": 1.1355,
+      "step": 6605
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37359067456526e-06,
+      "loss": 1.1246,
+      "step": 6606
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373495127078159e-06,
+      "loss": 1.0782,
+      "step": 6607
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373399579591057e-06,
+      "loss": 1.3738,
+      "step": 6608
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373304032103956e-06,
+      "loss": 1.4176,
+      "step": 6609
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373208484616855e-06,
+      "loss": 1.5212,
+      "step": 6610
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373112937129754e-06,
+      "loss": 0.9982,
+      "step": 6611
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.373017389642652e-06,
+      "loss": 0.8889,
+      "step": 6612
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372921842155552e-06,
+      "loss": 0.8952,
+      "step": 6613
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372826294668451e-06,
+      "loss": 1.3594,
+      "step": 6614
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372730747181349e-06,
+      "loss": 1.5785,
+      "step": 6615
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372635199694248e-06,
+      "loss": 1.061,
+      "step": 6616
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372539652207147e-06,
+      "loss": 0.7943,
+      "step": 6617
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372444104720046e-06,
+      "loss": 0.8563,
+      "step": 6618
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3723485572329445e-06,
+      "loss": 1.2081,
+      "step": 6619
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372253009745844e-06,
+      "loss": 1.0948,
+      "step": 6620
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3721574622587435e-06,
+      "loss": 1.4578,
+      "step": 6621
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372061914771642e-06,
+      "loss": 0.583,
+      "step": 6622
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.371966367284541e-06,
+      "loss": 1.1292,
+      "step": 6623
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37187081979744e-06,
+      "loss": 1.5995,
+      "step": 6624
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.371775272310339e-06,
+      "loss": 0.9567,
+      "step": 6625
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.371679724823237e-06,
+      "loss": 1.2212,
+      "step": 6626
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371584177336136e-06,
+      "loss": 1.5685,
+      "step": 6627
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371488629849036e-06,
+      "loss": 1.0339,
+      "step": 6628
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371393082361934e-06,
+      "loss": 0.9331,
+      "step": 6629
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371297534874833e-06,
+      "loss": 1.1186,
+      "step": 6630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371201987387732e-06,
+      "loss": 0.7721,
+      "step": 6631
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371106439900631e-06,
+      "loss": 1.0811,
+      "step": 6632
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371010892413529e-06,
+      "loss": 1.2082,
+      "step": 6633
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3709153449264284e-06,
+      "loss": 1.5856,
+      "step": 6634
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370819797439328e-06,
+      "loss": 1.3506,
+      "step": 6635
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3707242499522265e-06,
+      "loss": 1.1527,
+      "step": 6636
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3706287024651255e-06,
+      "loss": 1.2328,
+      "step": 6637
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370533154978025e-06,
+      "loss": 0.8583,
+      "step": 6638
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370437607490924e-06,
+      "loss": 1.1481,
+      "step": 6639
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370342060003822e-06,
+      "loss": 1.2239,
+      "step": 6640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370246512516721e-06,
+      "loss": 1.4489,
+      "step": 6641
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370150965029621e-06,
+      "loss": 1.1049,
+      "step": 6642
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370055417542519e-06,
+      "loss": 0.9897,
+      "step": 6643
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369959870055418e-06,
+      "loss": 1.1676,
+      "step": 6644
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369864322568317e-06,
+      "loss": 1.7289,
+      "step": 6645
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369768775081216e-06,
+      "loss": 1.4379,
+      "step": 6646
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369673227594114e-06,
+      "loss": 1.1804,
+      "step": 6647
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369577680107013e-06,
+      "loss": 1.5455,
+      "step": 6648
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369482132619912e-06,
+      "loss": 0.9179,
+      "step": 6649
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369386585132811e-06,
+      "loss": 1.1924,
+      "step": 6650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.36929103764571e-06,
+      "loss": 0.9814,
+      "step": 6651
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3691954901586094e-06,
+      "loss": 1.2495,
+      "step": 6652
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3690999426715085e-06,
+      "loss": 1.1241,
+      "step": 6653
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369004395184407e-06,
+      "loss": 0.8235,
+      "step": 6654
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368908847697306e-06,
+      "loss": 1.8281,
+      "step": 6655
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368813300210205e-06,
+      "loss": 1.073,
+      "step": 6656
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368717752723104e-06,
+      "loss": 1.572,
+      "step": 6657
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368622205236003e-06,
+      "loss": 1.0673,
+      "step": 6658
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368526657748902e-06,
+      "loss": 0.8743,
+      "step": 6659
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3684311102618e-06,
+      "loss": 0.9504,
+      "step": 6660
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368335562774699e-06,
+      "loss": 1.5313,
+      "step": 6661
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368240015287598e-06,
+      "loss": 0.9191,
+      "step": 6662
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368144467800497e-06,
+      "loss": 1.3666,
+      "step": 6663
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368048920313396e-06,
+      "loss": 1.3748,
+      "step": 6664
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367953372826295e-06,
+      "loss": 1.3027,
+      "step": 6665
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367857825339194e-06,
+      "loss": 1.5201,
+      "step": 6666
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3677622778520925e-06,
+      "loss": 0.6826,
+      "step": 6667
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3676667303649915e-06,
+      "loss": 0.9732,
+      "step": 6668
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3675711828778905e-06,
+      "loss": 0.9827,
+      "step": 6669
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.36747563539079e-06,
+      "loss": 1.3484,
+      "step": 6670
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367380087903689e-06,
+      "loss": 0.9347,
+      "step": 6671
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367284540416588e-06,
+      "loss": 0.9965,
+      "step": 6672
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367188992929487e-06,
+      "loss": 1.1487,
+      "step": 6673
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367093445442385e-06,
+      "loss": 1.3633,
+      "step": 6674
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366997897955284e-06,
+      "loss": 1.0784,
+      "step": 6675
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366902350468183e-06,
+      "loss": 1.58,
+      "step": 6676
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366806802981082e-06,
+      "loss": 1.1531,
+      "step": 6677
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366711255493981e-06,
+      "loss": 1.6382,
+      "step": 6678
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36661570800688e-06,
+      "loss": 1.3228,
+      "step": 6679
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.366520160519779e-06,
+      "loss": 1.0155,
+      "step": 6680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.366424613032677e-06,
+      "loss": 1.2929,
+      "step": 6681
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.366329065545576e-06,
+      "loss": 1.4708,
+      "step": 6682
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.366233518058475e-06,
+      "loss": 1.2576,
+      "step": 6683
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3661379705713744e-06,
+      "loss": 0.8188,
+      "step": 6684
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3660424230842735e-06,
+      "loss": 1.3971,
+      "step": 6685
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3659468755971725e-06,
+      "loss": 1.1668,
+      "step": 6686
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3658513281100715e-06,
+      "loss": 1.1766,
+      "step": 6687
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36575578062297e-06,
+      "loss": 1.5765,
+      "step": 6688
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365660233135869e-06,
+      "loss": 1.1727,
+      "step": 6689
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365564685648768e-06,
+      "loss": 1.4835,
+      "step": 6690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365469138161667e-06,
+      "loss": 1.4451,
+      "step": 6691
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365373590674565e-06,
+      "loss": 1.0866,
+      "step": 6692
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365278043187465e-06,
+      "loss": 1.5315,
+      "step": 6693
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365182495700364e-06,
+      "loss": 1.7624,
+      "step": 6694
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365086948213262e-06,
+      "loss": 1.4279,
+      "step": 6695
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364991400726161e-06,
+      "loss": 0.9838,
+      "step": 6696
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36489585323906e-06,
+      "loss": 1.158,
+      "step": 6697
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364800305751959e-06,
+      "loss": 0.966,
+      "step": 6698
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3647047582648575e-06,
+      "loss": 1.0703,
+      "step": 6699
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364609210777757e-06,
+      "loss": 1.3405,
+      "step": 6700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3645136632906555e-06,
+      "loss": 1.5059,
+      "step": 6701
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364418115803555e-06,
+      "loss": 1.2541,
+      "step": 6702
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364322568316454e-06,
+      "loss": 0.899,
+      "step": 6703
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364227020829353e-06,
+      "loss": 1.7633,
+      "step": 6704
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364131473342252e-06,
+      "loss": 1.5453,
+      "step": 6705
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36403592585515e-06,
+      "loss": 0.9941,
+      "step": 6706
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36394037836805e-06,
+      "loss": 1.1221,
+      "step": 6707
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363844830880948e-06,
+      "loss": 1.4151,
+      "step": 6708
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363749283393847e-06,
+      "loss": 0.8568,
+      "step": 6709
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363653735906746e-06,
+      "loss": 1.2618,
+      "step": 6710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363558188419645e-06,
+      "loss": 1.1532,
+      "step": 6711
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363462640932543e-06,
+      "loss": 1.7571,
+      "step": 6712
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363367093445442e-06,
+      "loss": 0.8901,
+      "step": 6713
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363271545958342e-06,
+      "loss": 1.8222,
+      "step": 6714
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36317599847124e-06,
+      "loss": 0.8109,
+      "step": 6715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3630804509841394e-06,
+      "loss": 1.1073,
+      "step": 6716
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3629849034970385e-06,
+      "loss": 1.1963,
+      "step": 6717
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3628893560099375e-06,
+      "loss": 0.8558,
+      "step": 6718
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362793808522836e-06,
+      "loss": 1.1851,
+      "step": 6719
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362698261035735e-06,
+      "loss": 2.0135,
+      "step": 6720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362602713548635e-06,
+      "loss": 1.4745,
+      "step": 6721
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362507166061533e-06,
+      "loss": 1.2401,
+      "step": 6722
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362411618574432e-06,
+      "loss": 1.1521,
+      "step": 6723
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362316071087331e-06,
+      "loss": 0.9229,
+      "step": 6724
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36222052360023e-06,
+      "loss": 1.8264,
+      "step": 6725
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362124976113128e-06,
+      "loss": 0.6826,
+      "step": 6726
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362029428626027e-06,
+      "loss": 0.9604,
+      "step": 6727
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361933881138926e-06,
+      "loss": 0.8585,
+      "step": 6728
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361838333651825e-06,
+      "loss": 1.4578,
+      "step": 6729
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361742786164724e-06,
+      "loss": 1.0212,
+      "step": 6730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.361647238677623e-06,
+      "loss": 1.3392,
+      "step": 6731
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.361551691190522e-06,
+      "loss": 1.7304,
+      "step": 6732
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3614561437034205e-06,
+      "loss": 1.1821,
+      "step": 6733
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36136059621632e-06,
+      "loss": 0.7203,
+      "step": 6734
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.361265048729219e-06,
+      "loss": 1.1535,
+      "step": 6735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.361169501242118e-06,
+      "loss": 0.9709,
+      "step": 6736
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.361073953755017e-06,
+      "loss": 0.5851,
+      "step": 6737
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360978406267916e-06,
+      "loss": 0.757,
+      "step": 6738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360882858780815e-06,
+      "loss": 0.9741,
+      "step": 6739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360787311293713e-06,
+      "loss": 1.4463,
+      "step": 6740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360691763806612e-06,
+      "loss": 1.4504,
+      "step": 6741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360596216319511e-06,
+      "loss": 0.4687,
+      "step": 6742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.36050066883241e-06,
+      "loss": 1.0666,
+      "step": 6743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360405121345309e-06,
+      "loss": 0.8639,
+      "step": 6744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360309573858208e-06,
+      "loss": 1.0467,
+      "step": 6745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360214026371107e-06,
+      "loss": 0.7046,
+      "step": 6746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.360118478884005e-06,
+      "loss": 1.1429,
+      "step": 6747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3600229313969044e-06,
+      "loss": 1.0983,
+      "step": 6748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3599273839098035e-06,
+      "loss": 0.8538,
+      "step": 6749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3598318364227025e-06,
+      "loss": 1.5296,
+      "step": 6750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3597362889356015e-06,
+      "loss": 1.153,
+      "step": 6751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359640741448501e-06,
+      "loss": 0.9284,
+      "step": 6752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359545193961399e-06,
+      "loss": 1.004,
+      "step": 6753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359449646474298e-06,
+      "loss": 1.2418,
+      "step": 6754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359354098987197e-06,
+      "loss": 1.8573,
+      "step": 6755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359258551500096e-06,
+      "loss": 0.7688,
+      "step": 6756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359163004012995e-06,
+      "loss": 1.0593,
+      "step": 6757
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.359067456525894e-06,
+      "loss": 1.615,
+      "step": 6758
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358971909038793e-06,
+      "loss": 1.0616,
+      "step": 6759
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358876361551691e-06,
+      "loss": 1.0426,
+      "step": 6760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35878081406459e-06,
+      "loss": 1.2637,
+      "step": 6761
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358685266577489e-06,
+      "loss": 0.6775,
+      "step": 6762
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358589719090388e-06,
+      "loss": 1.0983,
+      "step": 6763
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3584941716032865e-06,
+      "loss": 0.8423,
+      "step": 6764
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358398624116186e-06,
+      "loss": 1.2124,
+      "step": 6765
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3583030766290854e-06,
+      "loss": 1.0812,
+      "step": 6766
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358207529141984e-06,
+      "loss": 1.5131,
+      "step": 6767
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358111981654883e-06,
+      "loss": 0.9388,
+      "step": 6768
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.358016434167782e-06,
+      "loss": 1.1541,
+      "step": 6769
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357920886680681e-06,
+      "loss": 1.7486,
+      "step": 6770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357825339193579e-06,
+      "loss": 1.3372,
+      "step": 6771
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357729791706479e-06,
+      "loss": 1.5201,
+      "step": 6772
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357634244219378e-06,
+      "loss": 1.6318,
+      "step": 6773
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357538696732276e-06,
+      "loss": 1.8091,
+      "step": 6774
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357443149245175e-06,
+      "loss": 0.8559,
+      "step": 6775
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357347601758074e-06,
+      "loss": 1.3614,
+      "step": 6776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357252054270973e-06,
+      "loss": 1.6328,
+      "step": 6777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357156506783871e-06,
+      "loss": 1.0836,
+      "step": 6778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357060959296771e-06,
+      "loss": 1.7543,
+      "step": 6779
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35696541180967e-06,
+      "loss": 1.4396,
+      "step": 6780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3568698643225685e-06,
+      "loss": 1.0022,
+      "step": 6781
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3567743168354675e-06,
+      "loss": 1.2295,
+      "step": 6782
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3566787693483665e-06,
+      "loss": 1.3966,
+      "step": 6783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356583221861266e-06,
+      "loss": 1.0339,
+      "step": 6784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356487674374164e-06,
+      "loss": 0.9579,
+      "step": 6785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356392126887064e-06,
+      "loss": 0.9863,
+      "step": 6786
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356296579399963e-06,
+      "loss": 1.3495,
+      "step": 6787
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356201031912861e-06,
+      "loss": 1.1652,
+      "step": 6788
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35610548442576e-06,
+      "loss": 0.8977,
+      "step": 6789
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.356009936938659e-06,
+      "loss": 1.261,
+      "step": 6790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355914389451558e-06,
+      "loss": 1.1551,
+      "step": 6791
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355818841964456e-06,
+      "loss": 1.5517,
+      "step": 6792
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355723294477356e-06,
+      "loss": 1.2538,
+      "step": 6793
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355627746990255e-06,
+      "loss": 1.3287,
+      "step": 6794
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355532199503153e-06,
+      "loss": 1.0885,
+      "step": 6795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355436652016052e-06,
+      "loss": 1.1124,
+      "step": 6796
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355341104528951e-06,
+      "loss": 0.8929,
+      "step": 6797
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3552455570418504e-06,
+      "loss": 1.526,
+      "step": 6798
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355150009554749e-06,
+      "loss": 1.0767,
+      "step": 6799
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.355054462067648e-06,
+      "loss": 0.9309,
+      "step": 6800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354958914580547e-06,
+      "loss": 1.3443,
+      "step": 6801
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354863367093446e-06,
+      "loss": 1.1405,
+      "step": 6802
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354767819606345e-06,
+      "loss": 1.426,
+      "step": 6803
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354672272119244e-06,
+      "loss": 1.0028,
+      "step": 6804
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354576724632142e-06,
+      "loss": 1.307,
+      "step": 6805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354481177145041e-06,
+      "loss": 0.6538,
+      "step": 6806
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35438562965794e-06,
+      "loss": 1.2119,
+      "step": 6807
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354290082170839e-06,
+      "loss": 1.0124,
+      "step": 6808
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354194534683738e-06,
+      "loss": 1.0937,
+      "step": 6809
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354098987196637e-06,
+      "loss": 0.9462,
+      "step": 6810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.354003439709536e-06,
+      "loss": 1.161,
+      "step": 6811
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3539078922224344e-06,
+      "loss": 1.3169,
+      "step": 6812
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3538123447353335e-06,
+      "loss": 1.242,
+      "step": 6813
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3537167972482325e-06,
+      "loss": 0.838,
+      "step": 6814
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3536212497611315e-06,
+      "loss": 0.9112,
+      "step": 6815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.353525702274031e-06,
+      "loss": 1.1758,
+      "step": 6816
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35343015478693e-06,
+      "loss": 1.3697,
+      "step": 6817
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.353334607299829e-06,
+      "loss": 1.5128,
+      "step": 6818
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.353239059812727e-06,
+      "loss": 1.0295,
+      "step": 6819
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.353143512325626e-06,
+      "loss": 1.1547,
+      "step": 6820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.353047964838525e-06,
+      "loss": 0.6564,
+      "step": 6821
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352952417351424e-06,
+      "loss": 1.4655,
+      "step": 6822
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352856869864323e-06,
+      "loss": 1.2135,
+      "step": 6823
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352761322377222e-06,
+      "loss": 1.4944,
+      "step": 6824
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352665774890121e-06,
+      "loss": 1.8781,
+      "step": 6825
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352570227403019e-06,
+      "loss": 0.9179,
+      "step": 6826
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352474679915918e-06,
+      "loss": 1.4167,
+      "step": 6827
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352379132428817e-06,
+      "loss": 0.9777,
+      "step": 6828
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352283584941716e-06,
+      "loss": 1.121,
+      "step": 6829
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3521880374546154e-06,
+      "loss": 1.126,
+      "step": 6830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3520924899675145e-06,
+      "loss": 0.8804,
+      "step": 6831
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3519969424804135e-06,
+      "loss": 1.0986,
+      "step": 6832
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351901394993312e-06,
+      "loss": 1.0677,
+      "step": 6833
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351805847506211e-06,
+      "loss": 1.471,
+      "step": 6834
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35171030001911e-06,
+      "loss": 1.0238,
+      "step": 6835
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351614752532009e-06,
+      "loss": 1.3853,
+      "step": 6836
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351519205044908e-06,
+      "loss": 1.5881,
+      "step": 6837
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351423657557807e-06,
+      "loss": 1.199,
+      "step": 6838
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351328110070706e-06,
+      "loss": 0.9426,
+      "step": 6839
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351232562583604e-06,
+      "loss": 1.0691,
+      "step": 6840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351137015096503e-06,
+      "loss": 0.8018,
+      "step": 6841
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.351041467609402e-06,
+      "loss": 1.834,
+      "step": 6842
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350945920122301e-06,
+      "loss": 1.1195,
+      "step": 6843
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3508503726352e-06,
+      "loss": 1.5927,
+      "step": 6844
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350754825148099e-06,
+      "loss": 0.8951,
+      "step": 6845
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350659277660998e-06,
+      "loss": 0.7468,
+      "step": 6846
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3505637301738965e-06,
+      "loss": 0.9178,
+      "step": 6847
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350468182686796e-06,
+      "loss": 0.4467,
+      "step": 6848
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350372635199695e-06,
+      "loss": 1.2999,
+      "step": 6849
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350277087712594e-06,
+      "loss": 1.3537,
+      "step": 6850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350181540225493e-06,
+      "loss": 1.4431,
+      "step": 6851
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.350085992738392e-06,
+      "loss": 1.0931,
+      "step": 6852
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34999044525129e-06,
+      "loss": 1.3665,
+      "step": 6853
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349894897764189e-06,
+      "loss": 1.099,
+      "step": 6854
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349799350277088e-06,
+      "loss": 1.0688,
+      "step": 6855
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349703802789987e-06,
+      "loss": 0.8235,
+      "step": 6856
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349608255302886e-06,
+      "loss": 0.9151,
+      "step": 6857
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349512707815785e-06,
+      "loss": 1.2725,
+      "step": 6858
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349417160328684e-06,
+      "loss": 1.4163,
+      "step": 6859
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349321612841582e-06,
+      "loss": 1.1789,
+      "step": 6860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.349226065354481e-06,
+      "loss": 1.1892,
+      "step": 6861
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3491305178673804e-06,
+      "loss": 1.0044,
+      "step": 6862
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3490349703802795e-06,
+      "loss": 1.3785,
+      "step": 6863
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348939422893178e-06,
+      "loss": 1.4365,
+      "step": 6864
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3488438754060775e-06,
+      "loss": 1.1897,
+      "step": 6865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348748327918977e-06,
+      "loss": 1.4434,
+      "step": 6866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348652780431875e-06,
+      "loss": 1.4412,
+      "step": 6867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348557232944774e-06,
+      "loss": 0.9605,
+      "step": 6868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348461685457673e-06,
+      "loss": 1.1543,
+      "step": 6869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348366137970572e-06,
+      "loss": 1.5886,
+      "step": 6870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34827059048347e-06,
+      "loss": 1.145,
+      "step": 6871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348175042996369e-06,
+      "loss": 1.0916,
+      "step": 6872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.348079495509269e-06,
+      "loss": 0.8768,
+      "step": 6873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347983948022167e-06,
+      "loss": 0.8522,
+      "step": 6874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347888400535066e-06,
+      "loss": 1.6598,
+      "step": 6875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347792853047965e-06,
+      "loss": 1.2421,
+      "step": 6876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347697305560864e-06,
+      "loss": 1.2192,
+      "step": 6877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3476017580737625e-06,
+      "loss": 1.4457,
+      "step": 6878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3475062105866615e-06,
+      "loss": 1.2398,
+      "step": 6879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3474106630995614e-06,
+      "loss": 1.6078,
+      "step": 6880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34731511561246e-06,
+      "loss": 1.2967,
+      "step": 6881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347219568125359e-06,
+      "loss": 1.1548,
+      "step": 6882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347124020638258e-06,
+      "loss": 1.1977,
+      "step": 6883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.347028473151157e-06,
+      "loss": 1.4485,
+      "step": 6884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346932925664055e-06,
+      "loss": 0.9277,
+      "step": 6885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346837378176954e-06,
+      "loss": 1.4946,
+      "step": 6886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346741830689854e-06,
+      "loss": 0.9378,
+      "step": 6887
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.346646283202752e-06,
+      "loss": 1.0835,
+      "step": 6888
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.346550735715651e-06,
+      "loss": 1.0261,
+      "step": 6889
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.34645518822855e-06,
+      "loss": 1.1717,
+      "step": 6890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.346359640741449e-06,
+      "loss": 0.9811,
+      "step": 6891
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.346264093254347e-06,
+      "loss": 0.9323,
+      "step": 6892
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.346168545767246e-06,
+      "loss": 1.3217,
+      "step": 6893
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3460729982801454e-06,
+      "loss": 1.0526,
+      "step": 6894
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3459774507930445e-06,
+      "loss": 1.3817,
+      "step": 6895
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3458819033059435e-06,
+      "loss": 1.0843,
+      "step": 6896
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3457863558188425e-06,
+      "loss": 1.4166,
+      "step": 6897
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345690808331742e-06,
+      "loss": 1.1396,
+      "step": 6898
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.34559526084464e-06,
+      "loss": 1.187,
+      "step": 6899
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345499713357539e-06,
+      "loss": 1.3014,
+      "step": 6900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345404165870438e-06,
+      "loss": 1.1978,
+      "step": 6901
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345308618383337e-06,
+      "loss": 1.7085,
+      "step": 6902
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345213070896236e-06,
+      "loss": 1.6179,
+      "step": 6903
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345117523409135e-06,
+      "loss": 0.9194,
+      "step": 6904
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.345021975922033e-06,
+      "loss": 1.3922,
+      "step": 6905
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344926428434932e-06,
+      "loss": 0.8922,
+      "step": 6906
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344830880947831e-06,
+      "loss": 1.3155,
+      "step": 6907
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.34473533346073e-06,
+      "loss": 1.7407,
+      "step": 6908
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344639785973629e-06,
+      "loss": 1.1,
+      "step": 6909
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344544238486528e-06,
+      "loss": 0.974,
+      "step": 6910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344448690999427e-06,
+      "loss": 1.1202,
+      "step": 6911
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344353143512326e-06,
+      "loss": 1.4522,
+      "step": 6912
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344257596025225e-06,
+      "loss": 1.3366,
+      "step": 6913
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344162048538124e-06,
+      "loss": 1.3984,
+      "step": 6914
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.344066501051023e-06,
+      "loss": 1.0485,
+      "step": 6915
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343970953563922e-06,
+      "loss": 1.3157,
+      "step": 6916
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343875406076821e-06,
+      "loss": 1.1344,
+      "step": 6917
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.34377985858972e-06,
+      "loss": 1.0213,
+      "step": 6918
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343684311102618e-06,
+      "loss": 1.05,
+      "step": 6919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343588763615517e-06,
+      "loss": 1.6537,
+      "step": 6920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343493216128416e-06,
+      "loss": 0.8114,
+      "step": 6921
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343397668641315e-06,
+      "loss": 1.4801,
+      "step": 6922
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343302121154214e-06,
+      "loss": 1.4004,
+      "step": 6923
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343206573667113e-06,
+      "loss": 1.4272,
+      "step": 6924
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.343111026180012e-06,
+      "loss": 0.9325,
+      "step": 6925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3430154786929104e-06,
+      "loss": 1.4257,
+      "step": 6926
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3429199312058095e-06,
+      "loss": 1.0893,
+      "step": 6927
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3428243837187085e-06,
+      "loss": 1.0199,
+      "step": 6928
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3427288362316075e-06,
+      "loss": 1.099,
+      "step": 6929
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342633288744507e-06,
+      "loss": 0.5736,
+      "step": 6930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342537741257406e-06,
+      "loss": 0.7525,
+      "step": 6931
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342442193770305e-06,
+      "loss": 1.2172,
+      "step": 6932
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342346646283203e-06,
+      "loss": 1.1117,
+      "step": 6933
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342251098796102e-06,
+      "loss": 1.1104,
+      "step": 6934
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.342155551309001e-06,
+      "loss": 0.9757,
+      "step": 6935
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3420600038219e-06,
+      "loss": 1.4177,
+      "step": 6936
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341964456334799e-06,
+      "loss": 0.9934,
+      "step": 6937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341868908847698e-06,
+      "loss": 1.337,
+      "step": 6938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341773361360597e-06,
+      "loss": 1.007,
+      "step": 6939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341677813873495e-06,
+      "loss": 1.1088,
+      "step": 6940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341582266386394e-06,
+      "loss": 1.1237,
+      "step": 6941
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341486718899293e-06,
+      "loss": 0.8762,
+      "step": 6942
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341391171412192e-06,
+      "loss": 0.8627,
+      "step": 6943
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341295623925091e-06,
+      "loss": 1.4658,
+      "step": 6944
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3412000764379905e-06,
+      "loss": 1.0393,
+      "step": 6945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341104528950889e-06,
+      "loss": 1.1065,
+      "step": 6946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.341008981463788e-06,
+      "loss": 1.2223,
+      "step": 6947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340913433976687e-06,
+      "loss": 0.9393,
+      "step": 6948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340817886489586e-06,
+      "loss": 1.1398,
+      "step": 6949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340722339002485e-06,
+      "loss": 0.7427,
+      "step": 6950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340626791515383e-06,
+      "loss": 1.2425,
+      "step": 6951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340531244028283e-06,
+      "loss": 0.8777,
+      "step": 6952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340435696541181e-06,
+      "loss": 1.025,
+      "step": 6953
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.34034014905408e-06,
+      "loss": 0.7237,
+      "step": 6954
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340244601566979e-06,
+      "loss": 1.2289,
+      "step": 6955
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340149054079878e-06,
+      "loss": 1.3645,
+      "step": 6956
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.340053506592776e-06,
+      "loss": 1.1172,
+      "step": 6957
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3399579591056754e-06,
+      "loss": 0.702,
+      "step": 6958
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339862411618575e-06,
+      "loss": 1.051,
+      "step": 6959
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3397668641314735e-06,
+      "loss": 1.0306,
+      "step": 6960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3396713166443725e-06,
+      "loss": 1.0949,
+      "step": 6961
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339575769157272e-06,
+      "loss": 1.0601,
+      "step": 6962
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339480221670171e-06,
+      "loss": 0.9849,
+      "step": 6963
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339384674183069e-06,
+      "loss": 1.1737,
+      "step": 6964
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339289126695968e-06,
+      "loss": 1.4815,
+      "step": 6965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339193579208868e-06,
+      "loss": 1.3534,
+      "step": 6966
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339098031721766e-06,
+      "loss": 1.3451,
+      "step": 6967
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.339002484234665e-06,
+      "loss": 1.2582,
+      "step": 6968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338906936747564e-06,
+      "loss": 1.4397,
+      "step": 6969
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338811389260463e-06,
+      "loss": 1.0849,
+      "step": 6970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338715841773361e-06,
+      "loss": 1.073,
+      "step": 6971
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.33862029428626e-06,
+      "loss": 1.1879,
+      "step": 6972
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.33852474679916e-06,
+      "loss": 1.4621,
+      "step": 6973
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338429199312058e-06,
+      "loss": 1.4628,
+      "step": 6974
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338333651824957e-06,
+      "loss": 1.8503,
+      "step": 6975
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3382381043378564e-06,
+      "loss": 1.6279,
+      "step": 6976
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3381425568507555e-06,
+      "loss": 1.2035,
+      "step": 6977
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.338047009363654e-06,
+      "loss": 1.4041,
+      "step": 6978
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337951461876553e-06,
+      "loss": 1.3402,
+      "step": 6979
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337855914389452e-06,
+      "loss": 0.8922,
+      "step": 6980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337760366902351e-06,
+      "loss": 1.3753,
+      "step": 6981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.33766481941525e-06,
+      "loss": 1.0752,
+      "step": 6982
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337569271928149e-06,
+      "loss": 1.6505,
+      "step": 6983
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337473724441048e-06,
+      "loss": 0.9574,
+      "step": 6984
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337378176953946e-06,
+      "loss": 1.6022,
+      "step": 6985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337282629466845e-06,
+      "loss": 0.9675,
+      "step": 6986
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337187081979744e-06,
+      "loss": 0.9423,
+      "step": 6987
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337091534492643e-06,
+      "loss": 1.412,
+      "step": 6988
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.336995987005542e-06,
+      "loss": 1.5275,
+      "step": 6989
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.336900439518441e-06,
+      "loss": 1.1563,
+      "step": 6990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.33680489203134e-06,
+      "loss": 1.0788,
+      "step": 6991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3367093445442385e-06,
+      "loss": 1.0297,
+      "step": 6992
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3366137970571375e-06,
+      "loss": 1.1189,
+      "step": 6993
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336518249570037e-06,
+      "loss": 1.2675,
+      "step": 6994
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336422702082936e-06,
+      "loss": 1.2141,
+      "step": 6995
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336327154595835e-06,
+      "loss": 1.3223,
+      "step": 6996
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336231607108734e-06,
+      "loss": 1.1235,
+      "step": 6997
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336136059621632e-06,
+      "loss": 1.1638,
+      "step": 6998
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.336040512134531e-06,
+      "loss": 1.0357,
+      "step": 6999
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33594496464743e-06,
+      "loss": 1.1833,
+      "step": 7000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335849417160329e-06,
+      "loss": 1.1648,
+      "step": 7001
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335753869673228e-06,
+      "loss": 0.9885,
+      "step": 7002
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335658322186127e-06,
+      "loss": 1.543,
+      "step": 7003
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335562774699026e-06,
+      "loss": 1.2722,
+      "step": 7004
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335467227211924e-06,
+      "loss": 0.8919,
+      "step": 7005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335371679724823e-06,
+      "loss": 1.585,
+      "step": 7006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.335276132237722e-06,
+      "loss": 1.1719,
+      "step": 7007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3351805847506214e-06,
+      "loss": 1.2484,
+      "step": 7008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3350850372635205e-06,
+      "loss": 1.0642,
+      "step": 7009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3349894897764195e-06,
+      "loss": 0.7673,
+      "step": 7010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3348939422893185e-06,
+      "loss": 1.0386,
+      "step": 7011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334798394802217e-06,
+      "loss": 1.5605,
+      "step": 7012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334702847315116e-06,
+      "loss": 0.9194,
+      "step": 7013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334607299828015e-06,
+      "loss": 1.0836,
+      "step": 7014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334511752340914e-06,
+      "loss": 1.3494,
+      "step": 7015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334416204853812e-06,
+      "loss": 1.2774,
+      "step": 7016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334320657366712e-06,
+      "loss": 1.0683,
+      "step": 7017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334225109879611e-06,
+      "loss": 1.4393,
+      "step": 7018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334129562392509e-06,
+      "loss": 1.0057,
+      "step": 7019
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.334034014905408e-06,
+      "loss": 1.5988,
+      "step": 7020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333938467418307e-06,
+      "loss": 1.1371,
+      "step": 7021
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333842919931206e-06,
+      "loss": 1.1414,
+      "step": 7022
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3337473724441045e-06,
+      "loss": 1.6941,
+      "step": 7023
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333651824957004e-06,
+      "loss": 1.2509,
+      "step": 7024
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333556277469903e-06,
+      "loss": 1.0991,
+      "step": 7025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333460729982802e-06,
+      "loss": 1.1788,
+      "step": 7026
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333365182495701e-06,
+      "loss": 1.1641,
+      "step": 7027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3332696350086e-06,
+      "loss": 1.0739,
+      "step": 7028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333174087521499e-06,
+      "loss": 1.429,
+      "step": 7029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.333078540034397e-06,
+      "loss": 1.4306,
+      "step": 7030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332982992547297e-06,
+      "loss": 1.303,
+      "step": 7031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332887445060196e-06,
+      "loss": 1.4366,
+      "step": 7032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332791897573094e-06,
+      "loss": 0.845,
+      "step": 7033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332696350085993e-06,
+      "loss": 0.9221,
+      "step": 7034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332600802598892e-06,
+      "loss": 1.6511,
+      "step": 7035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332505255111791e-06,
+      "loss": 0.8908,
+      "step": 7036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332409707624689e-06,
+      "loss": 1.0969,
+      "step": 7037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332314160137589e-06,
+      "loss": 0.8251,
+      "step": 7038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332218612650488e-06,
+      "loss": 1.0433,
+      "step": 7039
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3321230651633864e-06,
+      "loss": 1.3619,
+      "step": 7040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3320275176762855e-06,
+      "loss": 1.4955,
+      "step": 7041
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3319319701891845e-06,
+      "loss": 1.3555,
+      "step": 7042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3318364227020835e-06,
+      "loss": 1.4175,
+      "step": 7043
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.331740875214982e-06,
+      "loss": 1.1926,
+      "step": 7044
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.331645327727882e-06,
+      "loss": 1.7343,
+      "step": 7045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.33154978024078e-06,
+      "loss": 1.4133,
+      "step": 7046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.331454232753679e-06,
+      "loss": 1.2448,
+      "step": 7047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.331358685266578e-06,
+      "loss": 0.8088,
+      "step": 7048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.331263137779477e-06,
+      "loss": 1.1042,
+      "step": 7049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.331167590292376e-06,
+      "loss": 1.0981,
+      "step": 7050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.331072042805274e-06,
+      "loss": 1.4562,
+      "step": 7051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330976495318173e-06,
+      "loss": 1.5044,
+      "step": 7052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330880947831072e-06,
+      "loss": 1.2144,
+      "step": 7053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330785400343971e-06,
+      "loss": 0.7976,
+      "step": 7054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.33068985285687e-06,
+      "loss": 1.0804,
+      "step": 7055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330594305369769e-06,
+      "loss": 1.7181,
+      "step": 7056
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3304987578826675e-06,
+      "loss": 0.744,
+      "step": 7057
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330403210395567e-06,
+      "loss": 1.1349,
+      "step": 7058
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330307662908466e-06,
+      "loss": 1.0237,
+      "step": 7059
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330212115421365e-06,
+      "loss": 1.441,
+      "step": 7060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330116567934264e-06,
+      "loss": 1.1317,
+      "step": 7061
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.330021020447163e-06,
+      "loss": 1.2504,
+      "step": 7062
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329925472960062e-06,
+      "loss": 1.0933,
+      "step": 7063
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.32982992547296e-06,
+      "loss": 1.0024,
+      "step": 7064
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329734377985859e-06,
+      "loss": 1.3258,
+      "step": 7065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329638830498758e-06,
+      "loss": 1.103,
+      "step": 7066
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329543283011657e-06,
+      "loss": 1.0879,
+      "step": 7067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329447735524556e-06,
+      "loss": 1.2855,
+      "step": 7068
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329352188037455e-06,
+      "loss": 0.8502,
+      "step": 7069
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329256640550354e-06,
+      "loss": 1.4851,
+      "step": 7070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.329161093063252e-06,
+      "loss": 1.0293,
+      "step": 7071
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3290655455761514e-06,
+      "loss": 1.1075,
+      "step": 7072
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3289699980890505e-06,
+      "loss": 0.8071,
+      "step": 7073
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3288744506019495e-06,
+      "loss": 0.8035,
+      "step": 7074
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3287789031148485e-06,
+      "loss": 1.1746,
+      "step": 7075
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328683355627748e-06,
+      "loss": 0.8375,
+      "step": 7076
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328587808140647e-06,
+      "loss": 0.9754,
+      "step": 7077
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328492260653545e-06,
+      "loss": 1.0931,
+      "step": 7078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328396713166444e-06,
+      "loss": 1.1158,
+      "step": 7079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328301165679343e-06,
+      "loss": 1.4222,
+      "step": 7080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328205618192242e-06,
+      "loss": 1.0109,
+      "step": 7081
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.328110070705141e-06,
+      "loss": 1.2334,
+      "step": 7082
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.32801452321804e-06,
+      "loss": 0.7849,
+      "step": 7083
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327918975730939e-06,
+      "loss": 1.1687,
+      "step": 7084
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327823428243837e-06,
+      "loss": 1.2132,
+      "step": 7085
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327727880756736e-06,
+      "loss": 1.1089,
+      "step": 7086
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327632333269635e-06,
+      "loss": 1.3665,
+      "step": 7087
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327536785782534e-06,
+      "loss": 0.8342,
+      "step": 7088
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327441238295433e-06,
+      "loss": 0.875,
+      "step": 7089
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3273456908083324e-06,
+      "loss": 1.0219,
+      "step": 7090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3272501433212315e-06,
+      "loss": 1.0781,
+      "step": 7091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.32715459583413e-06,
+      "loss": 1.4749,
+      "step": 7092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327059048347029e-06,
+      "loss": 0.9795,
+      "step": 7093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326963500859928e-06,
+      "loss": 1.2593,
+      "step": 7094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326867953372827e-06,
+      "loss": 0.9666,
+      "step": 7095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326772405885726e-06,
+      "loss": 1.1699,
+      "step": 7096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326676858398625e-06,
+      "loss": 1.431,
+      "step": 7097
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326581310911523e-06,
+      "loss": 1.1714,
+      "step": 7098
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326485763424422e-06,
+      "loss": 1.4325,
+      "step": 7099
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326390215937321e-06,
+      "loss": 1.3572,
+      "step": 7100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32629466845022e-06,
+      "loss": 1.1279,
+      "step": 7101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326199120963119e-06,
+      "loss": 1.1322,
+      "step": 7102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326103573476018e-06,
+      "loss": 0.8078,
+      "step": 7103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.326008025988917e-06,
+      "loss": 1.266,
+      "step": 7104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3259124785018155e-06,
+      "loss": 0.862,
+      "step": 7105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3258169310147145e-06,
+      "loss": 1.2629,
+      "step": 7106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3257213835276135e-06,
+      "loss": 0.8128,
+      "step": 7107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325625836040513e-06,
+      "loss": 1.0697,
+      "step": 7108
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325530288553411e-06,
+      "loss": 1.2455,
+      "step": 7109
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325434741066311e-06,
+      "loss": 1.3666,
+      "step": 7110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32533919357921e-06,
+      "loss": 0.8554,
+      "step": 7111
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325243646092108e-06,
+      "loss": 1.103,
+      "step": 7112
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325148098605007e-06,
+      "loss": 1.0659,
+      "step": 7113
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.325052551117906e-06,
+      "loss": 1.3291,
+      "step": 7114
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324957003630805e-06,
+      "loss": 1.0753,
+      "step": 7115
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324861456143703e-06,
+      "loss": 1.111,
+      "step": 7116
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324765908656603e-06,
+      "loss": 2.0303,
+      "step": 7117
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324670361169502e-06,
+      "loss": 0.8703,
+      "step": 7118
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3245748136824e-06,
+      "loss": 1.3009,
+      "step": 7119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324479266195299e-06,
+      "loss": 0.8409,
+      "step": 7120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324383718708198e-06,
+      "loss": 1.1671,
+      "step": 7121
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3242881712210974e-06,
+      "loss": 1.4583,
+      "step": 7122
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324192623733996e-06,
+      "loss": 1.2051,
+      "step": 7123
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.324097076246895e-06,
+      "loss": 0.7149,
+      "step": 7124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3240015287597945e-06,
+      "loss": 1.2308,
+      "step": 7125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323905981272693e-06,
+      "loss": 1.246,
+      "step": 7126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323810433785592e-06,
+      "loss": 1.0291,
+      "step": 7127
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323714886298491e-06,
+      "loss": 1.2884,
+      "step": 7128
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32361933881139e-06,
+      "loss": 1.1763,
+      "step": 7129
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323523791324288e-06,
+      "loss": 1.0045,
+      "step": 7130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323428243837187e-06,
+      "loss": 1.1468,
+      "step": 7131
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323332696350087e-06,
+      "loss": 1.3757,
+      "step": 7132
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323237148862985e-06,
+      "loss": 1.2243,
+      "step": 7133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323141601375884e-06,
+      "loss": 0.892,
+      "step": 7134
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.323046053888783e-06,
+      "loss": 1.2853,
+      "step": 7135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322950506401682e-06,
+      "loss": 1.1287,
+      "step": 7136
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3228549589145805e-06,
+      "loss": 0.8461,
+      "step": 7137
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3227594114274795e-06,
+      "loss": 1.3054,
+      "step": 7138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3226638639403785e-06,
+      "loss": 1.3836,
+      "step": 7139
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322568316453278e-06,
+      "loss": 1.2606,
+      "step": 7140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322472768966177e-06,
+      "loss": 1.2353,
+      "step": 7141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322377221479076e-06,
+      "loss": 1.3972,
+      "step": 7142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322281673991975e-06,
+      "loss": 1.2549,
+      "step": 7143
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322186126504873e-06,
+      "loss": 1.2043,
+      "step": 7144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322090579017772e-06,
+      "loss": 0.7533,
+      "step": 7145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.321995031530671e-06,
+      "loss": 1.4093,
+      "step": 7146
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.32189948404357e-06,
+      "loss": 1.1987,
+      "step": 7147
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.321803936556469e-06,
+      "loss": 1.539,
+      "step": 7148
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.321708389069368e-06,
+      "loss": 1.5777,
+      "step": 7149
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.321612841582266e-06,
+      "loss": 1.2911,
+      "step": 7150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.321517294095165e-06,
+      "loss": 1.1163,
+      "step": 7151
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.321421746608064e-06,
+      "loss": 1.1676,
+      "step": 7152
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.321326199120963e-06,
+      "loss": 1.2785,
+      "step": 7153
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3212306516338624e-06,
+      "loss": 1.2012,
+      "step": 7154
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3211351041467615e-06,
+      "loss": 1.1746,
+      "step": 7155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3210395566596605e-06,
+      "loss": 1.3719,
+      "step": 7156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320944009172559e-06,
+      "loss": 0.9302,
+      "step": 7157
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320848461685458e-06,
+      "loss": 1.0314,
+      "step": 7158
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320752914198357e-06,
+      "loss": 1.2448,
+      "step": 7159
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320657366711256e-06,
+      "loss": 1.0633,
+      "step": 7160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320561819224155e-06,
+      "loss": 1.0543,
+      "step": 7161
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320466271737054e-06,
+      "loss": 0.4879,
+      "step": 7162
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320370724249953e-06,
+      "loss": 0.8712,
+      "step": 7163
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320275176762851e-06,
+      "loss": 0.6987,
+      "step": 7164
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.32017962927575e-06,
+      "loss": 1.1397,
+      "step": 7165
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.320084081788649e-06,
+      "loss": 1.4908,
+      "step": 7166
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319988534301548e-06,
+      "loss": 1.2295,
+      "step": 7167
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319892986814447e-06,
+      "loss": 1.1965,
+      "step": 7168
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319797439327346e-06,
+      "loss": 1.336,
+      "step": 7169
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319701891840245e-06,
+      "loss": 1.6525,
+      "step": 7170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3196063443531435e-06,
+      "loss": 1.1805,
+      "step": 7171
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319510796866043e-06,
+      "loss": 1.2776,
+      "step": 7172
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319415249378942e-06,
+      "loss": 1.0609,
+      "step": 7173
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319319701891841e-06,
+      "loss": 1.1058,
+      "step": 7174
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.31922415440474e-06,
+      "loss": 1.6919,
+      "step": 7175
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319128606917639e-06,
+      "loss": 1.3277,
+      "step": 7176
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.319033059430538e-06,
+      "loss": 0.7355,
+      "step": 7177
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318937511943436e-06,
+      "loss": 0.9053,
+      "step": 7178
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318841964456335e-06,
+      "loss": 1.3723,
+      "step": 7179
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318746416969234e-06,
+      "loss": 1.1752,
+      "step": 7180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318650869482133e-06,
+      "loss": 1.1753,
+      "step": 7181
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318555321995032e-06,
+      "loss": 1.2953,
+      "step": 7182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318459774507931e-06,
+      "loss": 0.6723,
+      "step": 7183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.31836422702083e-06,
+      "loss": 1.5737,
+      "step": 7184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318268679533728e-06,
+      "loss": 1.3572,
+      "step": 7185
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.318173132046627e-06,
+      "loss": 0.9786,
+      "step": 7186
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3180775845595265e-06,
+      "loss": 1.3049,
+      "step": 7187
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3179820370724255e-06,
+      "loss": 1.2189,
+      "step": 7188
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3178864895853245e-06,
+      "loss": 1.4574,
+      "step": 7189
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317790942098224e-06,
+      "loss": 1.3771,
+      "step": 7190
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317695394611123e-06,
+      "loss": 1.3148,
+      "step": 7191
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317599847124021e-06,
+      "loss": 1.3318,
+      "step": 7192
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.31750429963692e-06,
+      "loss": 1.2346,
+      "step": 7193
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317408752149819e-06,
+      "loss": 1.1773,
+      "step": 7194
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317313204662718e-06,
+      "loss": 1.3136,
+      "step": 7195
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317217657175616e-06,
+      "loss": 1.1444,
+      "step": 7196
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317122109688516e-06,
+      "loss": 1.8548,
+      "step": 7197
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317026562201414e-06,
+      "loss": 1.1894,
+      "step": 7198
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316931014714313e-06,
+      "loss": 1.4301,
+      "step": 7199
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316835467227212e-06,
+      "loss": 0.7959,
+      "step": 7200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316739919740111e-06,
+      "loss": 1.1343,
+      "step": 7201
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3166443722530095e-06,
+      "loss": 1.2351,
+      "step": 7202
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3165488247659085e-06,
+      "loss": 1.1381,
+      "step": 7203
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.316453277278808e-06,
+      "loss": 0.9663,
+      "step": 7204
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.316357729791707e-06,
+      "loss": 1.1343,
+      "step": 7205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.316262182304606e-06,
+      "loss": 0.9533,
+      "step": 7206
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.316166634817505e-06,
+      "loss": 1.4384,
+      "step": 7207
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.316071087330404e-06,
+      "loss": 1.6204,
+      "step": 7208
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315975539843302e-06,
+      "loss": 0.9976,
+      "step": 7209
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315879992356201e-06,
+      "loss": 1.4723,
+      "step": 7210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315784444869101e-06,
+      "loss": 1.5605,
+      "step": 7211
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315688897381999e-06,
+      "loss": 1.1242,
+      "step": 7212
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315593349894898e-06,
+      "loss": 1.1102,
+      "step": 7213
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315497802407797e-06,
+      "loss": 0.9214,
+      "step": 7214
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315402254920696e-06,
+      "loss": 1.6993,
+      "step": 7215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315306707433594e-06,
+      "loss": 0.8855,
+      "step": 7216
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315211159946493e-06,
+      "loss": 0.6239,
+      "step": 7217
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315115612459393e-06,
+      "loss": 1.3285,
+      "step": 7218
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3150200649722915e-06,
+      "loss": 1.187,
+      "step": 7219
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3149245174851905e-06,
+      "loss": 1.2542,
+      "step": 7220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3148289699980895e-06,
+      "loss": 1.1386,
+      "step": 7221
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314733422510989e-06,
+      "loss": 0.9087,
+      "step": 7222
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314637875023887e-06,
+      "loss": 1.072,
+      "step": 7223
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314542327536786e-06,
+      "loss": 1.7055,
+      "step": 7224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314446780049686e-06,
+      "loss": 1.1321,
+      "step": 7225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314351232562584e-06,
+      "loss": 1.6517,
+      "step": 7226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314255685075483e-06,
+      "loss": 1.1895,
+      "step": 7227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314160137588382e-06,
+      "loss": 1.0178,
+      "step": 7228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.314064590101281e-06,
+      "loss": 0.7218,
+      "step": 7229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313969042614179e-06,
+      "loss": 1.12,
+      "step": 7230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313873495127078e-06,
+      "loss": 1.1976,
+      "step": 7231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313777947639977e-06,
+      "loss": 1.1115,
+      "step": 7232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313682400152876e-06,
+      "loss": 1.2253,
+      "step": 7233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313586852665775e-06,
+      "loss": 0.9237,
+      "step": 7234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313491305178674e-06,
+      "loss": 1.6872,
+      "step": 7235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313395757691573e-06,
+      "loss": 0.7268,
+      "step": 7236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313300210204472e-06,
+      "loss": 1.0491,
+      "step": 7237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313204662717371e-06,
+      "loss": 0.9131,
+      "step": 7238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.31310911523027e-06,
+      "loss": 1.0405,
+      "step": 7239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.313013567743169e-06,
+      "loss": 0.7986,
+      "step": 7240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312918020256068e-06,
+      "loss": 1.0541,
+      "step": 7241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312822472768967e-06,
+      "loss": 1.4506,
+      "step": 7242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312726925281866e-06,
+      "loss": 1.5815,
+      "step": 7243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312631377794764e-06,
+      "loss": 1.0778,
+      "step": 7244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312535830307663e-06,
+      "loss": 0.9295,
+      "step": 7245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312440282820562e-06,
+      "loss": 0.8782,
+      "step": 7246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312344735333461e-06,
+      "loss": 1.4586,
+      "step": 7247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.31224918784636e-06,
+      "loss": 0.7266,
+      "step": 7248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312153640359259e-06,
+      "loss": 1.4291,
+      "step": 7249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312058092872157e-06,
+      "loss": 1.4577,
+      "step": 7250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3119625453850565e-06,
+      "loss": 1.1903,
+      "step": 7251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3118669978979555e-06,
+      "loss": 0.6962,
+      "step": 7252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3117714504108545e-06,
+      "loss": 1.0224,
+      "step": 7253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3116759029237536e-06,
+      "loss": 0.9329,
+      "step": 7254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311580355436653e-06,
+      "loss": 0.9079,
+      "step": 7255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311484807949552e-06,
+      "loss": 1.3601,
+      "step": 7256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.31138926046245e-06,
+      "loss": 1.2807,
+      "step": 7257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311293712975349e-06,
+      "loss": 1.2661,
+      "step": 7258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311198165488248e-06,
+      "loss": 1.0234,
+      "step": 7259
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311102618001147e-06,
+      "loss": 0.8838,
+      "step": 7260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.311007070514046e-06,
+      "loss": 1.04,
+      "step": 7261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310911523026945e-06,
+      "loss": 1.8068,
+      "step": 7262
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310815975539844e-06,
+      "loss": 1.1045,
+      "step": 7263
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310720428052742e-06,
+      "loss": 1.1304,
+      "step": 7264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310624880565641e-06,
+      "loss": 1.0968,
+      "step": 7265
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.31052933307854e-06,
+      "loss": 1.33,
+      "step": 7266
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310433785591439e-06,
+      "loss": 0.9878,
+      "step": 7267
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3103382381043376e-06,
+      "loss": 1.0979,
+      "step": 7268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3102426906172375e-06,
+      "loss": 1.2324,
+      "step": 7269
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3101471431301365e-06,
+      "loss": 1.1724,
+      "step": 7270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.310051595643035e-06,
+      "loss": 1.2053,
+      "step": 7271
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309956048155934e-06,
+      "loss": 1.0197,
+      "step": 7272
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309860500668833e-06,
+      "loss": 1.5181,
+      "step": 7273
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309764953181732e-06,
+      "loss": 1.2875,
+      "step": 7274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30966940569463e-06,
+      "loss": 0.9687,
+      "step": 7275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30957385820753e-06,
+      "loss": 0.8247,
+      "step": 7276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309478310720429e-06,
+      "loss": 1.2,
+      "step": 7277
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309382763233327e-06,
+      "loss": 1.4958,
+      "step": 7278
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309287215746226e-06,
+      "loss": 1.1722,
+      "step": 7279
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309191668259125e-06,
+      "loss": 1.0797,
+      "step": 7280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309096120772024e-06,
+      "loss": 0.9877,
+      "step": 7281
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.309000573284922e-06,
+      "loss": 1.6448,
+      "step": 7282
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308905025797822e-06,
+      "loss": 0.9632,
+      "step": 7283
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308809478310721e-06,
+      "loss": 1.2522,
+      "step": 7284
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3087139308236195e-06,
+      "loss": 1.2273,
+      "step": 7285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3086183833365186e-06,
+      "loss": 1.0151,
+      "step": 7286
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308522835849418e-06,
+      "loss": 1.1601,
+      "step": 7287
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308427288362317e-06,
+      "loss": 1.044,
+      "step": 7288
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308331740875215e-06,
+      "loss": 1.5468,
+      "step": 7289
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308236193388115e-06,
+      "loss": 1.363,
+      "step": 7290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308140645901013e-06,
+      "loss": 0.9339,
+      "step": 7291
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.308045098413912e-06,
+      "loss": 1.0652,
+      "step": 7292
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307949550926811e-06,
+      "loss": 1.14,
+      "step": 7293
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30785400343971e-06,
+      "loss": 1.1941,
+      "step": 7294
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307758455952609e-06,
+      "loss": 1.0103,
+      "step": 7295
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307662908465507e-06,
+      "loss": 1.1375,
+      "step": 7296
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307567360978407e-06,
+      "loss": 1.3636,
+      "step": 7297
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307471813491305e-06,
+      "loss": 1.3398,
+      "step": 7298
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307376266004204e-06,
+      "loss": 1.7629,
+      "step": 7299
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307280718517103e-06,
+      "loss": 1.2208,
+      "step": 7300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3071851710300025e-06,
+      "loss": 0.9054,
+      "step": 7301
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307089623542901e-06,
+      "loss": 0.774,
+      "step": 7302
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3069940760558e-06,
+      "loss": 1.1184,
+      "step": 7303
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306898528568699e-06,
+      "loss": 1.0028,
+      "step": 7304
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306802981081598e-06,
+      "loss": 1.3469,
+      "step": 7305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306707433594497e-06,
+      "loss": 1.2571,
+      "step": 7306
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.306611886107396e-06,
+      "loss": 1.6643,
+      "step": 7307
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306516338620295e-06,
+      "loss": 0.8015,
+      "step": 7308
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306420791133193e-06,
+      "loss": 1.5185,
+      "step": 7309
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306325243646092e-06,
+      "loss": 1.6281,
+      "step": 7310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306229696158991e-06,
+      "loss": 1.0058,
+      "step": 7311
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30613414867189e-06,
+      "loss": 1.1408,
+      "step": 7312
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306038601184789e-06,
+      "loss": 1.4032,
+      "step": 7313
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305943053697688e-06,
+      "loss": 1.1728,
+      "step": 7314
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305847506210587e-06,
+      "loss": 1.1521,
+      "step": 7315
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3057519587234855e-06,
+      "loss": 0.988,
+      "step": 7316
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3056564112363845e-06,
+      "loss": 1.2811,
+      "step": 7317
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3055608637492836e-06,
+      "loss": 0.8576,
+      "step": 7318
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305465316262183e-06,
+      "loss": 1.3456,
+      "step": 7319
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305369768775082e-06,
+      "loss": 1.2029,
+      "step": 7320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305274221287981e-06,
+      "loss": 0.7334,
+      "step": 7321
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30517867380088e-06,
+      "loss": 0.9273,
+      "step": 7322
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.305083126313778e-06,
+      "loss": 0.8627,
+      "step": 7323
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304987578826677e-06,
+      "loss": 1.1187,
+      "step": 7324
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304892031339576e-06,
+      "loss": 1.4272,
+      "step": 7325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304796483852475e-06,
+      "loss": 1.6586,
+      "step": 7326
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304700936365374e-06,
+      "loss": 1.2714,
+      "step": 7327
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304605388878273e-06,
+      "loss": 0.6318,
+      "step": 7328
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304509841391172e-06,
+      "loss": 0.8946,
+      "step": 7329
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30441429390407e-06,
+      "loss": 1.1363,
+      "step": 7330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304318746416969e-06,
+      "loss": 1.1615,
+      "step": 7331
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.304223198929868e-06,
+      "loss": 1.2651,
+      "step": 7332
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3041276514427675e-06,
+      "loss": 1.2206,
+      "step": 7333
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3040321039556665e-06,
+      "loss": 0.7573,
+      "step": 7334
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3039365564685655e-06,
+      "loss": 0.989,
+      "step": 7335
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3038410089814646e-06,
+      "loss": 1.1115,
+      "step": 7336
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303745461494363e-06,
+      "loss": 1.5756,
+      "step": 7337
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303649914007262e-06,
+      "loss": 1.052,
+      "step": 7338
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303554366520161e-06,
+      "loss": 1.3188,
+      "step": 7339
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30345881903306e-06,
+      "loss": 1.3168,
+      "step": 7340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303363271545959e-06,
+      "loss": 0.8075,
+      "step": 7341
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303267724058858e-06,
+      "loss": 1.1569,
+      "step": 7342
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303172176571756e-06,
+      "loss": 0.8232,
+      "step": 7343
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303076629084655e-06,
+      "loss": 0.9449,
+      "step": 7344
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302981081597554e-06,
+      "loss": 1.3859,
+      "step": 7345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302885534110453e-06,
+      "loss": 0.9717,
+      "step": 7346
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302789986623352e-06,
+      "loss": 1.0996,
+      "step": 7347
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302694439136251e-06,
+      "loss": 1.3606,
+      "step": 7348
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30259889164915e-06,
+      "loss": 0.9865,
+      "step": 7349
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3025033441620486e-06,
+      "loss": 1.1708,
+      "step": 7350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302407796674948e-06,
+      "loss": 1.1159,
+      "step": 7351
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302312249187847e-06,
+      "loss": 1.2467,
+      "step": 7352
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302216701700746e-06,
+      "loss": 1.1051,
+      "step": 7353
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302121154213644e-06,
+      "loss": 1.212,
+      "step": 7354
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302025606726544e-06,
+      "loss": 0.709,
+      "step": 7355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301930059239443e-06,
+      "loss": 1.1751,
+      "step": 7356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301834511752341e-06,
+      "loss": 1.3367,
+      "step": 7357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30173896426524e-06,
+      "loss": 1.0363,
+      "step": 7358
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301643416778139e-06,
+      "loss": 1.1911,
+      "step": 7359
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.301547869291038e-06,
+      "loss": 1.2573,
+      "step": 7360
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.301452321803936e-06,
+      "loss": 1.4043,
+      "step": 7361
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.301356774316836e-06,
+      "loss": 0.8127,
+      "step": 7362
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.301261226829735e-06,
+      "loss": 1.1067,
+      "step": 7363
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.301165679342633e-06,
+      "loss": 0.8718,
+      "step": 7364
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3010701318555325e-06,
+      "loss": 1.5485,
+      "step": 7365
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3009745843684315e-06,
+      "loss": 1.1203,
+      "step": 7366
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3008790368813305e-06,
+      "loss": 0.6851,
+      "step": 7367
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300783489394229e-06,
+      "loss": 1.2107,
+      "step": 7368
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300687941907129e-06,
+      "loss": 1.1438,
+      "step": 7369
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300592394420028e-06,
+      "loss": 0.8859,
+      "step": 7370
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300496846932926e-06,
+      "loss": 1.4323,
+      "step": 7371
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300401299445825e-06,
+      "loss": 1.1085,
+      "step": 7372
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300305751958724e-06,
+      "loss": 1.5912,
+      "step": 7373
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300210204471623e-06,
+      "loss": 1.0759,
+      "step": 7374
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.300114656984521e-06,
+      "loss": 1.1353,
+      "step": 7375
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.30001910949742e-06,
+      "loss": 1.2717,
+      "step": 7376
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.29992356201032e-06,
+      "loss": 1.0022,
+      "step": 7377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299828014523218e-06,
+      "loss": 0.6897,
+      "step": 7378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299732467036117e-06,
+      "loss": 1.3469,
+      "step": 7379
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299636919549016e-06,
+      "loss": 1.0395,
+      "step": 7380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299541372061915e-06,
+      "loss": 1.3881,
+      "step": 7381
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2994458245748136e-06,
+      "loss": 1.348,
+      "step": 7382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299350277087713e-06,
+      "loss": 1.631,
+      "step": 7383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2992547296006125e-06,
+      "loss": 0.788,
+      "step": 7384
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.299159182113511e-06,
+      "loss": 1.3415,
+      "step": 7385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.29906363462641e-06,
+      "loss": 1.0016,
+      "step": 7386
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298968087139309e-06,
+      "loss": 0.7784,
+      "step": 7387
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298872539652208e-06,
+      "loss": 0.9288,
+      "step": 7388
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298776992165106e-06,
+      "loss": 1.6198,
+      "step": 7389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298681444678005e-06,
+      "loss": 1.1575,
+      "step": 7390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298585897190904e-06,
+      "loss": 1.2963,
+      "step": 7391
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298490349703803e-06,
+      "loss": 1.066,
+      "step": 7392
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298394802216702e-06,
+      "loss": 1.0849,
+      "step": 7393
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298299254729601e-06,
+      "loss": 0.9951,
+      "step": 7394
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298203707242499e-06,
+      "loss": 1.0929,
+      "step": 7395
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.298108159755398e-06,
+      "loss": 0.734,
+      "step": 7396
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2980126122682975e-06,
+      "loss": 1.3029,
+      "step": 7397
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2979170647811965e-06,
+      "loss": 1.1386,
+      "step": 7398
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2978215172940955e-06,
+      "loss": 1.376,
+      "step": 7399
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2977259698069946e-06,
+      "loss": 1.089,
+      "step": 7400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297630422319894e-06,
+      "loss": 1.9268,
+      "step": 7401
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297534874832792e-06,
+      "loss": 1.9314,
+      "step": 7402
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297439327345691e-06,
+      "loss": 1.4403,
+      "step": 7403
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.29734377985859e-06,
+      "loss": 0.9851,
+      "step": 7404
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297248232371489e-06,
+      "loss": 0.9448,
+      "step": 7405
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297152684884388e-06,
+      "loss": 1.325,
+      "step": 7406
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297057137397287e-06,
+      "loss": 1.4415,
+      "step": 7407
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296961589910186e-06,
+      "loss": 0.9833,
+      "step": 7408
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296866042423084e-06,
+      "loss": 1.498,
+      "step": 7409
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296770494935983e-06,
+      "loss": 1.0244,
+      "step": 7410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296674947448882e-06,
+      "loss": 1.2465,
+      "step": 7411
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.296579399961781e-06,
+      "loss": 1.2052,
+      "step": 7412
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.29648385247468e-06,
+      "loss": 1.6865,
+      "step": 7413
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.296388304987579e-06,
+      "loss": 0.8196,
+      "step": 7414
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2962927575004785e-06,
+      "loss": 0.8125,
+      "step": 7415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.296197210013377e-06,
+      "loss": 1.2575,
+      "step": 7416
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.296101662526276e-06,
+      "loss": 0.8767,
+      "step": 7417
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.296006115039175e-06,
+      "loss": 1.7062,
+      "step": 7418
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295910567552074e-06,
+      "loss": 1.0941,
+      "step": 7419
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295815020064973e-06,
+      "loss": 1.2096,
+      "step": 7420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295719472577872e-06,
+      "loss": 1.59,
+      "step": 7421
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295623925090771e-06,
+      "loss": 1.4298,
+      "step": 7422
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295528377603669e-06,
+      "loss": 0.8144,
+      "step": 7423
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295432830116568e-06,
+      "loss": 1.4558,
+      "step": 7424
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295337282629467e-06,
+      "loss": 1.1749,
+      "step": 7425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295241735142366e-06,
+      "loss": 1.1766,
+      "step": 7426
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295146187655265e-06,
+      "loss": 1.1569,
+      "step": 7427
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.295050640168164e-06,
+      "loss": 1.4499,
+      "step": 7428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294955092681063e-06,
+      "loss": 0.9187,
+      "step": 7429
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2948595451939615e-06,
+      "loss": 1.2857,
+      "step": 7430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2947639977068605e-06,
+      "loss": 0.9217,
+      "step": 7431
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2946684502197596e-06,
+      "loss": 0.8302,
+      "step": 7432
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294572902732659e-06,
+      "loss": 0.8173,
+      "step": 7433
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294477355245558e-06,
+      "loss": 0.8447,
+      "step": 7434
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294381807758457e-06,
+      "loss": 1.5559,
+      "step": 7435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294286260271356e-06,
+      "loss": 1.5498,
+      "step": 7436
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294190712784254e-06,
+      "loss": 1.4556,
+      "step": 7437
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.294095165297153e-06,
+      "loss": 0.5646,
+      "step": 7438
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293999617810052e-06,
+      "loss": 1.0523,
+      "step": 7439
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293904070322951e-06,
+      "loss": 1.1838,
+      "step": 7440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.29380852283585e-06,
+      "loss": 1.2093,
+      "step": 7441
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293712975348749e-06,
+      "loss": 1.3168,
+      "step": 7442
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293617427861647e-06,
+      "loss": 0.9907,
+      "step": 7443
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293521880374546e-06,
+      "loss": 1.1779,
+      "step": 7444
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293426332887445e-06,
+      "loss": 1.1512,
+      "step": 7445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293330785400344e-06,
+      "loss": 0.7762,
+      "step": 7446
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2932352379132435e-06,
+      "loss": 1.4809,
+      "step": 7447
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.293139690426142e-06,
+      "loss": 0.923,
+      "step": 7448
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2930441429390415e-06,
+      "loss": 1.5403,
+      "step": 7449
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.29294859545194e-06,
+      "loss": 0.9322,
+      "step": 7450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292853047964839e-06,
+      "loss": 1.1746,
+      "step": 7451
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292757500477738e-06,
+      "loss": 1.1264,
+      "step": 7452
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292661952990637e-06,
+      "loss": 1.2015,
+      "step": 7453
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292566405503535e-06,
+      "loss": 0.5087,
+      "step": 7454
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292470858016434e-06,
+      "loss": 1.2998,
+      "step": 7455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292375310529334e-06,
+      "loss": 1.1271,
+      "step": 7456
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292279763042232e-06,
+      "loss": 1.0956,
+      "step": 7457
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292184215555131e-06,
+      "loss": 0.9312,
+      "step": 7458
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.29208866806803e-06,
+      "loss": 1.1809,
+      "step": 7459
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.291993120580929e-06,
+      "loss": 1.4558,
+      "step": 7460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2918975730938275e-06,
+      "loss": 1.1988,
+      "step": 7461
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2918020256067265e-06,
+      "loss": 1.0098,
+      "step": 7462
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.291706478119626e-06,
+      "loss": 1.1774,
+      "step": 7463
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2916109306325246e-06,
+      "loss": 1.456,
+      "step": 7464
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.291515383145424e-06,
+      "loss": 1.1402,
+      "step": 7465
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.291419835658323e-06,
+      "loss": 1.4109,
+      "step": 7466
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.291324288171222e-06,
+      "loss": 1.0744,
+      "step": 7467
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.29122874068412e-06,
+      "loss": 1.2147,
+      "step": 7468
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.291133193197019e-06,
+      "loss": 1.2105,
+      "step": 7469
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.291037645709919e-06,
+      "loss": 0.804,
+      "step": 7470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290942098222817e-06,
+      "loss": 1.364,
+      "step": 7471
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290846550735716e-06,
+      "loss": 1.368,
+      "step": 7472
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290751003248615e-06,
+      "loss": 1.4976,
+      "step": 7473
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290655455761514e-06,
+      "loss": 1.3635,
+      "step": 7474
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290559908274412e-06,
+      "loss": 1.0906,
+      "step": 7475
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290464360787311e-06,
+      "loss": 1.4124,
+      "step": 7476
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290368813300211e-06,
+      "loss": 1.058,
+      "step": 7477
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.290273265813109e-06,
+      "loss": 0.8998,
+      "step": 7478
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2901777183260085e-06,
+      "loss": 1.6424,
+      "step": 7479
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2900821708389075e-06,
+      "loss": 1.1449,
+      "step": 7480
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2899866233518065e-06,
+      "loss": 0.9678,
+      "step": 7481
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289891075864705e-06,
+      "loss": 1.194,
+      "step": 7482
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289795528377604e-06,
+      "loss": 1.2069,
+      "step": 7483
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289699980890503e-06,
+      "loss": 1.0209,
+      "step": 7484
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289604433403402e-06,
+      "loss": 1.8901,
+      "step": 7485
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289508885916301e-06,
+      "loss": 1.2892,
+      "step": 7486
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2894133384292e-06,
+      "loss": 1.2225,
+      "step": 7487
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289317790942099e-06,
+      "loss": 1.0967,
+      "step": 7488
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289222243454997e-06,
+      "loss": 1.1692,
+      "step": 7489
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289126695967896e-06,
+      "loss": 1.0565,
+      "step": 7490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.289031148480795e-06,
+      "loss": 0.8415,
+      "step": 7491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288935600993694e-06,
+      "loss": 1.0465,
+      "step": 7492
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288840053506593e-06,
+      "loss": 1.15,
+      "step": 7493
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288744506019492e-06,
+      "loss": 1.6084,
+      "step": 7494
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2886489585323905e-06,
+      "loss": 1.1482,
+      "step": 7495
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2885534110452896e-06,
+      "loss": 1.8456,
+      "step": 7496
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288457863558189e-06,
+      "loss": 1.3115,
+      "step": 7497
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288362316071088e-06,
+      "loss": 1.3589,
+      "step": 7498
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288266768583987e-06,
+      "loss": 1.4395,
+      "step": 7499
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288171221096886e-06,
+      "loss": 0.6611,
+      "step": 7500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.288075673609785e-06,
+      "loss": 1.089,
+      "step": 7501
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287980126122683e-06,
+      "loss": 1.3885,
+      "step": 7502
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287884578635582e-06,
+      "loss": 1.2097,
+      "step": 7503
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287789031148481e-06,
+      "loss": 1.0639,
+      "step": 7504
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.28769348366138e-06,
+      "loss": 1.2062,
+      "step": 7505
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287597936174279e-06,
+      "loss": 1.1925,
+      "step": 7506
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287502388687178e-06,
+      "loss": 0.8487,
+      "step": 7507
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287406841200077e-06,
+      "loss": 0.932,
+      "step": 7508
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287311293712975e-06,
+      "loss": 1.3895,
+      "step": 7509
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287215746225874e-06,
+      "loss": 1.2564,
+      "step": 7510
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2871201987387735e-06,
+      "loss": 1.2283,
+      "step": 7511
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2870246512516725e-06,
+      "loss": 0.858,
+      "step": 7512
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2869291037645715e-06,
+      "loss": 1.4507,
+      "step": 7513
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2868335562774706e-06,
+      "loss": 0.9992,
+      "step": 7514
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.28673800879037e-06,
+      "loss": 0.8137,
+      "step": 7515
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.286642461303268e-06,
+      "loss": 0.9396,
+      "step": 7516
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286546913816167e-06,
+      "loss": 1.5963,
+      "step": 7517
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286451366329066e-06,
+      "loss": 1.2236,
+      "step": 7518
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286355818841965e-06,
+      "loss": 0.7628,
+      "step": 7519
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286260271354863e-06,
+      "loss": 1.0128,
+      "step": 7520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286164723867763e-06,
+      "loss": 0.8934,
+      "step": 7521
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.286069176380662e-06,
+      "loss": 1.0903,
+      "step": 7522
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28597362889356e-06,
+      "loss": 1.5554,
+      "step": 7523
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285878081406459e-06,
+      "loss": 1.4572,
+      "step": 7524
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285782533919358e-06,
+      "loss": 0.9025,
+      "step": 7525
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285686986432257e-06,
+      "loss": 0.9984,
+      "step": 7526
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2855914389451555e-06,
+      "loss": 0.953,
+      "step": 7527
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285495891458055e-06,
+      "loss": 1.1714,
+      "step": 7528
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2854003439709545e-06,
+      "loss": 1.4646,
+      "step": 7529
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285304796483853e-06,
+      "loss": 1.313,
+      "step": 7530
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285209248996752e-06,
+      "loss": 1.0026,
+      "step": 7531
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.285113701509651e-06,
+      "loss": 1.1929,
+      "step": 7532
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28501815402255e-06,
+      "loss": 1.2041,
+      "step": 7533
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284922606535448e-06,
+      "loss": 1.1876,
+      "step": 7534
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284827059048348e-06,
+      "loss": 1.1985,
+      "step": 7535
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284731511561246e-06,
+      "loss": 0.7297,
+      "step": 7536
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284635964074145e-06,
+      "loss": 1.3164,
+      "step": 7537
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284540416587044e-06,
+      "loss": 0.9312,
+      "step": 7538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284444869099943e-06,
+      "loss": 1.8209,
+      "step": 7539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.284349321612842e-06,
+      "loss": 0.9694,
+      "step": 7540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28425377412574e-06,
+      "loss": 1.4992,
+      "step": 7541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28415822663864e-06,
+      "loss": 1.0561,
+      "step": 7542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2840626791515385e-06,
+      "loss": 1.5002,
+      "step": 7543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2839671316644375e-06,
+      "loss": 0.885,
+      "step": 7544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2838715841773365e-06,
+      "loss": 1.091,
+      "step": 7545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2837760366902356e-06,
+      "loss": 1.0828,
+      "step": 7546
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283680489203134e-06,
+      "loss": 1.1585,
+      "step": 7547
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283584941716033e-06,
+      "loss": 0.8708,
+      "step": 7548
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283489394228933e-06,
+      "loss": 1.309,
+      "step": 7549
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283393846741831e-06,
+      "loss": 1.2165,
+      "step": 7550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28329829925473e-06,
+      "loss": 1.3985,
+      "step": 7551
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283202751767629e-06,
+      "loss": 1.3076,
+      "step": 7552
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283107204280528e-06,
+      "loss": 1.2638,
+      "step": 7553
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.283011656793426e-06,
+      "loss": 1.4286,
+      "step": 7554
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282916109306325e-06,
+      "loss": 1.2563,
+      "step": 7555
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282820561819224e-06,
+      "loss": 1.4556,
+      "step": 7556
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282725014332123e-06,
+      "loss": 1.1807,
+      "step": 7557
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282629466845022e-06,
+      "loss": 0.719,
+      "step": 7558
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282533919357921e-06,
+      "loss": 0.8492,
+      "step": 7559
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.28243837187082e-06,
+      "loss": 1.0552,
+      "step": 7560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282342824383719e-06,
+      "loss": 1.4717,
+      "step": 7561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282247276896618e-06,
+      "loss": 1.1458,
+      "step": 7562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282151729409517e-06,
+      "loss": 1.2655,
+      "step": 7563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282056181922416e-06,
+      "loss": 0.9517,
+      "step": 7564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281960634435315e-06,
+      "loss": 1.2604,
+      "step": 7565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281865086948214e-06,
+      "loss": 1.1606,
+      "step": 7566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281769539461113e-06,
+      "loss": 0.96,
+      "step": 7567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.281673991974011e-06,
+      "loss": 1.2931,
+      "step": 7568
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.28157844448691e-06,
+      "loss": 1.3661,
+      "step": 7569
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.281482896999809e-06,
+      "loss": 1.3173,
+      "step": 7570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.281387349512708e-06,
+      "loss": 0.9107,
+      "step": 7571
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.281291802025607e-06,
+      "loss": 0.9414,
+      "step": 7572
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.281196254538506e-06,
+      "loss": 1.1656,
+      "step": 7573
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.281100707051405e-06,
+      "loss": 1.7468,
+      "step": 7574
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2810051595643035e-06,
+      "loss": 1.0819,
+      "step": 7575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2809096120772025e-06,
+      "loss": 1.5672,
+      "step": 7576
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2808140645901015e-06,
+      "loss": 1.3186,
+      "step": 7577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2807185171030006e-06,
+      "loss": 1.361,
+      "step": 7578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2806229696159e-06,
+      "loss": 0.676,
+      "step": 7579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280527422128799e-06,
+      "loss": 1.4274,
+      "step": 7580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280431874641698e-06,
+      "loss": 1.4256,
+      "step": 7581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280336327154596e-06,
+      "loss": 1.4114,
+      "step": 7582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280240779667495e-06,
+      "loss": 0.9415,
+      "step": 7583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280145232180394e-06,
+      "loss": 1.6287,
+      "step": 7584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.280049684693293e-06,
+      "loss": 1.3357,
+      "step": 7585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279954137206192e-06,
+      "loss": 0.942,
+      "step": 7586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279858589719091e-06,
+      "loss": 0.8185,
+      "step": 7587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279763042231989e-06,
+      "loss": 0.9661,
+      "step": 7588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279667494744888e-06,
+      "loss": 1.2111,
+      "step": 7589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279571947257787e-06,
+      "loss": 1.4272,
+      "step": 7590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279476399770686e-06,
+      "loss": 1.3838,
+      "step": 7591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279380852283585e-06,
+      "loss": 1.2513,
+      "step": 7592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2792853047964845e-06,
+      "loss": 1.2942,
+      "step": 7593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2791897573093835e-06,
+      "loss": 1.7094,
+      "step": 7594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.279094209822282e-06,
+      "loss": 1.0647,
+      "step": 7595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278998662335181e-06,
+      "loss": 1.3021,
+      "step": 7596
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.27890311484808e-06,
+      "loss": 1.2686,
+      "step": 7597
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278807567360979e-06,
+      "loss": 1.2432,
+      "step": 7598
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278712019873877e-06,
+      "loss": 0.8007,
+      "step": 7599
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278616472386777e-06,
+      "loss": 0.8597,
+      "step": 7600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278520924899676e-06,
+      "loss": 0.9453,
+      "step": 7601
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278425377412574e-06,
+      "loss": 1.1575,
+      "step": 7602
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278329829925473e-06,
+      "loss": 0.6492,
+      "step": 7603
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278234282438372e-06,
+      "loss": 1.072,
+      "step": 7604
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278138734951271e-06,
+      "loss": 1.3033,
+      "step": 7605
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.278043187464169e-06,
+      "loss": 0.9897,
+      "step": 7606
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277947639977069e-06,
+      "loss": 1.3016,
+      "step": 7607
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277852092489968e-06,
+      "loss": 0.8367,
+      "step": 7608
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2777565450028665e-06,
+      "loss": 1.1109,
+      "step": 7609
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2776609975157656e-06,
+      "loss": 1.071,
+      "step": 7610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277565450028665e-06,
+      "loss": 0.9304,
+      "step": 7611
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277469902541564e-06,
+      "loss": 1.2417,
+      "step": 7612
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277374355054462e-06,
+      "loss": 0.879,
+      "step": 7613
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277278807567362e-06,
+      "loss": 1.6005,
+      "step": 7614
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277183260080261e-06,
+      "loss": 1.1469,
+      "step": 7615
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.277087712593159e-06,
+      "loss": 1.3295,
+      "step": 7616
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276992165106058e-06,
+      "loss": 1.0189,
+      "step": 7617
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276896617618957e-06,
+      "loss": 1.4205,
+      "step": 7618
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276801070131856e-06,
+      "loss": 0.9742,
+      "step": 7619
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276705522644754e-06,
+      "loss": 0.8155,
+      "step": 7620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.276609975157654e-06,
+      "loss": 1.1019,
+      "step": 7621
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.276514427670553e-06,
+      "loss": 0.8899,
+      "step": 7622
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.276418880183451e-06,
+      "loss": 0.8736,
+      "step": 7623
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27632333269635e-06,
+      "loss": 1.21,
+      "step": 7624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2762277852092495e-06,
+      "loss": 1.0316,
+      "step": 7625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2761322377221485e-06,
+      "loss": 1.1473,
+      "step": 7626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.276036690235047e-06,
+      "loss": 1.0677,
+      "step": 7627
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275941142747946e-06,
+      "loss": 1.1014,
+      "step": 7628
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275845595260846e-06,
+      "loss": 1.1371,
+      "step": 7629
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275750047773744e-06,
+      "loss": 1.067,
+      "step": 7630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275654500286643e-06,
+      "loss": 1.0649,
+      "step": 7631
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275558952799542e-06,
+      "loss": 1.4046,
+      "step": 7632
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275463405312441e-06,
+      "loss": 1.3823,
+      "step": 7633
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275367857825339e-06,
+      "loss": 1.1199,
+      "step": 7634
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275272310338238e-06,
+      "loss": 1.2007,
+      "step": 7635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275176762851137e-06,
+      "loss": 1.1227,
+      "step": 7636
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.275081215364036e-06,
+      "loss": 1.0305,
+      "step": 7637
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274985667876935e-06,
+      "loss": 0.9833,
+      "step": 7638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274890120389834e-06,
+      "loss": 1.1587,
+      "step": 7639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274794572902733e-06,
+      "loss": 1.542,
+      "step": 7640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2746990254156315e-06,
+      "loss": 1.1223,
+      "step": 7641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2746034779285306e-06,
+      "loss": 1.4325,
+      "step": 7642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27450793044143e-06,
+      "loss": 1.1576,
+      "step": 7643
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274412382954329e-06,
+      "loss": 1.1189,
+      "step": 7644
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274316835467228e-06,
+      "loss": 1.3381,
+      "step": 7645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274221287980127e-06,
+      "loss": 1.3365,
+      "step": 7646
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274125740493025e-06,
+      "loss": 1.1476,
+      "step": 7647
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.274030193005924e-06,
+      "loss": 1.0742,
+      "step": 7648
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273934645518823e-06,
+      "loss": 0.8503,
+      "step": 7649
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273839098031722e-06,
+      "loss": 0.9959,
+      "step": 7650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273743550544621e-06,
+      "loss": 1.1271,
+      "step": 7651
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27364800305752e-06,
+      "loss": 1.1596,
+      "step": 7652
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273552455570419e-06,
+      "loss": 1.0065,
+      "step": 7653
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273456908083317e-06,
+      "loss": 1.1827,
+      "step": 7654
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273361360596216e-06,
+      "loss": 1.4262,
+      "step": 7655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.273265813109115e-06,
+      "loss": 1.0909,
+      "step": 7656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2731702656220145e-06,
+      "loss": 1.0856,
+      "step": 7657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2730747181349135e-06,
+      "loss": 1.0069,
+      "step": 7658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2729791706478125e-06,
+      "loss": 1.1312,
+      "step": 7659
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2728836231607116e-06,
+      "loss": 1.4178,
+      "step": 7660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27278807567361e-06,
+      "loss": 0.922,
+      "step": 7661
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272692528186509e-06,
+      "loss": 1.0554,
+      "step": 7662
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272596980699408e-06,
+      "loss": 1.1577,
+      "step": 7663
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272501433212307e-06,
+      "loss": 0.8318,
+      "step": 7664
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272405885725206e-06,
+      "loss": 0.9497,
+      "step": 7665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272310338238105e-06,
+      "loss": 1.0103,
+      "step": 7666
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272214790751004e-06,
+      "loss": 1.098,
+      "step": 7667
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272119243263902e-06,
+      "loss": 1.0936,
+      "step": 7668
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.272023695776801e-06,
+      "loss": 1.0249,
+      "step": 7669
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2719281482897e-06,
+      "loss": 0.9955,
+      "step": 7670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271832600802599e-06,
+      "loss": 1.0189,
+      "step": 7671
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271737053315498e-06,
+      "loss": 1.0747,
+      "step": 7672
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271641505828397e-06,
+      "loss": 1.2185,
+      "step": 7673
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271545958341296e-06,
+      "loss": 1.2215,
+      "step": 7674
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271450410854195e-06,
+      "loss": 1.5946,
+      "step": 7675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271354863367094e-06,
+      "loss": 0.9211,
+      "step": 7676
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271259315879993e-06,
+      "loss": 0.9119,
+      "step": 7677
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271163768392892e-06,
+      "loss": 1.2367,
+      "step": 7678
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.271068220905791e-06,
+      "loss": 1.3629,
+      "step": 7679
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27097267341869e-06,
+      "loss": 0.9313,
+      "step": 7680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270877125931589e-06,
+      "loss": 1.0755,
+      "step": 7681
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270781578444487e-06,
+      "loss": 1.1576,
+      "step": 7682
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270686030957386e-06,
+      "loss": 1.331,
+      "step": 7683
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270590483470285e-06,
+      "loss": 1.2888,
+      "step": 7684
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270494935983184e-06,
+      "loss": 1.2791,
+      "step": 7685
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270399388496083e-06,
+      "loss": 1.3025,
+      "step": 7686
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.270303841008982e-06,
+      "loss": 1.1096,
+      "step": 7687
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.27020829352188e-06,
+      "loss": 0.8584,
+      "step": 7688
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2701127460347795e-06,
+      "loss": 0.9651,
+      "step": 7689
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2700171985476785e-06,
+      "loss": 1.084,
+      "step": 7690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2699216510605775e-06,
+      "loss": 0.9671,
+      "step": 7691
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2698261035734766e-06,
+      "loss": 1.1793,
+      "step": 7692
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269730556086376e-06,
+      "loss": 1.0425,
+      "step": 7693
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269635008599275e-06,
+      "loss": 1.0733,
+      "step": 7694
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269539461112173e-06,
+      "loss": 1.3384,
+      "step": 7695
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269443913625072e-06,
+      "loss": 1.1669,
+      "step": 7696
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269348366137971e-06,
+      "loss": 1.1509,
+      "step": 7697
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.26925281865087e-06,
+      "loss": 0.8549,
+      "step": 7698
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269157271163768e-06,
+      "loss": 1.2182,
+      "step": 7699
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.269061723676667e-06,
+      "loss": 1.4778,
+      "step": 7700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268966176189567e-06,
+      "loss": 1.8452,
+      "step": 7701
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268870628702465e-06,
+      "loss": 1.1312,
+      "step": 7702
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268775081215364e-06,
+      "loss": 1.0913,
+      "step": 7703
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268679533728263e-06,
+      "loss": 1.1613,
+      "step": 7704
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268583986241162e-06,
+      "loss": 1.2102,
+      "step": 7705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2684884387540606e-06,
+      "loss": 1.3028,
+      "step": 7706
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.26839289126696e-06,
+      "loss": 0.9132,
+      "step": 7707
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2682973437798595e-06,
+      "loss": 0.8365,
+      "step": 7708
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268201796292758e-06,
+      "loss": 1.2261,
+      "step": 7709
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268106248805657e-06,
+      "loss": 1.4132,
+      "step": 7710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.268010701318556e-06,
+      "loss": 1.411,
+      "step": 7711
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267915153831455e-06,
+      "loss": 1.2987,
+      "step": 7712
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267819606344353e-06,
+      "loss": 1.0772,
+      "step": 7713
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267724058857252e-06,
+      "loss": 1.2218,
+      "step": 7714
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267628511370152e-06,
+      "loss": 1.5803,
+      "step": 7715
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.26753296388305e-06,
+      "loss": 1.1503,
+      "step": 7716
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267437416395949e-06,
+      "loss": 1.6043,
+      "step": 7717
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267341868908848e-06,
+      "loss": 1.4387,
+      "step": 7718
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267246321421747e-06,
+      "loss": 1.2902,
+      "step": 7719
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267150773934645e-06,
+      "loss": 1.3857,
+      "step": 7720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2670552264475445e-06,
+      "loss": 1.3346,
+      "step": 7721
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.266959678960444e-06,
+      "loss": 1.0878,
+      "step": 7722
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2668641314733425e-06,
+      "loss": 0.9776,
+      "step": 7723
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2667685839862416e-06,
+      "loss": 1.0629,
+      "step": 7724
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.266673036499141e-06,
+      "loss": 1.1508,
+      "step": 7725
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.26657748901204e-06,
+      "loss": 1.2239,
+      "step": 7726
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266481941524938e-06,
+      "loss": 1.2254,
+      "step": 7727
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266386394037837e-06,
+      "loss": 0.8084,
+      "step": 7728
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266290846550736e-06,
+      "loss": 1.4827,
+      "step": 7729
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266195299063635e-06,
+      "loss": 1.4617,
+      "step": 7730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266099751576534e-06,
+      "loss": 1.7433,
+      "step": 7731
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.266004204089433e-06,
+      "loss": 1.505,
+      "step": 7732
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265908656602332e-06,
+      "loss": 0.8828,
+      "step": 7733
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26581310911523e-06,
+      "loss": 0.7603,
+      "step": 7734
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265717561628129e-06,
+      "loss": 0.9067,
+      "step": 7735
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265622014141028e-06,
+      "loss": 0.8696,
+      "step": 7736
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265526466653927e-06,
+      "loss": 1.3041,
+      "step": 7737
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265430919166826e-06,
+      "loss": 0.8657,
+      "step": 7738
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2653353716797255e-06,
+      "loss": 1.4509,
+      "step": 7739
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265239824192624e-06,
+      "loss": 1.1223,
+      "step": 7740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265144276705523e-06,
+      "loss": 0.9603,
+      "step": 7741
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.265048729218422e-06,
+      "loss": 1.4948,
+      "step": 7742
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264953181731321e-06,
+      "loss": 1.5419,
+      "step": 7743
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26485763424422e-06,
+      "loss": 1.1483,
+      "step": 7744
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264762086757119e-06,
+      "loss": 1.2058,
+      "step": 7745
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264666539270018e-06,
+      "loss": 1.5026,
+      "step": 7746
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264570991782916e-06,
+      "loss": 1.0195,
+      "step": 7747
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264475444295815e-06,
+      "loss": 1.4285,
+      "step": 7748
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264379896808714e-06,
+      "loss": 0.9532,
+      "step": 7749
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264284349321613e-06,
+      "loss": 1.0278,
+      "step": 7750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264188801834512e-06,
+      "loss": 0.9435,
+      "step": 7751
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.264093254347411e-06,
+      "loss": 1.1755,
+      "step": 7752
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.26399770686031e-06,
+      "loss": 1.3675,
+      "step": 7753
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2639021593732085e-06,
+      "loss": 1.059,
+      "step": 7754
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2638066118861075e-06,
+      "loss": 1.4247,
+      "step": 7755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2637110643990066e-06,
+      "loss": 1.2872,
+      "step": 7756
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263615516911906e-06,
+      "loss": 1.2488,
+      "step": 7757
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263519969424805e-06,
+      "loss": 1.4131,
+      "step": 7758
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263424421937704e-06,
+      "loss": 0.7822,
+      "step": 7759
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263328874450603e-06,
+      "loss": 1.4557,
+      "step": 7760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263233326963501e-06,
+      "loss": 0.886,
+      "step": 7761
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2631377794764e-06,
+      "loss": 1.3843,
+      "step": 7762
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.263042231989299e-06,
+      "loss": 1.3154,
+      "step": 7763
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262946684502198e-06,
+      "loss": 1.0437,
+      "step": 7764
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262851137015097e-06,
+      "loss": 1.1232,
+      "step": 7765
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262755589527996e-06,
+      "loss": 1.8095,
+      "step": 7766
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262660042040895e-06,
+      "loss": 0.9228,
+      "step": 7767
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262564494553793e-06,
+      "loss": 1.0709,
+      "step": 7768
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262468947066692e-06,
+      "loss": 1.7711,
+      "step": 7769
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262373399579591e-06,
+      "loss": 1.3174,
+      "step": 7770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2622778520924905e-06,
+      "loss": 0.8781,
+      "step": 7771
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262182304605389e-06,
+      "loss": 1.1481,
+      "step": 7772
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2620867571182885e-06,
+      "loss": 1.1668,
+      "step": 7773
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2619912096311876e-06,
+      "loss": 0.8839,
+      "step": 7774
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261895662144086e-06,
+      "loss": 1.822,
+      "step": 7775
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261800114656985e-06,
+      "loss": 1.1258,
+      "step": 7776
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261704567169884e-06,
+      "loss": 1.162,
+      "step": 7777
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.261609019682783e-06,
+      "loss": 1.2513,
+      "step": 7778
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.261513472195681e-06,
+      "loss": 0.9621,
+      "step": 7779
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.261417924708581e-06,
+      "loss": 1.2129,
+      "step": 7780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26132237722148e-06,
+      "loss": 1.068,
+      "step": 7781
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.261226829734378e-06,
+      "loss": 1.5885,
+      "step": 7782
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.261131282247277e-06,
+      "loss": 1.0341,
+      "step": 7783
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.261035734760176e-06,
+      "loss": 1.3871,
+      "step": 7784
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260940187273075e-06,
+      "loss": 1.1788,
+      "step": 7785
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2608446397859735e-06,
+      "loss": 1.1982,
+      "step": 7786
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260749092298873e-06,
+      "loss": 1.2226,
+      "step": 7787
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2606535448117716e-06,
+      "loss": 1.0812,
+      "step": 7788
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260557997324671e-06,
+      "loss": 0.9977,
+      "step": 7789
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.26046244983757e-06,
+      "loss": 1.5271,
+      "step": 7790
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260366902350469e-06,
+      "loss": 1.2592,
+      "step": 7791
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260271354863367e-06,
+      "loss": 0.7813,
+      "step": 7792
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260175807376266e-06,
+      "loss": 0.6846,
+      "step": 7793
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.260080259889166e-06,
+      "loss": 1.0184,
+      "step": 7794
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259984712402064e-06,
+      "loss": 1.0948,
+      "step": 7795
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259889164914963e-06,
+      "loss": 1.0896,
+      "step": 7796
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259793617427862e-06,
+      "loss": 1.0247,
+      "step": 7797
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259698069940761e-06,
+      "loss": 1.1422,
+      "step": 7798
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259602522453659e-06,
+      "loss": 1.2261,
+      "step": 7799
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259506974966558e-06,
+      "loss": 0.673,
+      "step": 7800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259411427479458e-06,
+      "loss": 1.3505,
+      "step": 7801
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.259315879992356e-06,
+      "loss": 1.2513,
+      "step": 7802
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2592203325052554e-06,
+      "loss": 1.7763,
+      "step": 7803
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2591247850181545e-06,
+      "loss": 1.1743,
+      "step": 7804
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2590292375310535e-06,
+      "loss": 1.372,
+      "step": 7805
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258933690043952e-06,
+      "loss": 1.1034,
+      "step": 7806
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258838142556851e-06,
+      "loss": 1.1321,
+      "step": 7807
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.25874259506975e-06,
+      "loss": 1.1168,
+      "step": 7808
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258647047582649e-06,
+      "loss": 0.9813,
+      "step": 7809
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258551500095548e-06,
+      "loss": 1.319,
+      "step": 7810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258455952608447e-06,
+      "loss": 0.9501,
+      "step": 7811
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258360405121346e-06,
+      "loss": 1.3183,
+      "step": 7812
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258264857634244e-06,
+      "loss": 1.2886,
+      "step": 7813
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258169310147143e-06,
+      "loss": 1.4428,
+      "step": 7814
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.258073762660042e-06,
+      "loss": 1.4008,
+      "step": 7815
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257978215172941e-06,
+      "loss": 1.1289,
+      "step": 7816
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.25788266768584e-06,
+      "loss": 1.2344,
+      "step": 7817
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257787120198739e-06,
+      "loss": 1.1533,
+      "step": 7818
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257691572711638e-06,
+      "loss": 0.8857,
+      "step": 7819
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2575960252245366e-06,
+      "loss": 0.7118,
+      "step": 7820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257500477737436e-06,
+      "loss": 1.1835,
+      "step": 7821
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257404930250335e-06,
+      "loss": 1.0691,
+      "step": 7822
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257309382763234e-06,
+      "loss": 1.1316,
+      "step": 7823
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257213835276133e-06,
+      "loss": 1.4285,
+      "step": 7824
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257118287789032e-06,
+      "loss": 1.4751,
+      "step": 7825
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257022740301931e-06,
+      "loss": 1.3704,
+      "step": 7826
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256927192814829e-06,
+      "loss": 1.0236,
+      "step": 7827
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256831645327728e-06,
+      "loss": 1.1355,
+      "step": 7828
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256736097840627e-06,
+      "loss": 0.5102,
+      "step": 7829
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256640550353526e-06,
+      "loss": 1.1847,
+      "step": 7830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.256545002866425e-06,
+      "loss": 1.1389,
+      "step": 7831
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.256449455379324e-06,
+      "loss": 1.5056,
+      "step": 7832
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.256353907892223e-06,
+      "loss": 1.5741,
+      "step": 7833
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.256258360405121e-06,
+      "loss": 1.3478,
+      "step": 7834
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2561628129180204e-06,
+      "loss": 0.9575,
+      "step": 7835
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2560672654309195e-06,
+      "loss": 1.0589,
+      "step": 7836
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2559717179438185e-06,
+      "loss": 1.2023,
+      "step": 7837
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2558761704567176e-06,
+      "loss": 1.6806,
+      "step": 7838
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255780622969617e-06,
+      "loss": 1.373,
+      "step": 7839
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255685075482515e-06,
+      "loss": 1.5629,
+      "step": 7840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255589527995414e-06,
+      "loss": 0.9836,
+      "step": 7841
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255493980508313e-06,
+      "loss": 0.8985,
+      "step": 7842
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255398433021212e-06,
+      "loss": 1.1763,
+      "step": 7843
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255302885534111e-06,
+      "loss": 0.8833,
+      "step": 7844
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25520733804701e-06,
+      "loss": 1.3451,
+      "step": 7845
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255111790559909e-06,
+      "loss": 1.0119,
+      "step": 7846
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.255016243072807e-06,
+      "loss": 1.0278,
+      "step": 7847
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254920695585706e-06,
+      "loss": 1.3818,
+      "step": 7848
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254825148098605e-06,
+      "loss": 1.12,
+      "step": 7849
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254729600611504e-06,
+      "loss": 1.045,
+      "step": 7850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2546340531244025e-06,
+      "loss": 1.2649,
+      "step": 7851
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254538505637302e-06,
+      "loss": 1.2008,
+      "step": 7852
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2544429581502014e-06,
+      "loss": 1.319,
+      "step": 7853
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2543474106631e-06,
+      "loss": 0.9427,
+      "step": 7854
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254251863175999e-06,
+      "loss": 0.9148,
+      "step": 7855
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254156315688898e-06,
+      "loss": 1.3005,
+      "step": 7856
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.254060768201797e-06,
+      "loss": 1.0501,
+      "step": 7857
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253965220714695e-06,
+      "loss": 1.3158,
+      "step": 7858
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253869673227595e-06,
+      "loss": 1.4762,
+      "step": 7859
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253774125740494e-06,
+      "loss": 1.2725,
+      "step": 7860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253678578253392e-06,
+      "loss": 1.3793,
+      "step": 7861
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253583030766291e-06,
+      "loss": 1.6799,
+      "step": 7862
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25348748327919e-06,
+      "loss": 0.6501,
+      "step": 7863
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253391935792089e-06,
+      "loss": 1.6832,
+      "step": 7864
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253296388304987e-06,
+      "loss": 1.2725,
+      "step": 7865
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253200840817887e-06,
+      "loss": 1.4601,
+      "step": 7866
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.253105293330786e-06,
+      "loss": 1.3653,
+      "step": 7867
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2530097458436845e-06,
+      "loss": 0.9044,
+      "step": 7868
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2529141983565835e-06,
+      "loss": 1.569,
+      "step": 7869
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2528186508694826e-06,
+      "loss": 1.2764,
+      "step": 7870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252723103382382e-06,
+      "loss": 1.2215,
+      "step": 7871
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25262755589528e-06,
+      "loss": 0.9679,
+      "step": 7872
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25253200840818e-06,
+      "loss": 1.1552,
+      "step": 7873
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252436460921079e-06,
+      "loss": 0.9847,
+      "step": 7874
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252340913433977e-06,
+      "loss": 1.3396,
+      "step": 7875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252245365946876e-06,
+      "loss": 0.9115,
+      "step": 7876
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252149818459775e-06,
+      "loss": 1.2621,
+      "step": 7877
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252054270972674e-06,
+      "loss": 0.9172,
+      "step": 7878
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251958723485572e-06,
+      "loss": 1.3083,
+      "step": 7879
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251863175998471e-06,
+      "loss": 1.2289,
+      "step": 7880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.25176762851137e-06,
+      "loss": 0.7595,
+      "step": 7881
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251672081024269e-06,
+      "loss": 1.1709,
+      "step": 7882
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251576533537168e-06,
+      "loss": 0.8704,
+      "step": 7883
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.251480986050067e-06,
+      "loss": 1.0053,
+      "step": 7884
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2513854385629664e-06,
+      "loss": 1.1919,
+      "step": 7885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.251289891075865e-06,
+      "loss": 1.2863,
+      "step": 7886
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.251194343588764e-06,
+      "loss": 1.3527,
+      "step": 7887
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.251098796101663e-06,
+      "loss": 0.906,
+      "step": 7888
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.251003248614562e-06,
+      "loss": 0.9609,
+      "step": 7889
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250907701127461e-06,
+      "loss": 1.1317,
+      "step": 7890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.25081215364036e-06,
+      "loss": 1.0972,
+      "step": 7891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250716606153258e-06,
+      "loss": 0.8036,
+      "step": 7892
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250621058666157e-06,
+      "loss": 0.9651,
+      "step": 7893
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250525511179056e-06,
+      "loss": 0.9318,
+      "step": 7894
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250429963691955e-06,
+      "loss": 1.0031,
+      "step": 7895
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250334416204854e-06,
+      "loss": 0.889,
+      "step": 7896
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250238868717753e-06,
+      "loss": 1.0971,
+      "step": 7897
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.250143321230652e-06,
+      "loss": 0.8674,
+      "step": 7898
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2500477737435504e-06,
+      "loss": 1.0664,
+      "step": 7899
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2499522262564495e-06,
+      "loss": 1.3255,
+      "step": 7900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2498566787693485e-06,
+      "loss": 1.1747,
+      "step": 7901
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2497611312822476e-06,
+      "loss": 0.9582,
+      "step": 7902
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249665583795147e-06,
+      "loss": 1.2537,
+      "step": 7903
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249570036308046e-06,
+      "loss": 1.2108,
+      "step": 7904
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249474488820945e-06,
+      "loss": 1.2393,
+      "step": 7905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249378941333843e-06,
+      "loss": 1.0705,
+      "step": 7906
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249283393846742e-06,
+      "loss": 0.8222,
+      "step": 7907
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.249187846359641e-06,
+      "loss": 1.5212,
+      "step": 7908
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.24909229887254e-06,
+      "loss": 1.2976,
+      "step": 7909
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248996751385439e-06,
+      "loss": 1.0304,
+      "step": 7910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248901203898338e-06,
+      "loss": 1.3288,
+      "step": 7911
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248805656411237e-06,
+      "loss": 1.276,
+      "step": 7912
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248710108924135e-06,
+      "loss": 1.0169,
+      "step": 7913
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248614561437034e-06,
+      "loss": 0.8825,
+      "step": 7914
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248519013949933e-06,
+      "loss": 1.1718,
+      "step": 7915
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248423466462832e-06,
+      "loss": 0.8454,
+      "step": 7916
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2483279189757314e-06,
+      "loss": 1.3768,
+      "step": 7917
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2482323714886305e-06,
+      "loss": 0.9452,
+      "step": 7918
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2481368240015295e-06,
+      "loss": 0.9759,
+      "step": 7919
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.248041276514428e-06,
+      "loss": 1.3093,
+      "step": 7920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247945729027327e-06,
+      "loss": 1.2546,
+      "step": 7921
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247850181540226e-06,
+      "loss": 0.9225,
+      "step": 7922
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247754634053125e-06,
+      "loss": 1.0712,
+      "step": 7923
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247659086566024e-06,
+      "loss": 0.9508,
+      "step": 7924
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247563539078923e-06,
+      "loss": 1.2625,
+      "step": 7925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247467991591822e-06,
+      "loss": 0.8633,
+      "step": 7926
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.24737244410472e-06,
+      "loss": 1.0734,
+      "step": 7927
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247276896617619e-06,
+      "loss": 1.0665,
+      "step": 7928
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247181349130518e-06,
+      "loss": 0.7944,
+      "step": 7929
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247085801643417e-06,
+      "loss": 1.0762,
+      "step": 7930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246990254156316e-06,
+      "loss": 1.0293,
+      "step": 7931
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246894706669215e-06,
+      "loss": 1.1067,
+      "step": 7932
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2467991591821135e-06,
+      "loss": 0.8695,
+      "step": 7933
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2467036116950126e-06,
+      "loss": 1.0444,
+      "step": 7934
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246608064207912e-06,
+      "loss": 1.1323,
+      "step": 7935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246512516720811e-06,
+      "loss": 1.0831,
+      "step": 7936
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24641696923371e-06,
+      "loss": 1.2138,
+      "step": 7937
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246321421746609e-06,
+      "loss": 1.9837,
+      "step": 7938
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246225874259508e-06,
+      "loss": 1.1715,
+      "step": 7939
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246130326772406e-06,
+      "loss": 0.6565,
+      "step": 7940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.246034779285305e-06,
+      "loss": 1.2503,
+      "step": 7941
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245939231798204e-06,
+      "loss": 0.9137,
+      "step": 7942
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245843684311103e-06,
+      "loss": 0.9844,
+      "step": 7943
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245748136824001e-06,
+      "loss": 0.8089,
+      "step": 7944
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245652589336901e-06,
+      "loss": 0.9921,
+      "step": 7945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2455570418498e-06,
+      "loss": 0.7927,
+      "step": 7946
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245461494362698e-06,
+      "loss": 1.1653,
+      "step": 7947
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245365946875597e-06,
+      "loss": 1.1674,
+      "step": 7948
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2452703993884964e-06,
+      "loss": 1.3763,
+      "step": 7949
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2451748519013955e-06,
+      "loss": 0.8466,
+      "step": 7950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.245079304414294e-06,
+      "loss": 1.3504,
+      "step": 7951
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244983756927193e-06,
+      "loss": 1.2461,
+      "step": 7952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244888209440093e-06,
+      "loss": 1.1959,
+      "step": 7953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244792661952991e-06,
+      "loss": 1.2745,
+      "step": 7954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24469711446589e-06,
+      "loss": 0.8445,
+      "step": 7955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244601566978789e-06,
+      "loss": 0.94,
+      "step": 7956
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244506019491688e-06,
+      "loss": 1.1974,
+      "step": 7957
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244410472004586e-06,
+      "loss": 1.2136,
+      "step": 7958
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244314924517485e-06,
+      "loss": 1.1917,
+      "step": 7959
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244219377030385e-06,
+      "loss": 1.0525,
+      "step": 7960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244123829543283e-06,
+      "loss": 1.4612,
+      "step": 7961
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244028282056182e-06,
+      "loss": 1.0211,
+      "step": 7962
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243932734569081e-06,
+      "loss": 1.1465,
+      "step": 7963
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24383718708198e-06,
+      "loss": 1.1167,
+      "step": 7964
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2437416395948785e-06,
+      "loss": 1.1505,
+      "step": 7965
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2436460921077776e-06,
+      "loss": 0.9794,
+      "step": 7966
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2435505446206774e-06,
+      "loss": 1.2934,
+      "step": 7967
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243454997133576e-06,
+      "loss": 1.6496,
+      "step": 7968
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243359449646475e-06,
+      "loss": 1.0186,
+      "step": 7969
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243263902159374e-06,
+      "loss": 1.2546,
+      "step": 7970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243168354672273e-06,
+      "loss": 1.0952,
+      "step": 7971
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.243072807185171e-06,
+      "loss": 0.8327,
+      "step": 7972
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24297725969807e-06,
+      "loss": 0.8737,
+      "step": 7973
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.24288171221097e-06,
+      "loss": 0.8828,
+      "step": 7974
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242786164723868e-06,
+      "loss": 0.9655,
+      "step": 7975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242690617236767e-06,
+      "loss": 1.0841,
+      "step": 7976
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242595069749666e-06,
+      "loss": 1.0088,
+      "step": 7977
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242499522262565e-06,
+      "loss": 1.3236,
+      "step": 7978
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242403974775463e-06,
+      "loss": 0.9577,
+      "step": 7979
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242308427288362e-06,
+      "loss": 1.0442,
+      "step": 7980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2422128798012614e-06,
+      "loss": 0.8324,
+      "step": 7981
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2421173323141605e-06,
+      "loss": 0.8174,
+      "step": 7982
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2420217848270595e-06,
+      "loss": 1.0977,
+      "step": 7983
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2419262373399586e-06,
+      "loss": 1.152,
+      "step": 7984
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241830689852857e-06,
+      "loss": 1.3974,
+      "step": 7985
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241735142365756e-06,
+      "loss": 1.1967,
+      "step": 7986
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241639594878655e-06,
+      "loss": 1.3798,
+      "step": 7987
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241544047391554e-06,
+      "loss": 1.6801,
+      "step": 7988
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241448499904453e-06,
+      "loss": 1.0918,
+      "step": 7989
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241352952417352e-06,
+      "loss": 1.1966,
+      "step": 7990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241257404930251e-06,
+      "loss": 1.1238,
+      "step": 7991
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241161857443149e-06,
+      "loss": 1.0575,
+      "step": 7992
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.241066309956048e-06,
+      "loss": 1.5528,
+      "step": 7993
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240970762468947e-06,
+      "loss": 1.3968,
+      "step": 7994
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240875214981846e-06,
+      "loss": 0.7734,
+      "step": 7995
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240779667494745e-06,
+      "loss": 1.0717,
+      "step": 7996
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240684120007644e-06,
+      "loss": 1.2986,
+      "step": 7997
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240588572520543e-06,
+      "loss": 1.1728,
+      "step": 7998
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240493025033442e-06,
+      "loss": 0.9052,
+      "step": 7999
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240397477546341e-06,
+      "loss": 0.9919,
+      "step": 8000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.24030193005924e-06,
+      "loss": 1.0771,
+      "step": 8001
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240206382572139e-06,
+      "loss": 0.8564,
+      "step": 8002
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240110835085038e-06,
+      "loss": 0.8708,
+      "step": 8003
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.240015287597937e-06,
+      "loss": 1.1306,
+      "step": 8004
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239919740110836e-06,
+      "loss": 1.0733,
+      "step": 8005
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239824192623734e-06,
+      "loss": 1.2522,
+      "step": 8006
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239728645136633e-06,
+      "loss": 0.6976,
+      "step": 8007
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239633097649532e-06,
+      "loss": 1.6069,
+      "step": 8008
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239537550162431e-06,
+      "loss": 0.7986,
+      "step": 8009
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.23944200267533e-06,
+      "loss": 1.1385,
+      "step": 8010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239346455188229e-06,
+      "loss": 1.0919,
+      "step": 8011
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.239250907701128e-06,
+      "loss": 1.3266,
+      "step": 8012
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2391553602140264e-06,
+      "loss": 0.8582,
+      "step": 8013
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2390598127269255e-06,
+      "loss": 0.9124,
+      "step": 8014
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2389642652398245e-06,
+      "loss": 0.9233,
+      "step": 8015
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2388687177527236e-06,
+      "loss": 1.5131,
+      "step": 8016
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238773170265623e-06,
+      "loss": 0.9041,
+      "step": 8017
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238677622778522e-06,
+      "loss": 1.0285,
+      "step": 8018
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238582075291421e-06,
+      "loss": 1.3461,
+      "step": 8019
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238486527804319e-06,
+      "loss": 0.8709,
+      "step": 8020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238390980317218e-06,
+      "loss": 1.4308,
+      "step": 8021
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238295432830117e-06,
+      "loss": 1.1707,
+      "step": 8022
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238199885343016e-06,
+      "loss": 1.2873,
+      "step": 8023
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238104337855914e-06,
+      "loss": 1.4763,
+      "step": 8024
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.238008790368814e-06,
+      "loss": 1.0336,
+      "step": 8025
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237913242881713e-06,
+      "loss": 1.168,
+      "step": 8026
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237817695394611e-06,
+      "loss": 1.1229,
+      "step": 8027
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.23772214790751e-06,
+      "loss": 1.3318,
+      "step": 8028
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237626600420409e-06,
+      "loss": 1.1629,
+      "step": 8029
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237531052933308e-06,
+      "loss": 0.9719,
+      "step": 8030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237435505446207e-06,
+      "loss": 1.1095,
+      "step": 8031
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2373399579591065e-06,
+      "loss": 1.0289,
+      "step": 8032
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237244410472005e-06,
+      "loss": 1.0202,
+      "step": 8033
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237148862984904e-06,
+      "loss": 1.0866,
+      "step": 8034
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.237053315497803e-06,
+      "loss": 0.9195,
+      "step": 8035
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236957768010702e-06,
+      "loss": 1.2274,
+      "step": 8036
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236862220523601e-06,
+      "loss": 1.7076,
+      "step": 8037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236766673036499e-06,
+      "loss": 1.0656,
+      "step": 8038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236671125549399e-06,
+      "loss": 1.2851,
+      "step": 8039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236575578062297e-06,
+      "loss": 1.0142,
+      "step": 8040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.236480030575196e-06,
+      "loss": 1.3837,
+      "step": 8041
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.236384483088095e-06,
+      "loss": 1.0282,
+      "step": 8042
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.236288935600994e-06,
+      "loss": 1.6734,
+      "step": 8043
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.236193388113892e-06,
+      "loss": 1.122,
+      "step": 8044
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2360978406267914e-06,
+      "loss": 1.8456,
+      "step": 8045
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.236002293139691e-06,
+      "loss": 0.7909,
+      "step": 8046
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2359067456525895e-06,
+      "loss": 1.0947,
+      "step": 8047
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2358111981654886e-06,
+      "loss": 1.2454,
+      "step": 8048
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235715650678388e-06,
+      "loss": 1.1594,
+      "step": 8049
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235620103191287e-06,
+      "loss": 0.7748,
+      "step": 8050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235524555704185e-06,
+      "loss": 1.0641,
+      "step": 8051
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235429008217084e-06,
+      "loss": 1.4541,
+      "step": 8052
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235333460729984e-06,
+      "loss": 0.8343,
+      "step": 8053
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235237913242882e-06,
+      "loss": 0.9781,
+      "step": 8054
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.235142365755781e-06,
+      "loss": 1.1882,
+      "step": 8055
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23504681826868e-06,
+      "loss": 1.1779,
+      "step": 8056
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234951270781579e-06,
+      "loss": 1.1361,
+      "step": 8057
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234855723294477e-06,
+      "loss": 0.7896,
+      "step": 8058
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234760175807376e-06,
+      "loss": 1.4903,
+      "step": 8059
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234664628320275e-06,
+      "loss": 1.2406,
+      "step": 8060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234569080833174e-06,
+      "loss": 0.7009,
+      "step": 8061
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234473533346073e-06,
+      "loss": 0.9167,
+      "step": 8062
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2343779858589724e-06,
+      "loss": 1.074,
+      "step": 8063
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2342824383718715e-06,
+      "loss": 1.2625,
+      "step": 8064
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23418689088477e-06,
+      "loss": 1.7261,
+      "step": 8065
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.234091343397669e-06,
+      "loss": 1.3235,
+      "step": 8066
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233995795910568e-06,
+      "loss": 0.9198,
+      "step": 8067
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233900248423467e-06,
+      "loss": 1.232,
+      "step": 8068
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233804700936366e-06,
+      "loss": 1.4169,
+      "step": 8069
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233709153449265e-06,
+      "loss": 1.003,
+      "step": 8070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233613605962164e-06,
+      "loss": 1.648,
+      "step": 8071
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233518058475062e-06,
+      "loss": 1.014,
+      "step": 8072
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233422510987961e-06,
+      "loss": 1.0012,
+      "step": 8073
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23332696350086e-06,
+      "loss": 1.4824,
+      "step": 8074
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233231416013759e-06,
+      "loss": 1.2071,
+      "step": 8075
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233135868526658e-06,
+      "loss": 1.1925,
+      "step": 8076
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.233040321039557e-06,
+      "loss": 1.052,
+      "step": 8077
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232944773552456e-06,
+      "loss": 1.0893,
+      "step": 8078
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2328492260653545e-06,
+      "loss": 1.0327,
+      "step": 8079
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2327536785782536e-06,
+      "loss": 1.6089,
+      "step": 8080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232658131091153e-06,
+      "loss": 1.3408,
+      "step": 8081
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232562583604052e-06,
+      "loss": 1.2017,
+      "step": 8082
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232467036116951e-06,
+      "loss": 0.8378,
+      "step": 8083
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23237148862985e-06,
+      "loss": 1.3533,
+      "step": 8084
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232275941142748e-06,
+      "loss": 1.2111,
+      "step": 8085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232180393655647e-06,
+      "loss": 1.285,
+      "step": 8086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232084846168546e-06,
+      "loss": 0.8454,
+      "step": 8087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231989298681445e-06,
+      "loss": 0.9331,
+      "step": 8088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231893751194344e-06,
+      "loss": 1.1963,
+      "step": 8089
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231798203707243e-06,
+      "loss": 0.8225,
+      "step": 8090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231702656220142e-06,
+      "loss": 1.1006,
+      "step": 8091
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.23160710873304e-06,
+      "loss": 0.9158,
+      "step": 8092
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.231511561245939e-06,
+      "loss": 1.3749,
+      "step": 8093
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.231416013758838e-06,
+      "loss": 0.9664,
+      "step": 8094
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2313204662717374e-06,
+      "loss": 0.7983,
+      "step": 8095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.231224918784636e-06,
+      "loss": 1.0172,
+      "step": 8096
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2311293712975355e-06,
+      "loss": 0.6098,
+      "step": 8097
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2310338238104346e-06,
+      "loss": 1.2707,
+      "step": 8098
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230938276323333e-06,
+      "loss": 1.0306,
+      "step": 8099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230842728836232e-06,
+      "loss": 0.9592,
+      "step": 8100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230747181349131e-06,
+      "loss": 1.5183,
+      "step": 8101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.23065163386203e-06,
+      "loss": 1.1323,
+      "step": 8102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230556086374928e-06,
+      "loss": 1.1087,
+      "step": 8103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230460538887828e-06,
+      "loss": 1.0396,
+      "step": 8104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230364991400727e-06,
+      "loss": 1.1161,
+      "step": 8105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230269443913625e-06,
+      "loss": 1.2537,
+      "step": 8106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230173896426524e-06,
+      "loss": 1.1728,
+      "step": 8107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.230078348939423e-06,
+      "loss": 1.1682,
+      "step": 8108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229982801452322e-06,
+      "loss": 1.1635,
+      "step": 8109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2298872539652205e-06,
+      "loss": 1.1413,
+      "step": 8110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.22979170647812e-06,
+      "loss": 1.0891,
+      "step": 8111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229696158991019e-06,
+      "loss": 0.5951,
+      "step": 8112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229600611503918e-06,
+      "loss": 0.8259,
+      "step": 8113
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229505064016817e-06,
+      "loss": 1.1737,
+      "step": 8114
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229409516529716e-06,
+      "loss": 0.993,
+      "step": 8115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229313969042615e-06,
+      "loss": 1.165,
+      "step": 8116
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229218421555513e-06,
+      "loss": 1.303,
+      "step": 8117
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229122874068413e-06,
+      "loss": 1.1585,
+      "step": 8118
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.229027326581312e-06,
+      "loss": 0.7966,
+      "step": 8119
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.22893177909421e-06,
+      "loss": 1.0218,
+      "step": 8120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228836231607109e-06,
+      "loss": 1.2048,
+      "step": 8121
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228740684120008e-06,
+      "loss": 1.1698,
+      "step": 8122
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228645136632907e-06,
+      "loss": 0.4387,
+      "step": 8123
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228549589145805e-06,
+      "loss": 1.4004,
+      "step": 8124
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228454041658705e-06,
+      "loss": 0.9774,
+      "step": 8125
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.228358494171603e-06,
+      "loss": 1.2091,
+      "step": 8126
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2282629466845024e-06,
+      "loss": 0.7417,
+      "step": 8127
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2281673991974015e-06,
+      "loss": 0.9931,
+      "step": 8128
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2280718517103005e-06,
+      "loss": 0.9359,
+      "step": 8129
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2279763042231996e-06,
+      "loss": 1.8395,
+      "step": 8130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227880756736098e-06,
+      "loss": 1.0468,
+      "step": 8131
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227785209248997e-06,
+      "loss": 1.0945,
+      "step": 8132
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227689661761896e-06,
+      "loss": 0.7879,
+      "step": 8133
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227594114274795e-06,
+      "loss": 1.2046,
+      "step": 8134
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227498566787694e-06,
+      "loss": 1.3866,
+      "step": 8135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227403019300593e-06,
+      "loss": 1.5373,
+      "step": 8136
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227307471813491e-06,
+      "loss": 1.1977,
+      "step": 8137
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.22721192432639e-06,
+      "loss": 1.3964,
+      "step": 8138
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227116376839289e-06,
+      "loss": 1.4223,
+      "step": 8139
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.227020829352188e-06,
+      "loss": 1.749,
+      "step": 8140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226925281865087e-06,
+      "loss": 1.5804,
+      "step": 8141
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226829734377986e-06,
+      "loss": 0.9734,
+      "step": 8142
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226734186890885e-06,
+      "loss": 1.0578,
+      "step": 8143
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2266386394037836e-06,
+      "loss": 1.1483,
+      "step": 8144
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226543091916683e-06,
+      "loss": 1.6682,
+      "step": 8145
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.226447544429582e-06,
+      "loss": 1.0232,
+      "step": 8146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.226351996942481e-06,
+      "loss": 1.2993,
+      "step": 8147
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22625644945538e-06,
+      "loss": 1.3434,
+      "step": 8148
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.226160901968279e-06,
+      "loss": 1.4505,
+      "step": 8149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.226065354481178e-06,
+      "loss": 1.1721,
+      "step": 8150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225969806994076e-06,
+      "loss": 1.3271,
+      "step": 8151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225874259506975e-06,
+      "loss": 1.204,
+      "step": 8152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225778712019874e-06,
+      "loss": 0.8832,
+      "step": 8153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225683164532773e-06,
+      "loss": 1.5123,
+      "step": 8154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225587617045672e-06,
+      "loss": 1.1733,
+      "step": 8155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225492069558571e-06,
+      "loss": 1.2458,
+      "step": 8156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22539652207147e-06,
+      "loss": 1.3924,
+      "step": 8157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.225300974584368e-06,
+      "loss": 1.0524,
+      "step": 8158
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2252054270972674e-06,
+      "loss": 0.7797,
+      "step": 8159
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2251098796101665e-06,
+      "loss": 1.1329,
+      "step": 8160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2250143321230655e-06,
+      "loss": 1.0728,
+      "step": 8161
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2249187846359646e-06,
+      "loss": 1.4347,
+      "step": 8162
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224823237148864e-06,
+      "loss": 1.0023,
+      "step": 8163
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224727689661763e-06,
+      "loss": 1.1074,
+      "step": 8164
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224632142174661e-06,
+      "loss": 1.2074,
+      "step": 8165
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22453659468756e-06,
+      "loss": 1.5643,
+      "step": 8166
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224441047200459e-06,
+      "loss": 1.3617,
+      "step": 8167
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224345499713358e-06,
+      "loss": 1.2062,
+      "step": 8168
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224249952226257e-06,
+      "loss": 1.434,
+      "step": 8169
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224154404739156e-06,
+      "loss": 0.9456,
+      "step": 8170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.224058857252055e-06,
+      "loss": 0.783,
+      "step": 8171
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223963309764953e-06,
+      "loss": 1.8238,
+      "step": 8172
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223867762277852e-06,
+      "loss": 1.5718,
+      "step": 8173
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223772214790751e-06,
+      "loss": 1.147,
+      "step": 8174
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22367666730365e-06,
+      "loss": 0.8437,
+      "step": 8175
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223581119816549e-06,
+      "loss": 1.2126,
+      "step": 8176
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2234855723294484e-06,
+      "loss": 1.3416,
+      "step": 8177
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2233900248423475e-06,
+      "loss": 0.8527,
+      "step": 8178
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223294477355246e-06,
+      "loss": 0.6767,
+      "step": 8179
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223198929868145e-06,
+      "loss": 0.9641,
+      "step": 8180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223103382381044e-06,
+      "loss": 1.6314,
+      "step": 8181
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.223007834893943e-06,
+      "loss": 1.2212,
+      "step": 8182
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222912287406842e-06,
+      "loss": 1.4877,
+      "step": 8183
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222816739919741e-06,
+      "loss": 1.0589,
+      "step": 8184
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222721192432639e-06,
+      "loss": 0.6759,
+      "step": 8185
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222625644945538e-06,
+      "loss": 1.4092,
+      "step": 8186
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222530097458437e-06,
+      "loss": 1.1081,
+      "step": 8187
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222434549971336e-06,
+      "loss": 1.1074,
+      "step": 8188
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222339002484234e-06,
+      "loss": 1.0153,
+      "step": 8189
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222243454997134e-06,
+      "loss": 1.309,
+      "step": 8190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.222147907510033e-06,
+      "loss": 1.0984,
+      "step": 8191
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2220523600229315e-06,
+      "loss": 1.2253,
+      "step": 8192
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2219568125358305e-06,
+      "loss": 1.1132,
+      "step": 8193
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2218612650487296e-06,
+      "loss": 1.3169,
+      "step": 8194
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.221765717561629e-06,
+      "loss": 1.5612,
+      "step": 8195
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.221670170074527e-06,
+      "loss": 1.2648,
+      "step": 8196
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.221574622587427e-06,
+      "loss": 1.4425,
+      "step": 8197
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221479075100326e-06,
+      "loss": 1.3369,
+      "step": 8198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221383527613224e-06,
+      "loss": 1.0451,
+      "step": 8199
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221287980126123e-06,
+      "loss": 0.8511,
+      "step": 8200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221192432639022e-06,
+      "loss": 1.0925,
+      "step": 8201
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221096885151921e-06,
+      "loss": 1.0314,
+      "step": 8202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.221001337664819e-06,
+      "loss": 0.9751,
+      "step": 8203
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220905790177718e-06,
+      "loss": 1.5427,
+      "step": 8204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220810242690618e-06,
+      "loss": 0.9938,
+      "step": 8205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220714695203516e-06,
+      "loss": 0.9808,
+      "step": 8206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220619147716415e-06,
+      "loss": 0.9798,
+      "step": 8207
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220523600229314e-06,
+      "loss": 0.702,
+      "step": 8208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2204280527422134e-06,
+      "loss": 1.3236,
+      "step": 8209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220332505255112e-06,
+      "loss": 1.1041,
+      "step": 8210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220236957768011e-06,
+      "loss": 0.9554,
+      "step": 8211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2201414102809106e-06,
+      "loss": 1.2477,
+      "step": 8212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.220045862793809e-06,
+      "loss": 1.5027,
+      "step": 8213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219950315306708e-06,
+      "loss": 1.079,
+      "step": 8214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219854767819607e-06,
+      "loss": 0.7621,
+      "step": 8215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219759220332506e-06,
+      "loss": 0.9906,
+      "step": 8216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219663672845404e-06,
+      "loss": 1.2004,
+      "step": 8217
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219568125358303e-06,
+      "loss": 1.2163,
+      "step": 8218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219472577871203e-06,
+      "loss": 1.1775,
+      "step": 8219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219377030384101e-06,
+      "loss": 1.1764,
+      "step": 8220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219281482897e-06,
+      "loss": 0.9688,
+      "step": 8221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219185935409899e-06,
+      "loss": 1.1588,
+      "step": 8222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.219090387922798e-06,
+      "loss": 1.0111,
+      "step": 8223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2189948404356965e-06,
+      "loss": 1.4163,
+      "step": 8224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2188992929485955e-06,
+      "loss": 0.9683,
+      "step": 8225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2188037454614946e-06,
+      "loss": 1.1767,
+      "step": 8226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218708197974394e-06,
+      "loss": 1.1362,
+      "step": 8227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218612650487293e-06,
+      "loss": 1.0726,
+      "step": 8228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218517103000192e-06,
+      "loss": 0.9853,
+      "step": 8229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218421555513091e-06,
+      "loss": 1.182,
+      "step": 8230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218326008025989e-06,
+      "loss": 0.7744,
+      "step": 8231
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218230460538888e-06,
+      "loss": 1.3381,
+      "step": 8232
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218134913051787e-06,
+      "loss": 1.2202,
+      "step": 8233
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.218039365564686e-06,
+      "loss": 1.2679,
+      "step": 8234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217943818077585e-06,
+      "loss": 0.8922,
+      "step": 8235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217848270590484e-06,
+      "loss": 1.4116,
+      "step": 8236
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217752723103382e-06,
+      "loss": 1.1825,
+      "step": 8237
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217657175616281e-06,
+      "loss": 1.3735,
+      "step": 8238
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.21756162812918e-06,
+      "loss": 1.2965,
+      "step": 8239
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217466080642079e-06,
+      "loss": 1.1602,
+      "step": 8240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2173705331549784e-06,
+      "loss": 1.2803,
+      "step": 8241
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2172749856678775e-06,
+      "loss": 1.2072,
+      "step": 8242
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2171794381807765e-06,
+      "loss": 1.0047,
+      "step": 8243
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217083890693675e-06,
+      "loss": 1.0632,
+      "step": 8244
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216988343206574e-06,
+      "loss": 1.1117,
+      "step": 8245
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216892795719473e-06,
+      "loss": 1.1123,
+      "step": 8246
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216797248232372e-06,
+      "loss": 1.0323,
+      "step": 8247
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.216701700745271e-06,
+      "loss": 1.0279,
+      "step": 8248
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.21660615325817e-06,
+      "loss": 0.9273,
+      "step": 8249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216510605771069e-06,
+      "loss": 1.1778,
+      "step": 8250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216415058283967e-06,
+      "loss": 1.0183,
+      "step": 8251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216319510796866e-06,
+      "loss": 0.9921,
+      "step": 8252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216223963309765e-06,
+      "loss": 1.1656,
+      "step": 8253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216128415822664e-06,
+      "loss": 1.1169,
+      "step": 8254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.216032868335563e-06,
+      "loss": 1.1433,
+      "step": 8255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215937320848462e-06,
+      "loss": 0.926,
+      "step": 8256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215841773361361e-06,
+      "loss": 1.0922,
+      "step": 8257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2157462258742596e-06,
+      "loss": 0.9004,
+      "step": 8258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215650678387159e-06,
+      "loss": 1.5084,
+      "step": 8259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215555130900058e-06,
+      "loss": 1.1193,
+      "step": 8260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215459583412957e-06,
+      "loss": 0.8425,
+      "step": 8261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215364035925856e-06,
+      "loss": 1.7017,
+      "step": 8262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215268488438755e-06,
+      "loss": 0.9794,
+      "step": 8263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215172940951654e-06,
+      "loss": 1.1997,
+      "step": 8264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.215077393464552e-06,
+      "loss": 1.0018,
+      "step": 8265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214981845977451e-06,
+      "loss": 1.2903,
+      "step": 8266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21488629849035e-06,
+      "loss": 1.7787,
+      "step": 8267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214790751003249e-06,
+      "loss": 1.1474,
+      "step": 8268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214695203516148e-06,
+      "loss": 1.1874,
+      "step": 8269
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214599656029047e-06,
+      "loss": 0.6916,
+      "step": 8270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214504108541946e-06,
+      "loss": 0.8291,
+      "step": 8271
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.214408561054844e-06,
+      "loss": 1.1408,
+      "step": 8272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2143130135677434e-06,
+      "loss": 1.1443,
+      "step": 8273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2142174660806425e-06,
+      "loss": 1.2305,
+      "step": 8274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2141219185935415e-06,
+      "loss": 0.8727,
+      "step": 8275
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21402637110644e-06,
+      "loss": 1.47,
+      "step": 8276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21393082361934e-06,
+      "loss": 1.0838,
+      "step": 8277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213835276132238e-06,
+      "loss": 0.9612,
+      "step": 8278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213739728645137e-06,
+      "loss": 0.7357,
+      "step": 8279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213644181158036e-06,
+      "loss": 0.9207,
+      "step": 8280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213548633670935e-06,
+      "loss": 1.4281,
+      "step": 8281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213453086183834e-06,
+      "loss": 1.1297,
+      "step": 8282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213357538696732e-06,
+      "loss": 0.9159,
+      "step": 8283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213261991209632e-06,
+      "loss": 0.6892,
+      "step": 8284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21316644372253e-06,
+      "loss": 1.2271,
+      "step": 8285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.213070896235429e-06,
+      "loss": 1.2647,
+      "step": 8286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212975348748328e-06,
+      "loss": 0.8956,
+      "step": 8287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212879801261227e-06,
+      "loss": 1.2464,
+      "step": 8288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2127842537741255e-06,
+      "loss": 1.3026,
+      "step": 8289
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2126887062870246e-06,
+      "loss": 1.1332,
+      "step": 8290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2125931587999244e-06,
+      "loss": 1.1838,
+      "step": 8291
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212497611312823e-06,
+      "loss": 0.8828,
+      "step": 8292
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212402063825722e-06,
+      "loss": 1.5857,
+      "step": 8293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212306516338621e-06,
+      "loss": 1.1758,
+      "step": 8294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21221096885152e-06,
+      "loss": 0.9443,
+      "step": 8295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212115421364418e-06,
+      "loss": 1.2398,
+      "step": 8296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212019873877317e-06,
+      "loss": 1.5881,
+      "step": 8297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211924326390217e-06,
+      "loss": 1.2213,
+      "step": 8298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211828778903115e-06,
+      "loss": 1.1415,
+      "step": 8299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211733231416014e-06,
+      "loss": 1.4469,
+      "step": 8300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211637683928913e-06,
+      "loss": 1.1542,
+      "step": 8301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211542136441812e-06,
+      "loss": 0.7977,
+      "step": 8302
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.21144658895471e-06,
+      "loss": 1.4471,
+      "step": 8303
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.211351041467609e-06,
+      "loss": 1.0399,
+      "step": 8304
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.211255493980509e-06,
+      "loss": 1.3358,
+      "step": 8305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2111599464934075e-06,
+      "loss": 0.8831,
+      "step": 8306
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2110643990063065e-06,
+      "loss": 1.001,
+      "step": 8307
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2109688515192056e-06,
+      "loss": 1.0498,
+      "step": 8308
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210873304032105e-06,
+      "loss": 1.5694,
+      "step": 8309
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210777756545003e-06,
+      "loss": 1.0434,
+      "step": 8310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210682209057902e-06,
+      "loss": 1.2478,
+      "step": 8311
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210586661570801e-06,
+      "loss": 1.7065,
+      "step": 8312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2104911140837e-06,
+      "loss": 1.2238,
+      "step": 8313
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210395566596599e-06,
+      "loss": 0.8815,
+      "step": 8314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210300019109498e-06,
+      "loss": 1.3609,
+      "step": 8315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210204471622397e-06,
+      "loss": 1.1998,
+      "step": 8316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210108924135295e-06,
+      "loss": 1.0515,
+      "step": 8317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.210013376648194e-06,
+      "loss": 0.9607,
+      "step": 8318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209917829161093e-06,
+      "loss": 1.1876,
+      "step": 8319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209822281673992e-06,
+      "loss": 1.589,
+      "step": 8320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209726734186891e-06,
+      "loss": 1.1494,
+      "step": 8321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.20963118669979e-06,
+      "loss": 1.015,
+      "step": 8322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2095356392126894e-06,
+      "loss": 1.0214,
+      "step": 8323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209440091725588e-06,
+      "loss": 0.8266,
+      "step": 8324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209344544238487e-06,
+      "loss": 1.1661,
+      "step": 8325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209248996751386e-06,
+      "loss": 1.349,
+      "step": 8326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209153449264285e-06,
+      "loss": 1.1053,
+      "step": 8327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.209057901777184e-06,
+      "loss": 0.9369,
+      "step": 8328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208962354290083e-06,
+      "loss": 1.4581,
+      "step": 8329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208866806802981e-06,
+      "loss": 1.0481,
+      "step": 8330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.20877125931588e-06,
+      "loss": 0.8483,
+      "step": 8331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208675711828779e-06,
+      "loss": 1.2818,
+      "step": 8332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208580164341678e-06,
+      "loss": 1.2206,
+      "step": 8333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208484616854577e-06,
+      "loss": 0.9872,
+      "step": 8334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208389069367476e-06,
+      "loss": 1.4345,
+      "step": 8335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.208293521880375e-06,
+      "loss": 1.2729,
+      "step": 8336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2081979743932734e-06,
+      "loss": 1.0395,
+      "step": 8337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2081024269061725e-06,
+      "loss": 1.0748,
+      "step": 8338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2080068794190715e-06,
+      "loss": 1.008,
+      "step": 8339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2079113319319706e-06,
+      "loss": 1.2666,
+      "step": 8340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.20781578444487e-06,
+      "loss": 1.3073,
+      "step": 8341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207720236957769e-06,
+      "loss": 0.5799,
+      "step": 8342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207624689470668e-06,
+      "loss": 0.8033,
+      "step": 8343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207529141983566e-06,
+      "loss": 1.1486,
+      "step": 8344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207433594496465e-06,
+      "loss": 1.1771,
+      "step": 8345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207338047009364e-06,
+      "loss": 1.0302,
+      "step": 8346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207242499522263e-06,
+      "loss": 0.8469,
+      "step": 8347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207146952035161e-06,
+      "loss": 1.2984,
+      "step": 8348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.207051404548061e-06,
+      "loss": 1.5967,
+      "step": 8349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.20695585706096e-06,
+      "loss": 0.6297,
+      "step": 8350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206860309573858e-06,
+      "loss": 1.3806,
+      "step": 8351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206764762086757e-06,
+      "loss": 1.5762,
+      "step": 8352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206669214599656e-06,
+      "loss": 1.0305,
+      "step": 8353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206573667112555e-06,
+      "loss": 1.0632,
+      "step": 8354
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.206478119625454e-06,
+      "loss": 0.9669,
+      "step": 8355
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2063825721383535e-06,
+      "loss": 1.3295,
+      "step": 8356
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2062870246512525e-06,
+      "loss": 1.5108,
+      "step": 8357
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.206191477164151e-06,
+      "loss": 0.8262,
+      "step": 8358
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20609592967705e-06,
+      "loss": 1.2098,
+      "step": 8359
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.206000382189949e-06,
+      "loss": 1.558,
+      "step": 8360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205904834702848e-06,
+      "loss": 0.8657,
+      "step": 8361
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205809287215746e-06,
+      "loss": 0.896,
+      "step": 8362
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205713739728646e-06,
+      "loss": 0.9428,
+      "step": 8363
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205618192241545e-06,
+      "loss": 1.1081,
+      "step": 8364
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205522644754443e-06,
+      "loss": 1.3508,
+      "step": 8365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205427097267342e-06,
+      "loss": 1.1341,
+      "step": 8366
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205331549780241e-06,
+      "loss": 1.2686,
+      "step": 8367
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20523600229314e-06,
+      "loss": 1.5597,
+      "step": 8368
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2051404548060384e-06,
+      "loss": 0.957,
+      "step": 8369
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.205044907318938e-06,
+      "loss": 1.0346,
+      "step": 8370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204949359831837e-06,
+      "loss": 1.1475,
+      "step": 8371
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2048538123447356e-06,
+      "loss": 0.8536,
+      "step": 8372
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204758264857635e-06,
+      "loss": 1.1552,
+      "step": 8373
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204662717370534e-06,
+      "loss": 1.1742,
+      "step": 8374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204567169883433e-06,
+      "loss": 1.0952,
+      "step": 8375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204471622396331e-06,
+      "loss": 1.5936,
+      "step": 8376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204376074909231e-06,
+      "loss": 1.1126,
+      "step": 8377
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204280527422129e-06,
+      "loss": 1.0108,
+      "step": 8378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204184979935028e-06,
+      "loss": 0.6668,
+      "step": 8379
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.204089432447927e-06,
+      "loss": 1.0511,
+      "step": 8380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203993884960826e-06,
+      "loss": 1.1328,
+      "step": 8381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203898337473724e-06,
+      "loss": 1.0229,
+      "step": 8382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203802789986623e-06,
+      "loss": 1.7211,
+      "step": 8383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203707242499522e-06,
+      "loss": 1.4618,
+      "step": 8384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203611695012421e-06,
+      "loss": 1.1215,
+      "step": 8385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.20351614752532e-06,
+      "loss": 1.6344,
+      "step": 8386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2034206000382194e-06,
+      "loss": 0.8012,
+      "step": 8387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2033250525511185e-06,
+      "loss": 1.2224,
+      "step": 8388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203229505064017e-06,
+      "loss": 0.9951,
+      "step": 8389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203133957576916e-06,
+      "loss": 1.0738,
+      "step": 8390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.203038410089815e-06,
+      "loss": 1.1644,
+      "step": 8391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202942862602714e-06,
+      "loss": 1.4495,
+      "step": 8392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202847315115613e-06,
+      "loss": 0.8456,
+      "step": 8393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202751767628512e-06,
+      "loss": 1.3403,
+      "step": 8394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202656220141411e-06,
+      "loss": 0.9934,
+      "step": 8395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202560672654309e-06,
+      "loss": 1.0763,
+      "step": 8396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202465125167208e-06,
+      "loss": 1.3925,
+      "step": 8397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202369577680107e-06,
+      "loss": 1.3174,
+      "step": 8398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202274030193006e-06,
+      "loss": 0.9842,
+      "step": 8399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202178482705905e-06,
+      "loss": 2.0098,
+      "step": 8400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202082935218804e-06,
+      "loss": 0.8902,
+      "step": 8401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.201987387731703e-06,
+      "loss": 1.0603,
+      "step": 8402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2018918402446015e-06,
+      "loss": 1.7308,
+      "step": 8403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2017962927575006e-06,
+      "loss": 1.1848,
+      "step": 8404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2017007452704e-06,
+      "loss": 1.4887,
+      "step": 8405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.201605197783299e-06,
+      "loss": 0.938,
+      "step": 8406
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201509650296198e-06,
+      "loss": 0.7576,
+      "step": 8407
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201414102809097e-06,
+      "loss": 1.056,
+      "step": 8408
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201318555321996e-06,
+      "loss": 0.7376,
+      "step": 8409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201223007834894e-06,
+      "loss": 0.9497,
+      "step": 8410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201127460347793e-06,
+      "loss": 1.0464,
+      "step": 8411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.201031912860692e-06,
+      "loss": 1.0726,
+      "step": 8412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.200936365373591e-06,
+      "loss": 1.2748,
+      "step": 8413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.20084081788649e-06,
+      "loss": 1.5119,
+      "step": 8414
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.200745270399389e-06,
+      "loss": 0.6188,
+      "step": 8415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.200649722912288e-06,
+      "loss": 1.3523,
+      "step": 8416
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.200554175425186e-06,
+      "loss": 1.1138,
+      "step": 8417
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.200458627938085e-06,
+      "loss": 0.9854,
+      "step": 8418
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2003630804509844e-06,
+      "loss": 0.9053,
+      "step": 8419
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2002675329638835e-06,
+      "loss": 1.1035,
+      "step": 8420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2001719854767825e-06,
+      "loss": 0.7833,
+      "step": 8421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2000764379896816e-06,
+      "loss": 1.1242,
+      "step": 8422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199980890502581e-06,
+      "loss": 1.1739,
+      "step": 8423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199885343015479e-06,
+      "loss": 0.877,
+      "step": 8424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199789795528378e-06,
+      "loss": 0.9525,
+      "step": 8425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199694248041277e-06,
+      "loss": 1.1692,
+      "step": 8426
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199598700554176e-06,
+      "loss": 0.9762,
+      "step": 8427
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199503153067075e-06,
+      "loss": 1.2652,
+      "step": 8428
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199407605579974e-06,
+      "loss": 1.1365,
+      "step": 8429
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199312058092872e-06,
+      "loss": 1.1877,
+      "step": 8430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199216510605771e-06,
+      "loss": 1.2503,
+      "step": 8431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.19912096311867e-06,
+      "loss": 0.9014,
+      "step": 8432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.199025415631569e-06,
+      "loss": 1.2755,
+      "step": 8433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198929868144468e-06,
+      "loss": 1.2867,
+      "step": 8434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198834320657367e-06,
+      "loss": 1.5155,
+      "step": 8435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198738773170266e-06,
+      "loss": 1.2829,
+      "step": 8436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198643225683165e-06,
+      "loss": 1.4702,
+      "step": 8437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198547678196064e-06,
+      "loss": 0.985,
+      "step": 8438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198452130708963e-06,
+      "loss": 1.3658,
+      "step": 8439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198356583221862e-06,
+      "loss": 1.1691,
+      "step": 8440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.19826103573476e-06,
+      "loss": 1.3057,
+      "step": 8441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.19816548824766e-06,
+      "loss": 0.9693,
+      "step": 8442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.198069940760559e-06,
+      "loss": 1.291,
+      "step": 8443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197974393273457e-06,
+      "loss": 1.2072,
+      "step": 8444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197878845786356e-06,
+      "loss": 1.5692,
+      "step": 8445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197783298299255e-06,
+      "loss": 0.8332,
+      "step": 8446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197687750812154e-06,
+      "loss": 1.0377,
+      "step": 8447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197592203325052e-06,
+      "loss": 1.7605,
+      "step": 8448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197496655837952e-06,
+      "loss": 1.2437,
+      "step": 8449
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197401108350851e-06,
+      "loss": 0.8092,
+      "step": 8450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1973055608637494e-06,
+      "loss": 1.482,
+      "step": 8451
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1972100133766485e-06,
+      "loss": 1.179,
+      "step": 8452
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1971144658895475e-06,
+      "loss": 1.0582,
+      "step": 8453
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1970189184024466e-06,
+      "loss": 1.1637,
+      "step": 8454
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196923370915345e-06,
+      "loss": 0.9534,
+      "step": 8455
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196827823428244e-06,
+      "loss": 0.9842,
+      "step": 8456
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196732275941144e-06,
+      "loss": 1.0736,
+      "step": 8457
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196636728454042e-06,
+      "loss": 1.0988,
+      "step": 8458
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196541180966941e-06,
+      "loss": 1.5162,
+      "step": 8459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19644563347984e-06,
+      "loss": 1.4479,
+      "step": 8460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.196350085992739e-06,
+      "loss": 0.7941,
+      "step": 8461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.196254538505637e-06,
+      "loss": 1.0897,
+      "step": 8462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.196158991018536e-06,
+      "loss": 0.7987,
+      "step": 8463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.196063443531436e-06,
+      "loss": 1.0271,
+      "step": 8464
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195967896044334e-06,
+      "loss": 1.657,
+      "step": 8465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195872348557233e-06,
+      "loss": 1.0056,
+      "step": 8466
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195776801070132e-06,
+      "loss": 1.0281,
+      "step": 8467
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195681253583031e-06,
+      "loss": 1.4449,
+      "step": 8468
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19558570609593e-06,
+      "loss": 0.9291,
+      "step": 8469
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195490158608829e-06,
+      "loss": 0.8583,
+      "step": 8470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195394611121728e-06,
+      "loss": 1.3967,
+      "step": 8471
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195299063634627e-06,
+      "loss": 1.2119,
+      "step": 8472
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195203516147526e-06,
+      "loss": 1.4849,
+      "step": 8473
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195107968660425e-06,
+      "loss": 1.3152,
+      "step": 8474
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.195012421173324e-06,
+      "loss": 0.7645,
+      "step": 8475
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194916873686222e-06,
+      "loss": 1.1339,
+      "step": 8476
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194821326199121e-06,
+      "loss": 1.3096,
+      "step": 8477
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19472577871202e-06,
+      "loss": 1.1286,
+      "step": 8478
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194630231224919e-06,
+      "loss": 0.8813,
+      "step": 8479
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194534683737818e-06,
+      "loss": 1.2532,
+      "step": 8480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194439136250717e-06,
+      "loss": 1.2307,
+      "step": 8481
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.194343588763615e-06,
+      "loss": 1.1569,
+      "step": 8482
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1942480412765144e-06,
+      "loss": 1.2683,
+      "step": 8483
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1941524937894135e-06,
+      "loss": 0.9927,
+      "step": 8484
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1940569463023125e-06,
+      "loss": 1.4917,
+      "step": 8485
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1939613988152116e-06,
+      "loss": 1.0235,
+      "step": 8486
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193865851328111e-06,
+      "loss": 0.9138,
+      "step": 8487
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19377030384101e-06,
+      "loss": 1.1398,
+      "step": 8488
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193674756353908e-06,
+      "loss": 0.7825,
+      "step": 8489
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193579208866807e-06,
+      "loss": 1.1486,
+      "step": 8490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193483661379706e-06,
+      "loss": 1.3306,
+      "step": 8491
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193388113892605e-06,
+      "loss": 1.2963,
+      "step": 8492
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193292566405504e-06,
+      "loss": 0.7127,
+      "step": 8493
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193197018918403e-06,
+      "loss": 1.4634,
+      "step": 8494
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.193101471431302e-06,
+      "loss": 1.1739,
+      "step": 8495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1930059239442e-06,
+      "loss": 1.2041,
+      "step": 8496
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192910376457099e-06,
+      "loss": 0.7831,
+      "step": 8497
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192814828969998e-06,
+      "loss": 1.0896,
+      "step": 8498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192719281482897e-06,
+      "loss": 1.2234,
+      "step": 8499
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192623733995796e-06,
+      "loss": 1.3648,
+      "step": 8500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1925281865086954e-06,
+      "loss": 1.0572,
+      "step": 8501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1924326390215945e-06,
+      "loss": 0.9863,
+      "step": 8502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192337091534493e-06,
+      "loss": 0.9051,
+      "step": 8503
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192241544047392e-06,
+      "loss": 0.7751,
+      "step": 8504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192145996560291e-06,
+      "loss": 0.8654,
+      "step": 8505
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.19205044907319e-06,
+      "loss": 0.9968,
+      "step": 8506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191954901586089e-06,
+      "loss": 1.696,
+      "step": 8507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191859354098988e-06,
+      "loss": 0.959,
+      "step": 8508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191763806611887e-06,
+      "loss": 1.5199,
+      "step": 8509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191668259124785e-06,
+      "loss": 1.0858,
+      "step": 8510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.191572711637684e-06,
+      "loss": 1.7952,
+      "step": 8511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.191477164150583e-06,
+      "loss": 1.2586,
+      "step": 8512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.191381616663482e-06,
+      "loss": 1.1015,
+      "step": 8513
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.191286069176381e-06,
+      "loss": 1.2969,
+      "step": 8514
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19119052168928e-06,
+      "loss": 0.6618,
+      "step": 8515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.191094974202179e-06,
+      "loss": 0.8519,
+      "step": 8516
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1909994267150775e-06,
+      "loss": 1.3585,
+      "step": 8517
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1909038792279766e-06,
+      "loss": 1.0441,
+      "step": 8518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190808331740876e-06,
+      "loss": 1.1451,
+      "step": 8519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190712784253775e-06,
+      "loss": 0.8703,
+      "step": 8520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190617236766674e-06,
+      "loss": 1.3801,
+      "step": 8521
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190521689279573e-06,
+      "loss": 1.064,
+      "step": 8522
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190426141792471e-06,
+      "loss": 1.3409,
+      "step": 8523
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.19033059430537e-06,
+      "loss": 1.435,
+      "step": 8524
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190235046818269e-06,
+      "loss": 1.3887,
+      "step": 8525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190139499331168e-06,
+      "loss": 0.9634,
+      "step": 8526
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.190043951844067e-06,
+      "loss": 1.0185,
+      "step": 8527
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189948404356965e-06,
+      "loss": 1.1954,
+      "step": 8528
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189852856869865e-06,
+      "loss": 1.4364,
+      "step": 8529
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189757309382763e-06,
+      "loss": 1.0604,
+      "step": 8530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189661761895662e-06,
+      "loss": 0.7321,
+      "step": 8531
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189566214408561e-06,
+      "loss": 1.6028,
+      "step": 8532
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1894706669214604e-06,
+      "loss": 1.1388,
+      "step": 8533
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189375119434359e-06,
+      "loss": 0.8569,
+      "step": 8534
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189279571947258e-06,
+      "loss": 0.823,
+      "step": 8535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1891840244601576e-06,
+      "loss": 1.3783,
+      "step": 8536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.189088476973056e-06,
+      "loss": 1.0019,
+      "step": 8537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188992929485955e-06,
+      "loss": 1.1902,
+      "step": 8538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188897381998854e-06,
+      "loss": 1.3221,
+      "step": 8539
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188801834511753e-06,
+      "loss": 1.5929,
+      "step": 8540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188706287024651e-06,
+      "loss": 0.5917,
+      "step": 8541
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18861073953755e-06,
+      "loss": 0.8433,
+      "step": 8542
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18851519205045e-06,
+      "loss": 1.0451,
+      "step": 8543
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188419644563348e-06,
+      "loss": 0.9018,
+      "step": 8544
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188324097076247e-06,
+      "loss": 1.3076,
+      "step": 8545
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188228549589146e-06,
+      "loss": 1.2117,
+      "step": 8546
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.188133002102045e-06,
+      "loss": 0.9233,
+      "step": 8547
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1880374546149435e-06,
+      "loss": 0.8467,
+      "step": 8548
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1879419071278425e-06,
+      "loss": 0.8788,
+      "step": 8549
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187846359640742e-06,
+      "loss": 0.9238,
+      "step": 8550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187750812153641e-06,
+      "loss": 1.0097,
+      "step": 8551
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18765526466654e-06,
+      "loss": 1.1369,
+      "step": 8552
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187559717179439e-06,
+      "loss": 1.1751,
+      "step": 8553
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187464169692338e-06,
+      "loss": 1.116,
+      "step": 8554
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187368622205236e-06,
+      "loss": 0.8046,
+      "step": 8555
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187273074718135e-06,
+      "loss": 1.0522,
+      "step": 8556
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187177527231035e-06,
+      "loss": 1.3741,
+      "step": 8557
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.187081979743933e-06,
+      "loss": 1.0659,
+      "step": 8558
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186986432256832e-06,
+      "loss": 1.3582,
+      "step": 8559
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186890884769731e-06,
+      "loss": 1.1654,
+      "step": 8560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.18679533728263e-06,
+      "loss": 0.9763,
+      "step": 8561
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186699789795528e-06,
+      "loss": 1.6397,
+      "step": 8562
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186604242308427e-06,
+      "loss": 1.596,
+      "step": 8563
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186508694821326e-06,
+      "loss": 0.9234,
+      "step": 8564
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1864131473342254e-06,
+      "loss": 1.1112,
+      "step": 8565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1863175998471245e-06,
+      "loss": 1.8882,
+      "step": 8566
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1862220523600235e-06,
+      "loss": 1.487,
+      "step": 8567
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1861265048729226e-06,
+      "loss": 1.4058,
+      "step": 8568
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.186030957385821e-06,
+      "loss": 1.0967,
+      "step": 8569
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18593540989872e-06,
+      "loss": 1.2651,
+      "step": 8570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185839862411619e-06,
+      "loss": 1.1363,
+      "step": 8571
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185744314924518e-06,
+      "loss": 0.9403,
+      "step": 8572
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185648767437417e-06,
+      "loss": 0.9454,
+      "step": 8573
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185553219950316e-06,
+      "loss": 1.623,
+      "step": 8574
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185457672463214e-06,
+      "loss": 1.3478,
+      "step": 8575
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185362124976113e-06,
+      "loss": 1.2526,
+      "step": 8576
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185266577489012e-06,
+      "loss": 1.2422,
+      "step": 8577
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.185171030001911e-06,
+      "loss": 1.5688,
+      "step": 8578
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18507548251481e-06,
+      "loss": 1.168,
+      "step": 8579
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184979935027709e-06,
+      "loss": 1.154,
+      "step": 8580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184884387540608e-06,
+      "loss": 1.2683,
+      "step": 8581
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1847888400535065e-06,
+      "loss": 1.4682,
+      "step": 8582
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184693292566406e-06,
+      "loss": 1.0193,
+      "step": 8583
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184597745079305e-06,
+      "loss": 1.2792,
+      "step": 8584
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184502197592204e-06,
+      "loss": 1.0223,
+      "step": 8585
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184406650105103e-06,
+      "loss": 1.8552,
+      "step": 8586
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184311102618002e-06,
+      "loss": 1.3268,
+      "step": 8587
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184215555130901e-06,
+      "loss": 1.0248,
+      "step": 8588
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184120007643799e-06,
+      "loss": 0.7842,
+      "step": 8589
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.184024460156698e-06,
+      "loss": 0.9372,
+      "step": 8590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183928912669597e-06,
+      "loss": 0.8457,
+      "step": 8591
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183833365182496e-06,
+      "loss": 1.0409,
+      "step": 8592
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183737817695395e-06,
+      "loss": 1.2875,
+      "step": 8593
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183642270208294e-06,
+      "loss": 1.1479,
+      "step": 8594
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183546722721193e-06,
+      "loss": 0.612,
+      "step": 8595
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183451175234091e-06,
+      "loss": 1.0125,
+      "step": 8596
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1833556277469904e-06,
+      "loss": 1.1435,
+      "step": 8597
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1832600802598895e-06,
+      "loss": 1.1882,
+      "step": 8598
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1831645327727885e-06,
+      "loss": 0.7958,
+      "step": 8599
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.183068985285687e-06,
+      "loss": 1.2852,
+      "step": 8600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182973437798587e-06,
+      "loss": 0.9532,
+      "step": 8601
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182877890311486e-06,
+      "loss": 1.2367,
+      "step": 8602
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182782342824384e-06,
+      "loss": 0.7982,
+      "step": 8603
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182686795337283e-06,
+      "loss": 0.7997,
+      "step": 8604
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182591247850182e-06,
+      "loss": 0.9508,
+      "step": 8605
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182495700363081e-06,
+      "loss": 1.17,
+      "step": 8606
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182400152875979e-06,
+      "loss": 0.9984,
+      "step": 8607
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182304605388879e-06,
+      "loss": 1.3202,
+      "step": 8608
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182209057901778e-06,
+      "loss": 0.7358,
+      "step": 8609
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182113510414676e-06,
+      "loss": 1.3974,
+      "step": 8610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.182017962927575e-06,
+      "loss": 1.1788,
+      "step": 8611
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.181922415440474e-06,
+      "loss": 1.3231,
+      "step": 8612
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.181826867953373e-06,
+      "loss": 1.7125,
+      "step": 8613
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1817313204662715e-06,
+      "loss": 1.462,
+      "step": 8614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1816357729791714e-06,
+      "loss": 0.9827,
+      "step": 8615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1815402254920705e-06,
+      "loss": 1.3116,
+      "step": 8616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.181444678004969e-06,
+      "loss": 1.094,
+      "step": 8617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.181349130517868e-06,
+      "loss": 1.1594,
+      "step": 8618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.181253583030767e-06,
+      "loss": 0.9958,
+      "step": 8619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.181158035543666e-06,
+      "loss": 1.0839,
+      "step": 8620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.181062488056564e-06,
+      "loss": 1.3789,
+      "step": 8621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180966940569464e-06,
+      "loss": 1.5602,
+      "step": 8622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180871393082362e-06,
+      "loss": 0.996,
+      "step": 8623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180775845595261e-06,
+      "loss": 1.7371,
+      "step": 8624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.18068029810816e-06,
+      "loss": 1.1182,
+      "step": 8625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180584750621059e-06,
+      "loss": 1.212,
+      "step": 8626
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180489203133958e-06,
+      "loss": 1.0815,
+      "step": 8627
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180393655646856e-06,
+      "loss": 1.3749,
+      "step": 8628
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.180298108159756e-06,
+      "loss": 1.1297,
+      "step": 8629
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1802025606726545e-06,
+      "loss": 1.3897,
+      "step": 8630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1801070131855535e-06,
+      "loss": 0.97,
+      "step": 8631
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1800114656984525e-06,
+      "loss": 0.9143,
+      "step": 8632
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179915918211352e-06,
+      "loss": 1.045,
+      "step": 8633
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.17982037072425e-06,
+      "loss": 1.1595,
+      "step": 8634
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179724823237149e-06,
+      "loss": 1.1145,
+      "step": 8635
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179629275750048e-06,
+      "loss": 1.4821,
+      "step": 8636
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179533728262947e-06,
+      "loss": 1.1738,
+      "step": 8637
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179438180775846e-06,
+      "loss": 1.0864,
+      "step": 8638
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179342633288745e-06,
+      "loss": 1.3841,
+      "step": 8639
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179247085801644e-06,
+      "loss": 1.2824,
+      "step": 8640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179151538314542e-06,
+      "loss": 1.011,
+      "step": 8641
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.179055990827441e-06,
+      "loss": 0.9066,
+      "step": 8642
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.17896044334034e-06,
+      "loss": 1.1689,
+      "step": 8643
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178864895853239e-06,
+      "loss": 1.2428,
+      "step": 8644
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178769348366138e-06,
+      "loss": 1.375,
+      "step": 8645
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178673800879037e-06,
+      "loss": 1.7754,
+      "step": 8646
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1785782533919364e-06,
+      "loss": 0.9256,
+      "step": 8647
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178482705904835e-06,
+      "loss": 1.4275,
+      "step": 8648
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178387158417734e-06,
+      "loss": 1.4749,
+      "step": 8649
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178291610930633e-06,
+      "loss": 1.0498,
+      "step": 8650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178196063443532e-06,
+      "loss": 1.1877,
+      "step": 8651
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.178100515956431e-06,
+      "loss": 0.9721,
+      "step": 8652
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.17800496846933e-06,
+      "loss": 0.9515,
+      "step": 8653
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177909420982229e-06,
+      "loss": 0.8742,
+      "step": 8654
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177813873495127e-06,
+      "loss": 1.5514,
+      "step": 8655
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177718326008026e-06,
+      "loss": 1.3595,
+      "step": 8656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177622778520925e-06,
+      "loss": 1.2659,
+      "step": 8657
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177527231033824e-06,
+      "loss": 0.6804,
+      "step": 8658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177431683546723e-06,
+      "loss": 1.1541,
+      "step": 8659
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177336136059622e-06,
+      "loss": 0.9055,
+      "step": 8660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177240588572521e-06,
+      "loss": 0.6526,
+      "step": 8661
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1771450410854195e-06,
+      "loss": 0.9086,
+      "step": 8662
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1770494935983185e-06,
+      "loss": 0.8929,
+      "step": 8663
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1769539461112175e-06,
+      "loss": 1.3212,
+      "step": 8664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176858398624117e-06,
+      "loss": 1.2519,
+      "step": 8665
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176762851137016e-06,
+      "loss": 1.0482,
+      "step": 8666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176667303649915e-06,
+      "loss": 1.119,
+      "step": 8667
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.176571756162814e-06,
+      "loss": 1.194,
+      "step": 8668
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.176476208675712e-06,
+      "loss": 0.8147,
+      "step": 8669
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.176380661188611e-06,
+      "loss": 1.1897,
+      "step": 8670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17628511370151e-06,
+      "loss": 0.8319,
+      "step": 8671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.176189566214409e-06,
+      "loss": 1.0756,
+      "step": 8672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.176094018727308e-06,
+      "loss": 1.6021,
+      "step": 8673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175998471240207e-06,
+      "loss": 0.9773,
+      "step": 8674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175902923753105e-06,
+      "loss": 1.2628,
+      "step": 8675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175807376266004e-06,
+      "loss": 1.5561,
+      "step": 8676
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175711828778903e-06,
+      "loss": 0.9799,
+      "step": 8677
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175616281291802e-06,
+      "loss": 1.1712,
+      "step": 8678
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1755207338047014e-06,
+      "loss": 1.2825,
+      "step": 8679
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1754251863176005e-06,
+      "loss": 0.839,
+      "step": 8680
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1753296388304995e-06,
+      "loss": 0.5807,
+      "step": 8681
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175234091343398e-06,
+      "loss": 0.7066,
+      "step": 8682
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175138543856297e-06,
+      "loss": 1.1516,
+      "step": 8683
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.175042996369196e-06,
+      "loss": 0.942,
+      "step": 8684
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174947448882095e-06,
+      "loss": 0.7451,
+      "step": 8685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174851901394993e-06,
+      "loss": 1.1123,
+      "step": 8686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174756353907893e-06,
+      "loss": 1.3065,
+      "step": 8687
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174660806420792e-06,
+      "loss": 1.1373,
+      "step": 8688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17456525893369e-06,
+      "loss": 1.2163,
+      "step": 8689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174469711446589e-06,
+      "loss": 1.2123,
+      "step": 8690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174374163959488e-06,
+      "loss": 1.7121,
+      "step": 8691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174278616472387e-06,
+      "loss": 1.188,
+      "step": 8692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1741830689852854e-06,
+      "loss": 1.4328,
+      "step": 8693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.174087521498185e-06,
+      "loss": 0.9971,
+      "step": 8694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173991974011084e-06,
+      "loss": 0.9667,
+      "step": 8695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1738964265239825e-06,
+      "loss": 1.32,
+      "step": 8696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173800879036882e-06,
+      "loss": 1.3016,
+      "step": 8697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173705331549781e-06,
+      "loss": 1.0519,
+      "step": 8698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17360978406268e-06,
+      "loss": 0.9497,
+      "step": 8699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173514236575578e-06,
+      "loss": 1.1157,
+      "step": 8700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173418689088478e-06,
+      "loss": 0.6201,
+      "step": 8701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173323141601377e-06,
+      "loss": 1.3642,
+      "step": 8702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173227594114275e-06,
+      "loss": 1.0728,
+      "step": 8703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173132046627174e-06,
+      "loss": 1.07,
+      "step": 8704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.173036499140073e-06,
+      "loss": 0.6047,
+      "step": 8705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172940951652972e-06,
+      "loss": 1.7295,
+      "step": 8706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17284540416587e-06,
+      "loss": 0.9239,
+      "step": 8707
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172749856678769e-06,
+      "loss": 1.1886,
+      "step": 8708
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172654309191669e-06,
+      "loss": 1.131,
+      "step": 8709
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172558761704567e-06,
+      "loss": 1.142,
+      "step": 8710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1724632142174664e-06,
+      "loss": 1.4735,
+      "step": 8711
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1723676667303655e-06,
+      "loss": 0.9123,
+      "step": 8712
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1722721192432645e-06,
+      "loss": 1.1065,
+      "step": 8713
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172176571756163e-06,
+      "loss": 0.8508,
+      "step": 8714
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.172081024269062e-06,
+      "loss": 0.9674,
+      "step": 8715
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171985476781961e-06,
+      "loss": 1.26,
+      "step": 8716
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.17188992929486e-06,
+      "loss": 1.0441,
+      "step": 8717
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171794381807759e-06,
+      "loss": 1.2008,
+      "step": 8718
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171698834320658e-06,
+      "loss": 1.4554,
+      "step": 8719
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171603286833557e-06,
+      "loss": 1.2286,
+      "step": 8720
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171507739346455e-06,
+      "loss": 1.2655,
+      "step": 8721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.171412191859354e-06,
+      "loss": 0.8068,
+      "step": 8722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.171316644372253e-06,
+      "loss": 0.9974,
+      "step": 8723
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.171221096885152e-06,
+      "loss": 1.1769,
+      "step": 8724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.171125549398051e-06,
+      "loss": 1.1138,
+      "step": 8725
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17103000191095e-06,
+      "loss": 1.2665,
+      "step": 8726
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1709344544238485e-06,
+      "loss": 1.5953,
+      "step": 8727
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1708389069367475e-06,
+      "loss": 1.2461,
+      "step": 8728
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170743359449647e-06,
+      "loss": 0.7712,
+      "step": 8729
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170647811962546e-06,
+      "loss": 1.3762,
+      "step": 8730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170552264475445e-06,
+      "loss": 1.3139,
+      "step": 8731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170456716988344e-06,
+      "loss": 1.5146,
+      "step": 8732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170361169501243e-06,
+      "loss": 0.9909,
+      "step": 8733
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170265622014141e-06,
+      "loss": 0.9205,
+      "step": 8734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.17017007452704e-06,
+      "loss": 1.2918,
+      "step": 8735
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.170074527039939e-06,
+      "loss": 0.9685,
+      "step": 8736
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169978979552838e-06,
+      "loss": 0.9361,
+      "step": 8737
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169883432065737e-06,
+      "loss": 1.3359,
+      "step": 8738
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169787884578636e-06,
+      "loss": 1.523,
+      "step": 8739
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169692337091535e-06,
+      "loss": 1.3875,
+      "step": 8740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169596789604433e-06,
+      "loss": 1.0875,
+      "step": 8741
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169501242117332e-06,
+      "loss": 0.9735,
+      "step": 8742
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1694056946302314e-06,
+      "loss": 1.2908,
+      "step": 8743
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1693101471431305e-06,
+      "loss": 0.8314,
+      "step": 8744
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1692145996560295e-06,
+      "loss": 0.8575,
+      "step": 8745
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1691190521689285e-06,
+      "loss": 1.062,
+      "step": 8746
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.169023504681828e-06,
+      "loss": 0.8325,
+      "step": 8747
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168927957194726e-06,
+      "loss": 0.8287,
+      "step": 8748
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168832409707625e-06,
+      "loss": 0.6893,
+      "step": 8749
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168736862220524e-06,
+      "loss": 1.3764,
+      "step": 8750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168641314733423e-06,
+      "loss": 1.2604,
+      "step": 8751
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168545767246322e-06,
+      "loss": 1.1577,
+      "step": 8752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168450219759221e-06,
+      "loss": 0.6876,
+      "step": 8753
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.16835467227212e-06,
+      "loss": 0.744,
+      "step": 8754
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168259124785018e-06,
+      "loss": 1.4637,
+      "step": 8755
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168163577297917e-06,
+      "loss": 0.9714,
+      "step": 8756
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.168068029810816e-06,
+      "loss": 1.3705,
+      "step": 8757
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167972482323715e-06,
+      "loss": 1.1168,
+      "step": 8758
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167876934836614e-06,
+      "loss": 0.8036,
+      "step": 8759
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167781387349513e-06,
+      "loss": 1.0577,
+      "step": 8760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1676858398624124e-06,
+      "loss": 0.8538,
+      "step": 8761
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167590292375311e-06,
+      "loss": 0.9026,
+      "step": 8762
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.16749474488821e-06,
+      "loss": 1.1908,
+      "step": 8763
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167399197401109e-06,
+      "loss": 1.0946,
+      "step": 8764
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167303649914008e-06,
+      "loss": 1.2871,
+      "step": 8765
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167208102426907e-06,
+      "loss": 1.32,
+      "step": 8766
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167112554939806e-06,
+      "loss": 1.4657,
+      "step": 8767
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167017007452705e-06,
+      "loss": 1.2275,
+      "step": 8768
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166921459965603e-06,
+      "loss": 1.69,
+      "step": 8769
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166825912478502e-06,
+      "loss": 1.1353,
+      "step": 8770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166730364991401e-06,
+      "loss": 0.9967,
+      "step": 8771
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1666348175043e-06,
+      "loss": 1.2114,
+      "step": 8772
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166539270017199e-06,
+      "loss": 0.9819,
+      "step": 8773
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.166443722530098e-06,
+      "loss": 0.9974,
+      "step": 8774
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1663481750429964e-06,
+      "loss": 1.3725,
+      "step": 8775
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1662526275558955e-06,
+      "loss": 0.9958,
+      "step": 8776
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1661570800687945e-06,
+      "loss": 1.1111,
+      "step": 8777
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1660615325816935e-06,
+      "loss": 0.9737,
+      "step": 8778
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165965985094592e-06,
+      "loss": 1.2867,
+      "step": 8779
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165870437607491e-06,
+      "loss": 1.0592,
+      "step": 8780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165774890120391e-06,
+      "loss": 1.1039,
+      "step": 8781
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165679342633289e-06,
+      "loss": 1.1561,
+      "step": 8782
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165583795146188e-06,
+      "loss": 0.9242,
+      "step": 8783
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165488247659087e-06,
+      "loss": 1.0504,
+      "step": 8784
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165392700171986e-06,
+      "loss": 0.8307,
+      "step": 8785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165297152684884e-06,
+      "loss": 1.0895,
+      "step": 8786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165201605197783e-06,
+      "loss": 1.4534,
+      "step": 8787
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165106057710683e-06,
+      "loss": 1.4262,
+      "step": 8788
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.165010510223581e-06,
+      "loss": 1.132,
+      "step": 8789
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16491496273648e-06,
+      "loss": 0.8114,
+      "step": 8790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164819415249379e-06,
+      "loss": 1.3964,
+      "step": 8791
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164723867762278e-06,
+      "loss": 1.1142,
+      "step": 8792
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164628320275177e-06,
+      "loss": 1.154,
+      "step": 8793
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164532772788076e-06,
+      "loss": 0.7303,
+      "step": 8794
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1644372253009755e-06,
+      "loss": 1.2878,
+      "step": 8795
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164341677813874e-06,
+      "loss": 0.7544,
+      "step": 8796
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164246130326773e-06,
+      "loss": 1.2172,
+      "step": 8797
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164150582839672e-06,
+      "loss": 1.2183,
+      "step": 8798
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.164055035352571e-06,
+      "loss": 0.9638,
+      "step": 8799
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163959487865469e-06,
+      "loss": 1.0468,
+      "step": 8800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163863940378368e-06,
+      "loss": 0.8928,
+      "step": 8801
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163768392891268e-06,
+      "loss": 1.0504,
+      "step": 8802
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163672845404166e-06,
+      "loss": 0.9903,
+      "step": 8803
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163577297917065e-06,
+      "loss": 0.9869,
+      "step": 8804
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163481750429964e-06,
+      "loss": 0.9699,
+      "step": 8805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.163386202942863e-06,
+      "loss": 1.3293,
+      "step": 8806
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1632906554557614e-06,
+      "loss": 1.3551,
+      "step": 8807
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1631951079686605e-06,
+      "loss": 1.356,
+      "step": 8808
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16309956048156e-06,
+      "loss": 1.4046,
+      "step": 8809
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1630040129944585e-06,
+      "loss": 0.9404,
+      "step": 8810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162908465507358e-06,
+      "loss": 1.5756,
+      "step": 8811
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162812918020257e-06,
+      "loss": 0.9427,
+      "step": 8812
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162717370533156e-06,
+      "loss": 1.5231,
+      "step": 8813
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162621823046054e-06,
+      "loss": 0.9405,
+      "step": 8814
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162526275558953e-06,
+      "loss": 0.8303,
+      "step": 8815
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162430728071852e-06,
+      "loss": 1.4049,
+      "step": 8816
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162335180584751e-06,
+      "loss": 1.2989,
+      "step": 8817
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.16223963309765e-06,
+      "loss": 0.8269,
+      "step": 8818
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162144085610549e-06,
+      "loss": 0.5945,
+      "step": 8819
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162048538123448e-06,
+      "loss": 1.1366,
+      "step": 8820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161952990636346e-06,
+      "loss": 1.4792,
+      "step": 8821
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161857443149245e-06,
+      "loss": 1.2831,
+      "step": 8822
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161761895662144e-06,
+      "loss": 1.3725,
+      "step": 8823
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161666348175043e-06,
+      "loss": 1.0123,
+      "step": 8824
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1615708006879424e-06,
+      "loss": 0.9696,
+      "step": 8825
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1614752532008415e-06,
+      "loss": 1.034,
+      "step": 8826
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16137970571374e-06,
+      "loss": 1.1299,
+      "step": 8827
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.161284158226639e-06,
+      "loss": 1.0063,
+      "step": 8828
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.161188610739538e-06,
+      "loss": 1.4672,
+      "step": 8829
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.161093063252437e-06,
+      "loss": 1.6885,
+      "step": 8830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160997515765336e-06,
+      "loss": 1.3294,
+      "step": 8831
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160901968278235e-06,
+      "loss": 1.0289,
+      "step": 8832
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160806420791134e-06,
+      "loss": 1.3159,
+      "step": 8833
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160710873304032e-06,
+      "loss": 0.8285,
+      "step": 8834
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160615325816931e-06,
+      "loss": 1.0041,
+      "step": 8835
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.16051977832983e-06,
+      "loss": 1.2252,
+      "step": 8836
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160424230842729e-06,
+      "loss": 0.8194,
+      "step": 8837
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160328683355628e-06,
+      "loss": 1.1517,
+      "step": 8838
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160233135868527e-06,
+      "loss": 1.3615,
+      "step": 8839
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.160137588381426e-06,
+      "loss": 1.563,
+      "step": 8840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1600420408943245e-06,
+      "loss": 0.9384,
+      "step": 8841
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1599464934072235e-06,
+      "loss": 1.1895,
+      "step": 8842
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159850945920123e-06,
+      "loss": 0.649,
+      "step": 8843
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159755398433022e-06,
+      "loss": 1.0692,
+      "step": 8844
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159659850945921e-06,
+      "loss": 0.8167,
+      "step": 8845
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.15956430345882e-06,
+      "loss": 1.2968,
+      "step": 8846
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159468755971719e-06,
+      "loss": 1.4161,
+      "step": 8847
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159373208484617e-06,
+      "loss": 1.3714,
+      "step": 8848
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159277660997516e-06,
+      "loss": 1.4549,
+      "step": 8849
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159182113510415e-06,
+      "loss": 1.4287,
+      "step": 8850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.159086566023314e-06,
+      "loss": 1.1274,
+      "step": 8851
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158991018536212e-06,
+      "loss": 0.5883,
+      "step": 8852
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158895471049112e-06,
+      "loss": 1.296,
+      "step": 8853
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158799923562011e-06,
+      "loss": 1.1311,
+      "step": 8854
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158704376074909e-06,
+      "loss": 1.1736,
+      "step": 8855
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158608828587808e-06,
+      "loss": 1.0083,
+      "step": 8856
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1585132811007074e-06,
+      "loss": 0.9123,
+      "step": 8857
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1584177336136065e-06,
+      "loss": 1.4507,
+      "step": 8858
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158322186126505e-06,
+      "loss": 1.076,
+      "step": 8859
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1582266386394045e-06,
+      "loss": 1.0893,
+      "step": 8860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158131091152304e-06,
+      "loss": 0.9373,
+      "step": 8861
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.158035543665202e-06,
+      "loss": 1.364,
+      "step": 8862
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157939996178101e-06,
+      "loss": 1.1446,
+      "step": 8863
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157844448691e-06,
+      "loss": 0.8845,
+      "step": 8864
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157748901203899e-06,
+      "loss": 1.2592,
+      "step": 8865
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157653353716797e-06,
+      "loss": 1.591,
+      "step": 8866
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157557806229697e-06,
+      "loss": 1.0387,
+      "step": 8867
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157462258742595e-06,
+      "loss": 1.4722,
+      "step": 8868
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157366711255494e-06,
+      "loss": 1.412,
+      "step": 8869
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157271163768393e-06,
+      "loss": 1.2791,
+      "step": 8870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157175616281292e-06,
+      "loss": 1.3469,
+      "step": 8871
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157080068794191e-06,
+      "loss": 1.7064,
+      "step": 8872
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1569845213070895e-06,
+      "loss": 1.2052,
+      "step": 8873
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156888973819989e-06,
+      "loss": 0.9543,
+      "step": 8874
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156793426332888e-06,
+      "loss": 0.7855,
+      "step": 8875
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156697878845787e-06,
+      "loss": 0.9826,
+      "step": 8876
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156602331358686e-06,
+      "loss": 0.8342,
+      "step": 8877
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.156506783871585e-06,
+      "loss": 1.0924,
+      "step": 8878
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.156411236384483e-06,
+      "loss": 1.1692,
+      "step": 8879
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.156315688897382e-06,
+      "loss": 0.9231,
+      "step": 8880
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.156220141410282e-06,
+      "loss": 0.7557,
+      "step": 8881
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15612459392318e-06,
+      "loss": 1.1415,
+      "step": 8882
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.156029046436079e-06,
+      "loss": 0.861,
+      "step": 8883
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155933498948978e-06,
+      "loss": 1.1624,
+      "step": 8884
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155837951461877e-06,
+      "loss": 1.553,
+      "step": 8885
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155742403974775e-06,
+      "loss": 1.0961,
+      "step": 8886
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155646856487674e-06,
+      "loss": 1.5774,
+      "step": 8887
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155551309000573e-06,
+      "loss": 1.6918,
+      "step": 8888
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1554557615134724e-06,
+      "loss": 0.6828,
+      "step": 8889
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1553602140263715e-06,
+      "loss": 1.4461,
+      "step": 8890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1552646665392705e-06,
+      "loss": 0.9185,
+      "step": 8891
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1551691190521695e-06,
+      "loss": 1.1389,
+      "step": 8892
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.155073571565068e-06,
+      "loss": 1.1474,
+      "step": 8893
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154978024077967e-06,
+      "loss": 0.89,
+      "step": 8894
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154882476590866e-06,
+      "loss": 0.6083,
+      "step": 8895
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154786929103765e-06,
+      "loss": 1.2147,
+      "step": 8896
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154691381616664e-06,
+      "loss": 0.7951,
+      "step": 8897
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154595834129563e-06,
+      "loss": 1.1113,
+      "step": 8898
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154500286642462e-06,
+      "loss": 1.384,
+      "step": 8899
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15440473915536e-06,
+      "loss": 1.21,
+      "step": 8900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154309191668259e-06,
+      "loss": 1.3582,
+      "step": 8901
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154213644181158e-06,
+      "loss": 1.1377,
+      "step": 8902
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154118096694057e-06,
+      "loss": 1.4766,
+      "step": 8903
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.154022549206956e-06,
+      "loss": 0.6221,
+      "step": 8904
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153927001719855e-06,
+      "loss": 1.0883,
+      "step": 8905
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153831454232754e-06,
+      "loss": 1.2759,
+      "step": 8906
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153735906745653e-06,
+      "loss": 1.4191,
+      "step": 8907
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153640359258552e-06,
+      "loss": 1.2407,
+      "step": 8908
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153544811771451e-06,
+      "loss": 0.8791,
+      "step": 8909
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15344926428435e-06,
+      "loss": 0.8496,
+      "step": 8910
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153353716797249e-06,
+      "loss": 1.1521,
+      "step": 8911
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153258169310148e-06,
+      "loss": 1.6694,
+      "step": 8912
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153162621823047e-06,
+      "loss": 0.8964,
+      "step": 8913
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.153067074335945e-06,
+      "loss": 0.7556,
+      "step": 8914
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152971526848844e-06,
+      "loss": 0.9245,
+      "step": 8915
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152875979361743e-06,
+      "loss": 1.1058,
+      "step": 8916
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152780431874642e-06,
+      "loss": 1.1427,
+      "step": 8917
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152684884387541e-06,
+      "loss": 0.7787,
+      "step": 8918
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15258933690044e-06,
+      "loss": 1.2133,
+      "step": 8919
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152493789413338e-06,
+      "loss": 1.5511,
+      "step": 8920
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1523982419262374e-06,
+      "loss": 1.217,
+      "step": 8921
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1523026944391365e-06,
+      "loss": 0.6095,
+      "step": 8922
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1522071469520355e-06,
+      "loss": 1.1734,
+      "step": 8923
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1521115994649345e-06,
+      "loss": 1.153,
+      "step": 8924
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152016051977834e-06,
+      "loss": 1.6159,
+      "step": 8925
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151920504490733e-06,
+      "loss": 1.4056,
+      "step": 8926
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151824957003631e-06,
+      "loss": 1.276,
+      "step": 8927
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.15172940951653e-06,
+      "loss": 0.9865,
+      "step": 8928
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151633862029429e-06,
+      "loss": 1.3451,
+      "step": 8929
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151538314542328e-06,
+      "loss": 0.9843,
+      "step": 8930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.151442767055226e-06,
+      "loss": 0.7937,
+      "step": 8931
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.151347219568126e-06,
+      "loss": 1.1219,
+      "step": 8932
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.151251672081025e-06,
+      "loss": 0.9237,
+      "step": 8933
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.151156124593923e-06,
+      "loss": 1.6737,
+      "step": 8934
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.151060577106822e-06,
+      "loss": 1.1125,
+      "step": 8935
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150965029619721e-06,
+      "loss": 0.7782,
+      "step": 8936
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.15086948213262e-06,
+      "loss": 1.1986,
+      "step": 8937
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1507739346455185e-06,
+      "loss": 1.176,
+      "step": 8938
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1506783871584184e-06,
+      "loss": 1.0488,
+      "step": 8939
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1505828396713175e-06,
+      "loss": 1.1534,
+      "step": 8940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150487292184216e-06,
+      "loss": 1.5123,
+      "step": 8941
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150391744697115e-06,
+      "loss": 1.0541,
+      "step": 8942
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150296197210014e-06,
+      "loss": 1.2689,
+      "step": 8943
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150200649722913e-06,
+      "loss": 1.8106,
+      "step": 8944
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150105102235811e-06,
+      "loss": 1.1257,
+      "step": 8945
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.150009554748711e-06,
+      "loss": 1.1506,
+      "step": 8946
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14991400726161e-06,
+      "loss": 1.6348,
+      "step": 8947
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149818459774508e-06,
+      "loss": 1.2625,
+      "step": 8948
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149722912287407e-06,
+      "loss": 0.6839,
+      "step": 8949
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149627364800306e-06,
+      "loss": 1.4,
+      "step": 8950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149531817313205e-06,
+      "loss": 1.1213,
+      "step": 8951
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149436269826103e-06,
+      "loss": 1.351,
+      "step": 8952
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149340722339003e-06,
+      "loss": 1.0868,
+      "step": 8953
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.149245174851902e-06,
+      "loss": 1.0899,
+      "step": 8954
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1491496273648005e-06,
+      "loss": 1.4367,
+      "step": 8955
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1490540798776995e-06,
+      "loss": 0.8109,
+      "step": 8956
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148958532390599e-06,
+      "loss": 0.9918,
+      "step": 8957
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148862984903498e-06,
+      "loss": 1.3292,
+      "step": 8958
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148767437416396e-06,
+      "loss": 0.6334,
+      "step": 8959
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148671889929295e-06,
+      "loss": 1.0371,
+      "step": 8960
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148576342442195e-06,
+      "loss": 0.9572,
+      "step": 8961
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148480794955093e-06,
+      "loss": 1.3035,
+      "step": 8962
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148385247467992e-06,
+      "loss": 1.2958,
+      "step": 8963
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148289699980891e-06,
+      "loss": 1.1282,
+      "step": 8964
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14819415249379e-06,
+      "loss": 1.3006,
+      "step": 8965
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148098605006688e-06,
+      "loss": 0.9303,
+      "step": 8966
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.148003057519587e-06,
+      "loss": 1.0017,
+      "step": 8967
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147907510032486e-06,
+      "loss": 1.3393,
+      "step": 8968
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147811962545385e-06,
+      "loss": 1.5655,
+      "step": 8969
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147716415058284e-06,
+      "loss": 1.0614,
+      "step": 8970
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1476208675711834e-06,
+      "loss": 1.185,
+      "step": 8971
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147525320084082e-06,
+      "loss": 0.8329,
+      "step": 8972
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147429772596981e-06,
+      "loss": 1.3363,
+      "step": 8973
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14733422510988e-06,
+      "loss": 1.3552,
+      "step": 8974
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147238677622779e-06,
+      "loss": 1.0674,
+      "step": 8975
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147143130135678e-06,
+      "loss": 1.2014,
+      "step": 8976
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147047582648577e-06,
+      "loss": 1.5802,
+      "step": 8977
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146952035161476e-06,
+      "loss": 1.1224,
+      "step": 8978
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146856487674374e-06,
+      "loss": 1.0944,
+      "step": 8979
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146760940187273e-06,
+      "loss": 1.1963,
+      "step": 8980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146665392700172e-06,
+      "loss": 1.2393,
+      "step": 8981
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146569845213071e-06,
+      "loss": 0.7385,
+      "step": 8982
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14647429772597e-06,
+      "loss": 1.4636,
+      "step": 8983
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.146378750238869e-06,
+      "loss": 1.1482,
+      "step": 8984
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.146283202751768e-06,
+      "loss": 1.7152,
+      "step": 8985
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1461876552646665e-06,
+      "loss": 1.0733,
+      "step": 8986
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1460921077775655e-06,
+      "loss": 1.241,
+      "step": 8987
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1459965602904645e-06,
+      "loss": 1.2507,
+      "step": 8988
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145901012803364e-06,
+      "loss": 0.8575,
+      "step": 8989
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145805465316263e-06,
+      "loss": 1.2428,
+      "step": 8990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145709917829162e-06,
+      "loss": 0.6724,
+      "step": 8991
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145614370342061e-06,
+      "loss": 0.9382,
+      "step": 8992
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145518822854959e-06,
+      "loss": 1.8863,
+      "step": 8993
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145423275367858e-06,
+      "loss": 0.713,
+      "step": 8994
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145327727880757e-06,
+      "loss": 1.091,
+      "step": 8995
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145232180393656e-06,
+      "loss": 1.21,
+      "step": 8996
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145136632906555e-06,
+      "loss": 1.1122,
+      "step": 8997
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.145041085419454e-06,
+      "loss": 1.0966,
+      "step": 8998
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.144945537932353e-06,
+      "loss": 1.0934,
+      "step": 8999
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.144849990445251e-06,
+      "loss": 1.4554,
+      "step": 9000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14475444295815e-06,
+      "loss": 1.7776,
+      "step": 9001
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.144658895471049e-06,
+      "loss": 1.0893,
+      "step": 9002
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1445633479839484e-06,
+      "loss": 1.1237,
+      "step": 9003
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1444678004968475e-06,
+      "loss": 1.4568,
+      "step": 9004
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1443722530097465e-06,
+      "loss": 0.8268,
+      "step": 9005
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1442767055226455e-06,
+      "loss": 1.0942,
+      "step": 9006
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.144181158035544e-06,
+      "loss": 1.0683,
+      "step": 9007
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.144085610548443e-06,
+      "loss": 1.2664,
+      "step": 9008
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143990063061342e-06,
+      "loss": 0.9362,
+      "step": 9009
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143894515574241e-06,
+      "loss": 1.2118,
+      "step": 9010
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14379896808714e-06,
+      "loss": 1.3126,
+      "step": 9011
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143703420600039e-06,
+      "loss": 1.1954,
+      "step": 9012
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143607873112938e-06,
+      "loss": 0.7346,
+      "step": 9013
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143512325625836e-06,
+      "loss": 1.1161,
+      "step": 9014
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143416778138735e-06,
+      "loss": 1.6865,
+      "step": 9015
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143321230651634e-06,
+      "loss": 1.0806,
+      "step": 9016
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143225683164533e-06,
+      "loss": 1.6498,
+      "step": 9017
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143130135677432e-06,
+      "loss": 0.852,
+      "step": 9018
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.143034588190331e-06,
+      "loss": 0.9032,
+      "step": 9019
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1429390407032295e-06,
+      "loss": 0.863,
+      "step": 9020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142843493216129e-06,
+      "loss": 1.0561,
+      "step": 9021
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142747945729028e-06,
+      "loss": 1.1677,
+      "step": 9022
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142652398241927e-06,
+      "loss": 1.1544,
+      "step": 9023
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142556850754826e-06,
+      "loss": 0.6168,
+      "step": 9024
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142461303267725e-06,
+      "loss": 1.3214,
+      "step": 9025
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142365755780624e-06,
+      "loss": 0.9541,
+      "step": 9026
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142270208293522e-06,
+      "loss": 0.7958,
+      "step": 9027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142174660806421e-06,
+      "loss": 1.0622,
+      "step": 9028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.14207911331932e-06,
+      "loss": 1.1725,
+      "step": 9029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141983565832219e-06,
+      "loss": 1.3337,
+      "step": 9030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141888018345117e-06,
+      "loss": 1.4221,
+      "step": 9031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141792470858016e-06,
+      "loss": 1.5266,
+      "step": 9032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141696923370916e-06,
+      "loss": 1.1192,
+      "step": 9033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141601375883814e-06,
+      "loss": 1.1708,
+      "step": 9034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1415058283967134e-06,
+      "loss": 1.3581,
+      "step": 9035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1414102809096125e-06,
+      "loss": 0.7603,
+      "step": 9036
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1413147334225115e-06,
+      "loss": 0.949,
+      "step": 9037
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.14121918593541e-06,
+      "loss": 0.8693,
+      "step": 9038
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.141123638448309e-06,
+      "loss": 1.2036,
+      "step": 9039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.141028090961209e-06,
+      "loss": 1.0372,
+      "step": 9040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140932543474107e-06,
+      "loss": 1.3583,
+      "step": 9041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140836995987006e-06,
+      "loss": 1.0441,
+      "step": 9042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140741448499905e-06,
+      "loss": 0.667,
+      "step": 9043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140645901012804e-06,
+      "loss": 1.7863,
+      "step": 9044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140550353525702e-06,
+      "loss": 1.0515,
+      "step": 9045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140454806038601e-06,
+      "loss": 0.8597,
+      "step": 9046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140359258551501e-06,
+      "loss": 1.4402,
+      "step": 9047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140263711064399e-06,
+      "loss": 1.2694,
+      "step": 9048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140168163577298e-06,
+      "loss": 1.515,
+      "step": 9049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.140072616090197e-06,
+      "loss": 0.9088,
+      "step": 9050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139977068603096e-06,
+      "loss": 1.1285,
+      "step": 9051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1398815211159945e-06,
+      "loss": 1.2528,
+      "step": 9052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139785973628894e-06,
+      "loss": 0.9174,
+      "step": 9053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1396904261417935e-06,
+      "loss": 1.042,
+      "step": 9054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139594878654692e-06,
+      "loss": 1.3031,
+      "step": 9055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139499331167591e-06,
+      "loss": 0.6444,
+      "step": 9056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.13940378368049e-06,
+      "loss": 1.019,
+      "step": 9057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139308236193389e-06,
+      "loss": 0.8793,
+      "step": 9058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139212688706287e-06,
+      "loss": 0.9736,
+      "step": 9059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139117141219186e-06,
+      "loss": 1.2187,
+      "step": 9060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.139021593732085e-06,
+      "loss": 1.0862,
+      "step": 9061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138926046244984e-06,
+      "loss": 1.7637,
+      "step": 9062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138830498757883e-06,
+      "loss": 1.2938,
+      "step": 9063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138734951270782e-06,
+      "loss": 1.0397,
+      "step": 9064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138639403783681e-06,
+      "loss": 1.2241,
+      "step": 9065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138543856296579e-06,
+      "loss": 1.1041,
+      "step": 9066
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1384483088094784e-06,
+      "loss": 1.1998,
+      "step": 9067
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1383527613223775e-06,
+      "loss": 1.0041,
+      "step": 9068
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1382572138352765e-06,
+      "loss": 0.9785,
+      "step": 9069
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1381616663481755e-06,
+      "loss": 1.5968,
+      "step": 9070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.138066118861075e-06,
+      "loss": 1.0447,
+      "step": 9071
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137970571373973e-06,
+      "loss": 1.2099,
+      "step": 9072
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137875023886872e-06,
+      "loss": 1.5425,
+      "step": 9073
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137779476399771e-06,
+      "loss": 1.244,
+      "step": 9074
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.13768392891267e-06,
+      "loss": 1.4761,
+      "step": 9075
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137588381425569e-06,
+      "loss": 1.31,
+      "step": 9076
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137492833938468e-06,
+      "loss": 0.878,
+      "step": 9077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137397286451367e-06,
+      "loss": 1.5714,
+      "step": 9078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137301738964265e-06,
+      "loss": 1.3661,
+      "step": 9079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137206191477164e-06,
+      "loss": 1.1282,
+      "step": 9080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137110643990063e-06,
+      "loss": 0.6392,
+      "step": 9081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137015096502962e-06,
+      "loss": 0.8831,
+      "step": 9082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136919549015861e-06,
+      "loss": 1.0898,
+      "step": 9083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.13682400152876e-06,
+      "loss": 1.0327,
+      "step": 9084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1367284540416594e-06,
+      "loss": 1.4581,
+      "step": 9085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136632906554558e-06,
+      "loss": 1.1765,
+      "step": 9086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136537359067457e-06,
+      "loss": 1.5903,
+      "step": 9087
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.136441811580356e-06,
+      "loss": 1.0822,
+      "step": 9088
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.136346264093255e-06,
+      "loss": 1.1226,
+      "step": 9089
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.136250716606154e-06,
+      "loss": 0.8397,
+      "step": 9090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.136155169119053e-06,
+      "loss": 1.351,
+      "step": 9091
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.136059621631952e-06,
+      "loss": 1.1564,
+      "step": 9092
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13596407414485e-06,
+      "loss": 1.2109,
+      "step": 9093
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135868526657749e-06,
+      "loss": 1.0888,
+      "step": 9094
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135772979170648e-06,
+      "loss": 0.857,
+      "step": 9095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135677431683547e-06,
+      "loss": 1.8157,
+      "step": 9096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135581884196446e-06,
+      "loss": 0.7503,
+      "step": 9097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135486336709345e-06,
+      "loss": 1.4622,
+      "step": 9098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.135390789222244e-06,
+      "loss": 1.1479,
+      "step": 9099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1352952417351425e-06,
+      "loss": 1.3655,
+      "step": 9100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1351996942480415e-06,
+      "loss": 1.0266,
+      "step": 9101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1351041467609405e-06,
+      "loss": 0.9935,
+      "step": 9102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13500859927384e-06,
+      "loss": 1.1774,
+      "step": 9103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134913051786738e-06,
+      "loss": 0.9165,
+      "step": 9104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134817504299638e-06,
+      "loss": 1.1074,
+      "step": 9105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134721956812537e-06,
+      "loss": 1.0085,
+      "step": 9106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134626409325435e-06,
+      "loss": 0.9145,
+      "step": 9107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134530861838334e-06,
+      "loss": 0.9198,
+      "step": 9108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134435314351233e-06,
+      "loss": 0.7613,
+      "step": 9109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134339766864132e-06,
+      "loss": 1.158,
+      "step": 9110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13424421937703e-06,
+      "loss": 1.6491,
+      "step": 9111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13414867188993e-06,
+      "loss": 0.9076,
+      "step": 9112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.134053124402828e-06,
+      "loss": 1.3127,
+      "step": 9113
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133957576915727e-06,
+      "loss": 0.8725,
+      "step": 9114
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133862029428626e-06,
+      "loss": 1.0259,
+      "step": 9115
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133766481941525e-06,
+      "loss": 1.3376,
+      "step": 9116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1336709344544244e-06,
+      "loss": 0.3883,
+      "step": 9117
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133575386967323e-06,
+      "loss": 1.1777,
+      "step": 9118
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1334798394802225e-06,
+      "loss": 1.2003,
+      "step": 9119
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133384291993121e-06,
+      "loss": 0.678,
+      "step": 9120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13328874450602e-06,
+      "loss": 1.3748,
+      "step": 9121
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133193197018919e-06,
+      "loss": 0.9051,
+      "step": 9122
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133097649531818e-06,
+      "loss": 1.1717,
+      "step": 9123
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.133002102044716e-06,
+      "loss": 1.3897,
+      "step": 9124
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132906554557615e-06,
+      "loss": 1.1137,
+      "step": 9125
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132811007070515e-06,
+      "loss": 1.1427,
+      "step": 9126
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132715459583413e-06,
+      "loss": 1.1422,
+      "step": 9127
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132619912096312e-06,
+      "loss": 0.9233,
+      "step": 9128
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132524364609211e-06,
+      "loss": 1.3011,
+      "step": 9129
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13242881712211e-06,
+      "loss": 1.3191,
+      "step": 9130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132333269635008e-06,
+      "loss": 1.2861,
+      "step": 9131
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1322377221479075e-06,
+      "loss": 1.1251,
+      "step": 9132
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.132142174660807e-06,
+      "loss": 0.8097,
+      "step": 9133
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1320466271737055e-06,
+      "loss": 1.1743,
+      "step": 9134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131951079686605e-06,
+      "loss": 1.4356,
+      "step": 9135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131855532199504e-06,
+      "loss": 0.6985,
+      "step": 9136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131759984712403e-06,
+      "loss": 1.4489,
+      "step": 9137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131664437225301e-06,
+      "loss": 1.3161,
+      "step": 9138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1315688897382e-06,
+      "loss": 1.352,
+      "step": 9139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131473342251099e-06,
+      "loss": 1.2083,
+      "step": 9140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131377794763998e-06,
+      "loss": 1.2686,
+      "step": 9141
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131282247276897e-06,
+      "loss": 1.0299,
+      "step": 9142
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131186699789796e-06,
+      "loss": 0.9848,
+      "step": 9143
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131091152302695e-06,
+      "loss": 1.0817,
+      "step": 9144
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130995604815593e-06,
+      "loss": 0.8728,
+      "step": 9145
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130900057328492e-06,
+      "loss": 1.5888,
+      "step": 9146
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130804509841391e-06,
+      "loss": 1.1452,
+      "step": 9147
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.13070896235429e-06,
+      "loss": 1.225,
+      "step": 9148
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130613414867189e-06,
+      "loss": 1.472,
+      "step": 9149
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1305178673800885e-06,
+      "loss": 1.5033,
+      "step": 9150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1304223198929875e-06,
+      "loss": 1.278,
+      "step": 9151
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130326772405886e-06,
+      "loss": 1.2447,
+      "step": 9152
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130231224918785e-06,
+      "loss": 1.2213,
+      "step": 9153
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130135677431684e-06,
+      "loss": 0.7156,
+      "step": 9154
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130040129944583e-06,
+      "loss": 1.7449,
+      "step": 9155
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129944582457482e-06,
+      "loss": 1.1306,
+      "step": 9156
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129849034970381e-06,
+      "loss": 0.7857,
+      "step": 9157
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.12975348748328e-06,
+      "loss": 0.9986,
+      "step": 9158
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129657939996178e-06,
+      "loss": 1.0015,
+      "step": 9159
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129562392509077e-06,
+      "loss": 1.0249,
+      "step": 9160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129466845021976e-06,
+      "loss": 0.9148,
+      "step": 9161
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129371297534875e-06,
+      "loss": 0.9198,
+      "step": 9162
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129275750047774e-06,
+      "loss": 1.2638,
+      "step": 9163
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129180202560673e-06,
+      "loss": 1.0451,
+      "step": 9164
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.129084655073572e-06,
+      "loss": 1.2638,
+      "step": 9165
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1289891075864705e-06,
+      "loss": 1.0902,
+      "step": 9166
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.12889356009937e-06,
+      "loss": 1.1942,
+      "step": 9167
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128798012612269e-06,
+      "loss": 1.2364,
+      "step": 9168
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128702465125168e-06,
+      "loss": 1.3402,
+      "step": 9169
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128606917638067e-06,
+      "loss": 1.1473,
+      "step": 9170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128511370150966e-06,
+      "loss": 1.2355,
+      "step": 9171
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128415822663864e-06,
+      "loss": 1.1892,
+      "step": 9172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128320275176763e-06,
+      "loss": 1.0379,
+      "step": 9173
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128224727689662e-06,
+      "loss": 1.311,
+      "step": 9174
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128129180202561e-06,
+      "loss": 1.5499,
+      "step": 9175
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.128033632715459e-06,
+      "loss": 1.0804,
+      "step": 9176
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127938085228359e-06,
+      "loss": 0.9166,
+      "step": 9177
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127842537741258e-06,
+      "loss": 0.7685,
+      "step": 9178
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127746990254156e-06,
+      "loss": 1.0454,
+      "step": 9179
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127651442767055e-06,
+      "loss": 1.4076,
+      "step": 9180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127555895279954e-06,
+      "loss": 1.1196,
+      "step": 9181
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1274603477928535e-06,
+      "loss": 0.9733,
+      "step": 9182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127364800305752e-06,
+      "loss": 1.4599,
+      "step": 9183
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1272692528186515e-06,
+      "loss": 1.0545,
+      "step": 9184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127173705331551e-06,
+      "loss": 1.4633,
+      "step": 9185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127078157844449e-06,
+      "loss": 1.2487,
+      "step": 9186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126982610357348e-06,
+      "loss": 1.6283,
+      "step": 9187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126887062870247e-06,
+      "loss": 0.9379,
+      "step": 9188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126791515383146e-06,
+      "loss": 1.0715,
+      "step": 9189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126695967896044e-06,
+      "loss": 1.3479,
+      "step": 9190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126600420408944e-06,
+      "loss": 1.2804,
+      "step": 9191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126504872921843e-06,
+      "loss": 1.0854,
+      "step": 9192
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.126409325434741e-06,
+      "loss": 1.129,
+      "step": 9193
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12631377794764e-06,
+      "loss": 0.8385,
+      "step": 9194
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.126218230460539e-06,
+      "loss": 1.287,
+      "step": 9195
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.126122682973438e-06,
+      "loss": 0.9199,
+      "step": 9196
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1260271354863365e-06,
+      "loss": 1.2503,
+      "step": 9197
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125931587999236e-06,
+      "loss": 0.9008,
+      "step": 9198
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125836040512135e-06,
+      "loss": 1.4514,
+      "step": 9199
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125740493025034e-06,
+      "loss": 1.0652,
+      "step": 9200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125644945537933e-06,
+      "loss": 1.3653,
+      "step": 9201
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125549398050832e-06,
+      "loss": 1.3014,
+      "step": 9202
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125453850563731e-06,
+      "loss": 0.9168,
+      "step": 9203
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125358303076629e-06,
+      "loss": 1.0241,
+      "step": 9204
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125262755589529e-06,
+      "loss": 0.8102,
+      "step": 9205
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125167208102428e-06,
+      "loss": 1.0343,
+      "step": 9206
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.125071660615326e-06,
+      "loss": 1.0207,
+      "step": 9207
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124976113128225e-06,
+      "loss": 1.8224,
+      "step": 9208
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124880565641124e-06,
+      "loss": 1.6124,
+      "step": 9209
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124785018154023e-06,
+      "loss": 1.1665,
+      "step": 9210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124689470666921e-06,
+      "loss": 1.2344,
+      "step": 9211
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12459392317982e-06,
+      "loss": 0.8965,
+      "step": 9212
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124498375692719e-06,
+      "loss": 0.7462,
+      "step": 9213
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1244028282056185e-06,
+      "loss": 0.9659,
+      "step": 9214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1243072807185175e-06,
+      "loss": 1.1722,
+      "step": 9215
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1242117332314165e-06,
+      "loss": 1.1933,
+      "step": 9216
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1241161857443156e-06,
+      "loss": 1.1126,
+      "step": 9217
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124020638257214e-06,
+      "loss": 1.0088,
+      "step": 9218
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123925090770113e-06,
+      "loss": 1.0418,
+      "step": 9219
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123829543283012e-06,
+      "loss": 1.0522,
+      "step": 9220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123733995795911e-06,
+      "loss": 0.7385,
+      "step": 9221
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12363844830881e-06,
+      "loss": 1.0842,
+      "step": 9222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123542900821709e-06,
+      "loss": 0.9813,
+      "step": 9223
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123447353334607e-06,
+      "loss": 1.0896,
+      "step": 9224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123351805847506e-06,
+      "loss": 1.1195,
+      "step": 9225
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123256258360405e-06,
+      "loss": 1.5172,
+      "step": 9226
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123160710873304e-06,
+      "loss": 1.3384,
+      "step": 9227
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.123065163386203e-06,
+      "loss": 1.3755,
+      "step": 9228
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122969615899102e-06,
+      "loss": 1.5301,
+      "step": 9229
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122874068412001e-06,
+      "loss": 1.1021,
+      "step": 9230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1227785209248996e-06,
+      "loss": 1.1436,
+      "step": 9231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122682973437799e-06,
+      "loss": 1.2055,
+      "step": 9232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122587425950698e-06,
+      "loss": 1.1045,
+      "step": 9233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122491878463597e-06,
+      "loss": 0.986,
+      "step": 9234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122396330976496e-06,
+      "loss": 0.7614,
+      "step": 9235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122300783489395e-06,
+      "loss": 1.0313,
+      "step": 9236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122205236002294e-06,
+      "loss": 1.1772,
+      "step": 9237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122109688515192e-06,
+      "loss": 0.9631,
+      "step": 9238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122014141028091e-06,
+      "loss": 1.5033,
+      "step": 9239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.12191859354099e-06,
+      "loss": 0.7169,
+      "step": 9240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121823046053889e-06,
+      "loss": 0.7414,
+      "step": 9241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121727498566788e-06,
+      "loss": 1.259,
+      "step": 9242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121631951079687e-06,
+      "loss": 0.9115,
+      "step": 9243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121536403592586e-06,
+      "loss": 1.521,
+      "step": 9244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121440856105484e-06,
+      "loss": 1.1043,
+      "step": 9245
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1213453086183835e-06,
+      "loss": 1.0024,
+      "step": 9246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1212497611312825e-06,
+      "loss": 1.319,
+      "step": 9247
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1211542136441815e-06,
+      "loss": 0.6067,
+      "step": 9248
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1210586661570806e-06,
+      "loss": 1.05,
+      "step": 9249
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.12096311866998e-06,
+      "loss": 1.3374,
+      "step": 9250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120867571182879e-06,
+      "loss": 1.1771,
+      "step": 9251
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120772023695777e-06,
+      "loss": 1.4067,
+      "step": 9252
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120676476208676e-06,
+      "loss": 0.8095,
+      "step": 9253
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120580928721575e-06,
+      "loss": 1.0899,
+      "step": 9254
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120485381234474e-06,
+      "loss": 1.0516,
+      "step": 9255
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120389833747373e-06,
+      "loss": 0.8595,
+      "step": 9256
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120294286260272e-06,
+      "loss": 0.8579,
+      "step": 9257
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120198738773171e-06,
+      "loss": 0.9906,
+      "step": 9258
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120103191286069e-06,
+      "loss": 1.1326,
+      "step": 9259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.120007643798968e-06,
+      "loss": 1.0926,
+      "step": 9260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119912096311867e-06,
+      "loss": 1.0754,
+      "step": 9261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119816548824766e-06,
+      "loss": 1.3302,
+      "step": 9262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119721001337665e-06,
+      "loss": 1.0651,
+      "step": 9263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1196254538505645e-06,
+      "loss": 0.9728,
+      "step": 9264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119529906363463e-06,
+      "loss": 1.139,
+      "step": 9265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119434358876362e-06,
+      "loss": 0.6761,
+      "step": 9266
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119338811389261e-06,
+      "loss": 1.4361,
+      "step": 9267
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11924326390216e-06,
+      "loss": 1.1469,
+      "step": 9268
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119147716415059e-06,
+      "loss": 1.1132,
+      "step": 9269
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.119052168927958e-06,
+      "loss": 0.8905,
+      "step": 9270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118956621440857e-06,
+      "loss": 1.3529,
+      "step": 9271
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118861073953755e-06,
+      "loss": 0.9908,
+      "step": 9272
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118765526466654e-06,
+      "loss": 1.1192,
+      "step": 9273
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118669978979553e-06,
+      "loss": 1.1074,
+      "step": 9274
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118574431492452e-06,
+      "loss": 1.5157,
+      "step": 9275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11847888400535e-06,
+      "loss": 0.877,
+      "step": 9276
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11838333651825e-06,
+      "loss": 0.9037,
+      "step": 9277
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.118287789031149e-06,
+      "loss": 0.9429,
+      "step": 9278
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1181922415440475e-06,
+      "loss": 1.3437,
+      "step": 9279
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1180966940569465e-06,
+      "loss": 0.9341,
+      "step": 9280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1180011465698456e-06,
+      "loss": 0.747,
+      "step": 9281
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117905599082745e-06,
+      "loss": 0.9518,
+      "step": 9282
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117810051595643e-06,
+      "loss": 0.7781,
+      "step": 9283
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117714504108542e-06,
+      "loss": 1.5657,
+      "step": 9284
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117618956621442e-06,
+      "loss": 1.0917,
+      "step": 9285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11752340913434e-06,
+      "loss": 1.1569,
+      "step": 9286
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117427861647239e-06,
+      "loss": 1.1152,
+      "step": 9287
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117332314160138e-06,
+      "loss": 1.0855,
+      "step": 9288
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117236766673037e-06,
+      "loss": 1.3962,
+      "step": 9289
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117141219185935e-06,
+      "loss": 0.8781,
+      "step": 9290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117045671698834e-06,
+      "loss": 1.0288,
+      "step": 9291
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116950124211734e-06,
+      "loss": 1.0502,
+      "step": 9292
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116854576724632e-06,
+      "loss": 0.7425,
+      "step": 9293
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116759029237531e-06,
+      "loss": 1.1615,
+      "step": 9294
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11666348175043e-06,
+      "loss": 1.0444,
+      "step": 9295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1165679342633295e-06,
+      "loss": 0.76,
+      "step": 9296
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.116472386776228e-06,
+      "loss": 0.8716,
+      "step": 9297
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.116376839289127e-06,
+      "loss": 1.304,
+      "step": 9298
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1162812918020266e-06,
+      "loss": 1.0843,
+      "step": 9299
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.116185744314925e-06,
+      "loss": 1.7146,
+      "step": 9300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.116090196827824e-06,
+      "loss": 1.0242,
+      "step": 9301
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115994649340723e-06,
+      "loss": 1.0584,
+      "step": 9302
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115899101853622e-06,
+      "loss": 1.3616,
+      "step": 9303
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11580355436652e-06,
+      "loss": 0.5897,
+      "step": 9304
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115708006879419e-06,
+      "loss": 0.8652,
+      "step": 9305
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115612459392318e-06,
+      "loss": 1.3025,
+      "step": 9306
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115516911905217e-06,
+      "loss": 1.0841,
+      "step": 9307
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115421364418116e-06,
+      "loss": 0.9762,
+      "step": 9308
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115325816931015e-06,
+      "loss": 1.4294,
+      "step": 9309
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.115230269443914e-06,
+      "loss": 0.8639,
+      "step": 9310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1151347219568125e-06,
+      "loss": 1.2142,
+      "step": 9311
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1150391744697115e-06,
+      "loss": 1.2132,
+      "step": 9312
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1149436269826106e-06,
+      "loss": 0.8499,
+      "step": 9313
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11484807949551e-06,
+      "loss": 1.4914,
+      "step": 9314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114752532008409e-06,
+      "loss": 0.5441,
+      "step": 9315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114656984521308e-06,
+      "loss": 1.2903,
+      "step": 9316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114561437034206e-06,
+      "loss": 0.8763,
+      "step": 9317
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114465889547105e-06,
+      "loss": 1.5771,
+      "step": 9318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114370342060004e-06,
+      "loss": 1.058,
+      "step": 9319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114274794572903e-06,
+      "loss": 1.1919,
+      "step": 9320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114179247085802e-06,
+      "loss": 1.0398,
+      "step": 9321
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.114083699598701e-06,
+      "loss": 0.7885,
+      "step": 9322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1139881521116e-06,
+      "loss": 1.2221,
+      "step": 9323
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113892604624498e-06,
+      "loss": 1.2954,
+      "step": 9324
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113797057137397e-06,
+      "loss": 1.0197,
+      "step": 9325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113701509650296e-06,
+      "loss": 1.2103,
+      "step": 9326
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113605962163195e-06,
+      "loss": 1.033,
+      "step": 9327
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1135104146760945e-06,
+      "loss": 1.1738,
+      "step": 9328
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1134148671889935e-06,
+      "loss": 1.1207,
+      "step": 9329
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1133193197018925e-06,
+      "loss": 1.48,
+      "step": 9330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113223772214791e-06,
+      "loss": 1.1739,
+      "step": 9331
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11312822472769e-06,
+      "loss": 0.8317,
+      "step": 9332
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.113032677240589e-06,
+      "loss": 1.2524,
+      "step": 9333
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112937129753488e-06,
+      "loss": 0.9675,
+      "step": 9334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112841582266387e-06,
+      "loss": 1.0902,
+      "step": 9335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112746034779286e-06,
+      "loss": 1.22,
+      "step": 9336
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112650487292185e-06,
+      "loss": 0.8514,
+      "step": 9337
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112554939805083e-06,
+      "loss": 1.2182,
+      "step": 9338
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112459392317982e-06,
+      "loss": 1.2057,
+      "step": 9339
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112363844830881e-06,
+      "loss": 1.4094,
+      "step": 9340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11226829734378e-06,
+      "loss": 0.6664,
+      "step": 9341
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112172749856679e-06,
+      "loss": 1.4411,
+      "step": 9342
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.112077202369578e-06,
+      "loss": 1.3962,
+      "step": 9343
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111981654882477e-06,
+      "loss": 1.3102,
+      "step": 9344
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1118861073953756e-06,
+      "loss": 1.2712,
+      "step": 9345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111790559908275e-06,
+      "loss": 1.3011,
+      "step": 9346
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111695012421174e-06,
+      "loss": 0.9427,
+      "step": 9347
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111599464934073e-06,
+      "loss": 1.0342,
+      "step": 9348
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111503917446972e-06,
+      "loss": 1.4154,
+      "step": 9349
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.111408369959871e-06,
+      "loss": 1.4866,
+      "step": 9350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.11131282247277e-06,
+      "loss": 1.2797,
+      "step": 9351
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.111217274985668e-06,
+      "loss": 0.9612,
+      "step": 9352
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.111121727498567e-06,
+      "loss": 1.442,
+      "step": 9353
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.111026180011466e-06,
+      "loss": 1.6355,
+      "step": 9354
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110930632524365e-06,
+      "loss": 1.0969,
+      "step": 9355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110835085037263e-06,
+      "loss": 1.1084,
+      "step": 9356
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110739537550163e-06,
+      "loss": 1.1393,
+      "step": 9357
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110643990063062e-06,
+      "loss": 0.8416,
+      "step": 9358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.11054844257596e-06,
+      "loss": 1.2691,
+      "step": 9359
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1104528950888595e-06,
+      "loss": 0.891,
+      "step": 9360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1103573476017585e-06,
+      "loss": 1.5468,
+      "step": 9361
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1102618001146575e-06,
+      "loss": 1.4115,
+      "step": 9362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110166252627556e-06,
+      "loss": 1.311,
+      "step": 9363
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.110070705140456e-06,
+      "loss": 1.242,
+      "step": 9364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109975157653354e-06,
+      "loss": 1.2047,
+      "step": 9365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109879610166253e-06,
+      "loss": 1.2605,
+      "step": 9366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109784062679152e-06,
+      "loss": 1.0394,
+      "step": 9367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109688515192051e-06,
+      "loss": 1.0239,
+      "step": 9368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109592967704949e-06,
+      "loss": 1.0247,
+      "step": 9369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109497420217848e-06,
+      "loss": 1.1978,
+      "step": 9370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109401872730748e-06,
+      "loss": 1.2503,
+      "step": 9371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109306325243646e-06,
+      "loss": 1.3379,
+      "step": 9372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109210777756545e-06,
+      "loss": 0.7101,
+      "step": 9373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109115230269444e-06,
+      "loss": 1.5828,
+      "step": 9374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.109019682782343e-06,
+      "loss": 1.1599,
+      "step": 9375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1089241352952415e-06,
+      "loss": 1.0735,
+      "step": 9376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1088285878081406e-06,
+      "loss": 1.2816,
+      "step": 9377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1087330403210405e-06,
+      "loss": 1.0155,
+      "step": 9378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108637492833939e-06,
+      "loss": 0.7211,
+      "step": 9379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108541945346838e-06,
+      "loss": 0.7532,
+      "step": 9380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108446397859737e-06,
+      "loss": 0.8304,
+      "step": 9381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108350850372636e-06,
+      "loss": 0.8911,
+      "step": 9382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108255302885534e-06,
+      "loss": 1.2267,
+      "step": 9383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108159755398433e-06,
+      "loss": 1.1626,
+      "step": 9384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.108064207911333e-06,
+      "loss": 0.7652,
+      "step": 9385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107968660424231e-06,
+      "loss": 1.3832,
+      "step": 9386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.10787311293713e-06,
+      "loss": 1.2609,
+      "step": 9387
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107777565450029e-06,
+      "loss": 0.9491,
+      "step": 9388
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107682017962928e-06,
+      "loss": 1.0556,
+      "step": 9389
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107586470475826e-06,
+      "loss": 0.9314,
+      "step": 9390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107490922988725e-06,
+      "loss": 1.2455,
+      "step": 9391
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1073953755016245e-06,
+      "loss": 0.8836,
+      "step": 9392
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1072998280145235e-06,
+      "loss": 0.8884,
+      "step": 9393
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1072042805274225e-06,
+      "loss": 1.3746,
+      "step": 9394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1071087330403216e-06,
+      "loss": 1.3591,
+      "step": 9395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107013185553221e-06,
+      "loss": 1.4537,
+      "step": 9396
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106917638066119e-06,
+      "loss": 1.2131,
+      "step": 9397
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106822090579018e-06,
+      "loss": 0.9936,
+      "step": 9398
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106726543091917e-06,
+      "loss": 0.7641,
+      "step": 9399
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106630995604816e-06,
+      "loss": 1.0449,
+      "step": 9400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106535448117715e-06,
+      "loss": 1.1431,
+      "step": 9401
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106439900630614e-06,
+      "loss": 1.1201,
+      "step": 9402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.106344353143513e-06,
+      "loss": 1.0421,
+      "step": 9403
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.106248805656411e-06,
+      "loss": 1.4588,
+      "step": 9404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10615325816931e-06,
+      "loss": 0.8468,
+      "step": 9405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.106057710682209e-06,
+      "loss": 1.1074,
+      "step": 9406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105962163195108e-06,
+      "loss": 0.9832,
+      "step": 9407
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105866615708007e-06,
+      "loss": 0.8594,
+      "step": 9408
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105771068220906e-06,
+      "loss": 0.9893,
+      "step": 9409
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1056755207338055e-06,
+      "loss": 1.1408,
+      "step": 9410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105579973246704e-06,
+      "loss": 0.7505,
+      "step": 9411
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105484425759603e-06,
+      "loss": 1.3539,
+      "step": 9412
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105388878272502e-06,
+      "loss": 1.523,
+      "step": 9413
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105293330785401e-06,
+      "loss": 1.0028,
+      "step": 9414
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1051977832983e-06,
+      "loss": 1.52,
+      "step": 9415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105102235811199e-06,
+      "loss": 0.5132,
+      "step": 9416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.105006688324097e-06,
+      "loss": 1.4106,
+      "step": 9417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104911140836996e-06,
+      "loss": 1.4346,
+      "step": 9418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104815593349895e-06,
+      "loss": 1.4146,
+      "step": 9419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104720045862794e-06,
+      "loss": 1.2082,
+      "step": 9420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104624498375693e-06,
+      "loss": 1.0863,
+      "step": 9421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104528950888592e-06,
+      "loss": 1.4825,
+      "step": 9422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.104433403401491e-06,
+      "loss": 0.9777,
+      "step": 9423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1043378559143895e-06,
+      "loss": 1.3852,
+      "step": 9424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1042423084272885e-06,
+      "loss": 1.27,
+      "step": 9425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1041467609401875e-06,
+      "loss": 1.1455,
+      "step": 9426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1040512134530866e-06,
+      "loss": 1.5097,
+      "step": 9427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103955665965985e-06,
+      "loss": 0.8935,
+      "step": 9428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103860118478885e-06,
+      "loss": 0.933,
+      "step": 9429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103764570991784e-06,
+      "loss": 0.8311,
+      "step": 9430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103669023504682e-06,
+      "loss": 1.3396,
+      "step": 9431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103573476017581e-06,
+      "loss": 1.2644,
+      "step": 9432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10347792853048e-06,
+      "loss": 0.911,
+      "step": 9433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103382381043379e-06,
+      "loss": 1.1751,
+      "step": 9434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103286833556277e-06,
+      "loss": 1.1051,
+      "step": 9435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103191286069177e-06,
+      "loss": 1.0221,
+      "step": 9436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103095738582076e-06,
+      "loss": 1.4311,
+      "step": 9437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.103000191094974e-06,
+      "loss": 1.5466,
+      "step": 9438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102904643607873e-06,
+      "loss": 1.1694,
+      "step": 9439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102809096120772e-06,
+      "loss": 1.2353,
+      "step": 9440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102713548633671e-06,
+      "loss": 0.955,
+      "step": 9441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.10261800114657e-06,
+      "loss": 1.0057,
+      "step": 9442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1025224536594695e-06,
+      "loss": 1.6957,
+      "step": 9443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1024269061723685e-06,
+      "loss": 1.5995,
+      "step": 9444
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102331358685267e-06,
+      "loss": 0.85,
+      "step": 9445
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102235811198166e-06,
+      "loss": 0.9504,
+      "step": 9446
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102140263711065e-06,
+      "loss": 1.1669,
+      "step": 9447
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102044716223964e-06,
+      "loss": 1.1003,
+      "step": 9448
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101949168736862e-06,
+      "loss": 0.9797,
+      "step": 9449
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101853621249762e-06,
+      "loss": 0.9702,
+      "step": 9450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101758073762661e-06,
+      "loss": 1.6634,
+      "step": 9451
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101662526275559e-06,
+      "loss": 1.2227,
+      "step": 9452
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101566978788458e-06,
+      "loss": 1.5086,
+      "step": 9453
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101471431301357e-06,
+      "loss": 0.8739,
+      "step": 9454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.101375883814256e-06,
+      "loss": 0.9918,
+      "step": 9455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1012803363271545e-06,
+      "loss": 1.0102,
+      "step": 9456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.101184788840054e-06,
+      "loss": 1.2633,
+      "step": 9457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1010892413529525e-06,
+      "loss": 0.8992,
+      "step": 9458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1009936938658516e-06,
+      "loss": 0.9798,
+      "step": 9459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100898146378751e-06,
+      "loss": 0.7729,
+      "step": 9460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.10080259889165e-06,
+      "loss": 1.1825,
+      "step": 9461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100707051404549e-06,
+      "loss": 1.5702,
+      "step": 9462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100611503917447e-06,
+      "loss": 1.5171,
+      "step": 9463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100515956430346e-06,
+      "loss": 0.915,
+      "step": 9464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100420408943245e-06,
+      "loss": 1.299,
+      "step": 9465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100324861456144e-06,
+      "loss": 1.2128,
+      "step": 9466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100229313969043e-06,
+      "loss": 1.2658,
+      "step": 9467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.100133766481942e-06,
+      "loss": 1.2877,
+      "step": 9468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.10003821899484e-06,
+      "loss": 1.1276,
+      "step": 9469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099942671507739e-06,
+      "loss": 0.9119,
+      "step": 9470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099847124020638e-06,
+      "loss": 1.1196,
+      "step": 9471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099751576533537e-06,
+      "loss": 0.7991,
+      "step": 9472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099656029046436e-06,
+      "loss": 1.5499,
+      "step": 9473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0995604815593355e-06,
+      "loss": 1.0206,
+      "step": 9474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0994649340722345e-06,
+      "loss": 0.8049,
+      "step": 9475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099369386585133e-06,
+      "loss": 1.4965,
+      "step": 9476
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099273839098032e-06,
+      "loss": 0.9077,
+      "step": 9477
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.099178291610931e-06,
+      "loss": 1.096,
+      "step": 9478
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09908274412383e-06,
+      "loss": 1.5402,
+      "step": 9479
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098987196636729e-06,
+      "loss": 1.2743,
+      "step": 9480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098891649149628e-06,
+      "loss": 1.2655,
+      "step": 9481
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098796101662527e-06,
+      "loss": 1.2269,
+      "step": 9482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098700554175425e-06,
+      "loss": 1.198,
+      "step": 9483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098605006688324e-06,
+      "loss": 1.5308,
+      "step": 9484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098509459201223e-06,
+      "loss": 1.4113,
+      "step": 9485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098413911714122e-06,
+      "loss": 0.8949,
+      "step": 9486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098318364227021e-06,
+      "loss": 1.0307,
+      "step": 9487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09822281673992e-06,
+      "loss": 1.1019,
+      "step": 9488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.098127269252819e-06,
+      "loss": 1.0144,
+      "step": 9489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0980317217657175e-06,
+      "loss": 1.5942,
+      "step": 9490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0979361742786166e-06,
+      "loss": 1.0853,
+      "step": 9491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097840626791516e-06,
+      "loss": 1.3463,
+      "step": 9492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097745079304415e-06,
+      "loss": 1.2975,
+      "step": 9493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097649531817314e-06,
+      "loss": 0.8835,
+      "step": 9494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097553984330213e-06,
+      "loss": 1.0833,
+      "step": 9495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097458436843112e-06,
+      "loss": 0.7255,
+      "step": 9496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09736288935601e-06,
+      "loss": 1.1161,
+      "step": 9497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097267341868909e-06,
+      "loss": 1.3544,
+      "step": 9498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097171794381808e-06,
+      "loss": 0.8139,
+      "step": 9499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097076246894707e-06,
+      "loss": 1.0834,
+      "step": 9500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.096980699407606e-06,
+      "loss": 1.2436,
+      "step": 9501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.096885151920505e-06,
+      "loss": 1.1555,
+      "step": 9502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.096789604433404e-06,
+      "loss": 1.1206,
+      "step": 9503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.096694056946302e-06,
+      "loss": 1.2639,
+      "step": 9504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.096598509459201e-06,
+      "loss": 1.1044,
+      "step": 9505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0965029619721005e-06,
+      "loss": 0.9795,
+      "step": 9506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0964074144849995e-06,
+      "loss": 0.9645,
+      "step": 9507
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0963118669978985e-06,
+      "loss": 1.3331,
+      "step": 9508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0962163195107976e-06,
+      "loss": 1.296,
+      "step": 9509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.096120772023696e-06,
+      "loss": 0.9245,
+      "step": 9510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.096025224536595e-06,
+      "loss": 1.084,
+      "step": 9511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095929677049494e-06,
+      "loss": 1.2865,
+      "step": 9512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095834129562393e-06,
+      "loss": 1.5565,
+      "step": 9513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095738582075292e-06,
+      "loss": 1.4973,
+      "step": 9514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095643034588191e-06,
+      "loss": 1.3443,
+      "step": 9515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09554748710109e-06,
+      "loss": 1.2766,
+      "step": 9516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095451939613988e-06,
+      "loss": 1.2061,
+      "step": 9517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095356392126887e-06,
+      "loss": 0.8536,
+      "step": 9518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095260844639786e-06,
+      "loss": 1.3313,
+      "step": 9519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.095165297152685e-06,
+      "loss": 0.8957,
+      "step": 9520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0950697496655835e-06,
+      "loss": 1.1495,
+      "step": 9521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094974202178483e-06,
+      "loss": 0.5986,
+      "step": 9522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094878654691382e-06,
+      "loss": 1.192,
+      "step": 9523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094783107204281e-06,
+      "loss": 1.3142,
+      "step": 9524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09468755971718e-06,
+      "loss": 0.7193,
+      "step": 9525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094592012230079e-06,
+      "loss": 1.0208,
+      "step": 9526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094496464742978e-06,
+      "loss": 0.9933,
+      "step": 9527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094400917255876e-06,
+      "loss": 1.5908,
+      "step": 9528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094305369768776e-06,
+      "loss": 1.3389,
+      "step": 9529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094209822281675e-06,
+      "loss": 1.0584,
+      "step": 9530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094114274794573e-06,
+      "loss": 1.3089,
+      "step": 9531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094018727307472e-06,
+      "loss": 1.6857,
+      "step": 9532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.093923179820371e-06,
+      "loss": 1.1086,
+      "step": 9533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09382763233327e-06,
+      "loss": 1.664,
+      "step": 9534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.093732084846168e-06,
+      "loss": 0.995,
+      "step": 9535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.093636537359067e-06,
+      "loss": 0.8966,
+      "step": 9536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.093540989871967e-06,
+      "loss": 1.065,
+      "step": 9537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0934454423848655e-06,
+      "loss": 1.5859,
+      "step": 9538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0933498948977645e-06,
+      "loss": 1.3123,
+      "step": 9539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0932543474106635e-06,
+      "loss": 1.1766,
+      "step": 9540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0931587999235626e-06,
+      "loss": 1.2355,
+      "step": 9541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.093063252436461e-06,
+      "loss": 0.8965,
+      "step": 9542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09296770494936e-06,
+      "loss": 1.0496,
+      "step": 9543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09287215746226e-06,
+      "loss": 1.234,
+      "step": 9544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092776609975158e-06,
+      "loss": 0.9274,
+      "step": 9545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092681062488057e-06,
+      "loss": 1.202,
+      "step": 9546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092585515000956e-06,
+      "loss": 1.1455,
+      "step": 9547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092489967513855e-06,
+      "loss": 1.1097,
+      "step": 9548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092394420026753e-06,
+      "loss": 0.5942,
+      "step": 9549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092298872539652e-06,
+      "loss": 1.0334,
+      "step": 9550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092203325052552e-06,
+      "loss": 1.124,
+      "step": 9551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.09210777756545e-06,
+      "loss": 1.372,
+      "step": 9552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092012230078349e-06,
+      "loss": 1.5927,
+      "step": 9553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091916682591248e-06,
+      "loss": 0.8473,
+      "step": 9554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091821135104147e-06,
+      "loss": 0.9517,
+      "step": 9555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091725587617046e-06,
+      "loss": 0.8661,
+      "step": 9556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091630040129945e-06,
+      "loss": 0.6787,
+      "step": 9557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091534492642844e-06,
+      "loss": 1.2096,
+      "step": 9558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091438945155743e-06,
+      "loss": 0.9046,
+      "step": 9559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.091343397668642e-06,
+      "loss": 1.358,
+      "step": 9560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.091247850181541e-06,
+      "loss": 1.2055,
+      "step": 9561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.091152302694439e-06,
+      "loss": 0.8665,
+      "step": 9562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.091056755207338e-06,
+      "loss": 1.3017,
+      "step": 9563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090961207720237e-06,
+      "loss": 1.1871,
+      "step": 9564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090865660233136e-06,
+      "loss": 0.9502,
+      "step": 9565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090770112746035e-06,
+      "loss": 0.8881,
+      "step": 9566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090674565258934e-06,
+      "loss": 0.9079,
+      "step": 9567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090579017771833e-06,
+      "loss": 1.3034,
+      "step": 9568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090483470284731e-06,
+      "loss": 1.2549,
+      "step": 9569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0903879227976305e-06,
+      "loss": 1.0879,
+      "step": 9570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0902923753105295e-06,
+      "loss": 1.3591,
+      "step": 9571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0901968278234285e-06,
+      "loss": 0.8677,
+      "step": 9572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0901012803363276e-06,
+      "loss": 1.3154,
+      "step": 9573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.090005732849227e-06,
+      "loss": 0.7856,
+      "step": 9574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089910185362126e-06,
+      "loss": 1.053,
+      "step": 9575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089814637875024e-06,
+      "loss": 1.0153,
+      "step": 9576
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089719090387923e-06,
+      "loss": 1.6766,
+      "step": 9577
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089623542900822e-06,
+      "loss": 0.8943,
+      "step": 9578
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089527995413721e-06,
+      "loss": 0.8909,
+      "step": 9579
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.08943244792662e-06,
+      "loss": 1.2498,
+      "step": 9580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089336900439519e-06,
+      "loss": 0.9374,
+      "step": 9581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089241352952418e-06,
+      "loss": 1.1276,
+      "step": 9582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089145805465316e-06,
+      "loss": 0.9823,
+      "step": 9583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.089050257978215e-06,
+      "loss": 0.962,
+      "step": 9584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088954710491114e-06,
+      "loss": 1.0771,
+      "step": 9585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088859163004013e-06,
+      "loss": 1.3264,
+      "step": 9586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088763615516912e-06,
+      "loss": 1.7482,
+      "step": 9587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0886680680298115e-06,
+      "loss": 0.8178,
+      "step": 9588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0885725205427105e-06,
+      "loss": 1.1519,
+      "step": 9589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088476973055609e-06,
+      "loss": 0.9319,
+      "step": 9590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088381425568508e-06,
+      "loss": 1.1372,
+      "step": 9591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088285878081407e-06,
+      "loss": 0.6704,
+      "step": 9592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088190330594306e-06,
+      "loss": 0.9799,
+      "step": 9593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.088094783107205e-06,
+      "loss": 0.8723,
+      "step": 9594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087999235620104e-06,
+      "loss": 1.344,
+      "step": 9595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087903688133003e-06,
+      "loss": 1.1681,
+      "step": 9596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087808140645901e-06,
+      "loss": 0.9731,
+      "step": 9597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0877125931588e-06,
+      "loss": 0.903,
+      "step": 9598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087617045671699e-06,
+      "loss": 1.2968,
+      "step": 9599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087521498184598e-06,
+      "loss": 1.1256,
+      "step": 9600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087425950697497e-06,
+      "loss": 1.1025,
+      "step": 9601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087330403210396e-06,
+      "loss": 0.932,
+      "step": 9602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087234855723295e-06,
+      "loss": 0.6806,
+      "step": 9603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0871393082361935e-06,
+      "loss": 1.1959,
+      "step": 9604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0870437607490926e-06,
+      "loss": 1.3586,
+      "step": 9605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086948213261992e-06,
+      "loss": 1.1423,
+      "step": 9606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086852665774891e-06,
+      "loss": 0.7706,
+      "step": 9607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086757118287789e-06,
+      "loss": 1.4374,
+      "step": 9608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086661570800689e-06,
+      "loss": 0.8688,
+      "step": 9609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086566023313587e-06,
+      "loss": 1.3781,
+      "step": 9610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.086470475826486e-06,
+      "loss": 1.1603,
+      "step": 9611
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.086374928339385e-06,
+      "loss": 1.2073,
+      "step": 9612
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.086279380852284e-06,
+      "loss": 1.531,
+      "step": 9613
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.086183833365183e-06,
+      "loss": 1.5458,
+      "step": 9614
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.086088285878081e-06,
+      "loss": 0.8553,
+      "step": 9615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085992738390981e-06,
+      "loss": 1.2235,
+      "step": 9616
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085897190903879e-06,
+      "loss": 1.2993,
+      "step": 9617
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085801643416778e-06,
+      "loss": 1.4765,
+      "step": 9618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085706095929677e-06,
+      "loss": 1.4886,
+      "step": 9619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0856105484425765e-06,
+      "loss": 1.5605,
+      "step": 9620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085515000955475e-06,
+      "loss": 1.2709,
+      "step": 9621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085419453468374e-06,
+      "loss": 1.0462,
+      "step": 9622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0853239059812736e-06,
+      "loss": 1.5052,
+      "step": 9623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085228358494172e-06,
+      "loss": 1.0786,
+      "step": 9624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.085132811007071e-06,
+      "loss": 1.2053,
+      "step": 9625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08503726351997e-06,
+      "loss": 0.969,
+      "step": 9626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084941716032869e-06,
+      "loss": 1.0839,
+      "step": 9627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084846168545767e-06,
+      "loss": 1.3119,
+      "step": 9628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084750621058666e-06,
+      "loss": 0.943,
+      "step": 9629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084655073571566e-06,
+      "loss": 1.1248,
+      "step": 9630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084559526084464e-06,
+      "loss": 1.2004,
+      "step": 9631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084463978597363e-06,
+      "loss": 1.3062,
+      "step": 9632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084368431110262e-06,
+      "loss": 0.6165,
+      "step": 9633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.084272883623161e-06,
+      "loss": 0.9054,
+      "step": 9634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0841773361360595e-06,
+      "loss": 1.3843,
+      "step": 9635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0840817886489585e-06,
+      "loss": 1.1561,
+      "step": 9636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083986241161858e-06,
+      "loss": 1.1207,
+      "step": 9637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083890693674757e-06,
+      "loss": 1.0385,
+      "step": 9638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083795146187656e-06,
+      "loss": 1.1728,
+      "step": 9639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083699598700555e-06,
+      "loss": 0.8025,
+      "step": 9640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083604051213454e-06,
+      "loss": 1.3598,
+      "step": 9641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083508503726352e-06,
+      "loss": 1.1546,
+      "step": 9642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083412956239251e-06,
+      "loss": 1.3107,
+      "step": 9643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08331740875215e-06,
+      "loss": 1.4042,
+      "step": 9644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083221861265049e-06,
+      "loss": 1.0802,
+      "step": 9645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083126313777948e-06,
+      "loss": 1.2675,
+      "step": 9646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.083030766290847e-06,
+      "loss": 1.0687,
+      "step": 9647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082935218803746e-06,
+      "loss": 1.2811,
+      "step": 9648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082839671316644e-06,
+      "loss": 1.014,
+      "step": 9649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082744123829543e-06,
+      "loss": 0.763,
+      "step": 9650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082648576342442e-06,
+      "loss": 0.9068,
+      "step": 9651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0825530288553415e-06,
+      "loss": 1.2901,
+      "step": 9652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0824574813682405e-06,
+      "loss": 0.9035,
+      "step": 9653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0823619338811395e-06,
+      "loss": 1.375,
+      "step": 9654
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0822663863940386e-06,
+      "loss": 1.1396,
+      "step": 9655
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082170838906937e-06,
+      "loss": 1.139,
+      "step": 9656
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.082075291419836e-06,
+      "loss": 1.3924,
+      "step": 9657
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081979743932735e-06,
+      "loss": 1.2811,
+      "step": 9658
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081884196445634e-06,
+      "loss": 1.2564,
+      "step": 9659
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081788648958533e-06,
+      "loss": 1.1513,
+      "step": 9660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081693101471432e-06,
+      "loss": 0.9776,
+      "step": 9661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08159755398433e-06,
+      "loss": 1.142,
+      "step": 9662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081502006497229e-06,
+      "loss": 1.3539,
+      "step": 9663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081406459010128e-06,
+      "loss": 1.1558,
+      "step": 9664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.081310911523027e-06,
+      "loss": 1.0962,
+      "step": 9665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.081215364035926e-06,
+      "loss": 1.0162,
+      "step": 9666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.081119816548825e-06,
+      "loss": 1.226,
+      "step": 9667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.081024269061724e-06,
+      "loss": 1.3383,
+      "step": 9668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0809287215746226e-06,
+      "loss": 0.836,
+      "step": 9669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080833174087522e-06,
+      "loss": 1.0003,
+      "step": 9670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080737626600421e-06,
+      "loss": 0.9473,
+      "step": 9671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.08064207911332e-06,
+      "loss": 1.4998,
+      "step": 9672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080546531626219e-06,
+      "loss": 0.8576,
+      "step": 9673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080450984139118e-06,
+      "loss": 1.1053,
+      "step": 9674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080355436652017e-06,
+      "loss": 1.1873,
+      "step": 9675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080259889164915e-06,
+      "loss": 1.1449,
+      "step": 9676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080164341677814e-06,
+      "loss": 1.191,
+      "step": 9677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.080068794190713e-06,
+      "loss": 0.9111,
+      "step": 9678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079973246703612e-06,
+      "loss": 0.9944,
+      "step": 9679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07987769921651e-06,
+      "loss": 0.9714,
+      "step": 9680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07978215172941e-06,
+      "loss": 1.1564,
+      "step": 9681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079686604242309e-06,
+      "loss": 1.0309,
+      "step": 9682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079591056755207e-06,
+      "loss": 0.9079,
+      "step": 9683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0794955092681065e-06,
+      "loss": 0.7789,
+      "step": 9684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0793999617810055e-06,
+      "loss": 1.2795,
+      "step": 9685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0793044142939045e-06,
+      "loss": 0.996,
+      "step": 9686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079208866806803e-06,
+      "loss": 1.2869,
+      "step": 9687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079113319319703e-06,
+      "loss": 1.446,
+      "step": 9688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.079017771832602e-06,
+      "loss": 1.1563,
+      "step": 9689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0789222243455e-06,
+      "loss": 1.5198,
+      "step": 9690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078826676858399e-06,
+      "loss": 0.7757,
+      "step": 9691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078731129371298e-06,
+      "loss": 0.8174,
+      "step": 9692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078635581884197e-06,
+      "loss": 0.9681,
+      "step": 9693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078540034397095e-06,
+      "loss": 1.2392,
+      "step": 9694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078444486909995e-06,
+      "loss": 1.0171,
+      "step": 9695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078348939422894e-06,
+      "loss": 1.3684,
+      "step": 9696
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078253391935792e-06,
+      "loss": 1.0891,
+      "step": 9697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.078157844448691e-06,
+      "loss": 0.951,
+      "step": 9698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07806229696159e-06,
+      "loss": 1.1333,
+      "step": 9699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077966749474489e-06,
+      "loss": 1.0038,
+      "step": 9700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0778712019873876e-06,
+      "loss": 1.4941,
+      "step": 9701
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0777756545002875e-06,
+      "loss": 0.9609,
+      "step": 9702
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077680107013186e-06,
+      "loss": 1.0548,
+      "step": 9703
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077584559526085e-06,
+      "loss": 0.8653,
+      "step": 9704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077489012038984e-06,
+      "loss": 1.4782,
+      "step": 9705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077393464551883e-06,
+      "loss": 1.0677,
+      "step": 9706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077297917064782e-06,
+      "loss": 0.6452,
+      "step": 9707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07720236957768e-06,
+      "loss": 1.0616,
+      "step": 9708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.07710682209058e-06,
+      "loss": 1.4145,
+      "step": 9709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.077011274603478e-06,
+      "loss": 0.7189,
+      "step": 9710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076915727116377e-06,
+      "loss": 1.4563,
+      "step": 9711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076820179629276e-06,
+      "loss": 1.1705,
+      "step": 9712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076724632142175e-06,
+      "loss": 1.0859,
+      "step": 9713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076629084655073e-06,
+      "loss": 1.2176,
+      "step": 9714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076533537167972e-06,
+      "loss": 0.7034,
+      "step": 9715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0764379896808715e-06,
+      "loss": 1.0554,
+      "step": 9716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0763424421937705e-06,
+      "loss": 1.0825,
+      "step": 9717
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0762468947066695e-06,
+      "loss": 0.8264,
+      "step": 9718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0761513472195686e-06,
+      "loss": 1.3479,
+      "step": 9719
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.076055799732468e-06,
+      "loss": 1.1483,
+      "step": 9720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075960252245366e-06,
+      "loss": 0.9154,
+      "step": 9721
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075864704758265e-06,
+      "loss": 1.0038,
+      "step": 9722
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075769157271164e-06,
+      "loss": 1.084,
+      "step": 9723
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075673609784063e-06,
+      "loss": 0.8641,
+      "step": 9724
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075578062296962e-06,
+      "loss": 1.1414,
+      "step": 9725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075482514809861e-06,
+      "loss": 1.7224,
+      "step": 9726
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07538696732276e-06,
+      "loss": 1.2397,
+      "step": 9727
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075291419835658e-06,
+      "loss": 0.9368,
+      "step": 9728
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075195872348557e-06,
+      "loss": 1.1021,
+      "step": 9729
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075100324861456e-06,
+      "loss": 1.0343,
+      "step": 9730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.075004777374355e-06,
+      "loss": 1.5604,
+      "step": 9731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074909229887254e-06,
+      "loss": 1.2667,
+      "step": 9732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074813682400153e-06,
+      "loss": 1.2751,
+      "step": 9733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0747181349130525e-06,
+      "loss": 0.9584,
+      "step": 9734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074622587425951e-06,
+      "loss": 1.0933,
+      "step": 9735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07452703993885e-06,
+      "loss": 0.9068,
+      "step": 9736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074431492451749e-06,
+      "loss": 1.3925,
+      "step": 9737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074335944964648e-06,
+      "loss": 0.969,
+      "step": 9738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074240397477547e-06,
+      "loss": 1.218,
+      "step": 9739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074144849990446e-06,
+      "loss": 1.324,
+      "step": 9740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.074049302503345e-06,
+      "loss": 1.048,
+      "step": 9741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073953755016243e-06,
+      "loss": 1.1921,
+      "step": 9742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073858207529142e-06,
+      "loss": 1.0252,
+      "step": 9743
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073762660042041e-06,
+      "loss": 0.7305,
+      "step": 9744
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07366711255494e-06,
+      "loss": 1.3137,
+      "step": 9745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073571565067839e-06,
+      "loss": 0.9278,
+      "step": 9746
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073476017580738e-06,
+      "loss": 0.8772,
+      "step": 9747
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.073380470093637e-06,
+      "loss": 0.9101,
+      "step": 9748
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0732849226065355e-06,
+      "loss": 1.4293,
+      "step": 9749
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0731893751194345e-06,
+      "loss": 1.2709,
+      "step": 9750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0730938276323336e-06,
+      "loss": 1.5539,
+      "step": 9751
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072998280145233e-06,
+      "loss": 1.1483,
+      "step": 9752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072902732658132e-06,
+      "loss": 1.663,
+      "step": 9753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072807185171031e-06,
+      "loss": 0.8819,
+      "step": 9754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07271163768393e-06,
+      "loss": 1.2042,
+      "step": 9755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072616090196828e-06,
+      "loss": 0.8724,
+      "step": 9756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072520542709727e-06,
+      "loss": 1.1433,
+      "step": 9757
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072424995222626e-06,
+      "loss": 0.7593,
+      "step": 9758
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072329447735525e-06,
+      "loss": 1.4373,
+      "step": 9759
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072233900248424e-06,
+      "loss": 0.7634,
+      "step": 9760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072138352761323e-06,
+      "loss": 1.2211,
+      "step": 9761
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.072042805274221e-06,
+      "loss": 1.0105,
+      "step": 9762
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07194725778712e-06,
+      "loss": 1.3257,
+      "step": 9763
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.071851710300019e-06,
+      "loss": 1.2198,
+      "step": 9764
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.071756162812918e-06,
+      "loss": 0.6726,
+      "step": 9765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.071660615325817e-06,
+      "loss": 1.2364,
+      "step": 9766
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0715650678387165e-06,
+      "loss": 0.8762,
+      "step": 9767
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0714695203516155e-06,
+      "loss": 1.2923,
+      "step": 9768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.071373972864514e-06,
+      "loss": 1.4304,
+      "step": 9769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.071278425377413e-06,
+      "loss": 1.1744,
+      "step": 9770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.071182877890312e-06,
+      "loss": 1.1166,
+      "step": 9771
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.071087330403211e-06,
+      "loss": 0.8983,
+      "step": 9772
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070991782916109e-06,
+      "loss": 1.1972,
+      "step": 9773
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070896235429009e-06,
+      "loss": 0.5501,
+      "step": 9774
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070800687941908e-06,
+      "loss": 1.7578,
+      "step": 9775
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070705140454806e-06,
+      "loss": 1.5257,
+      "step": 9776
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070609592967705e-06,
+      "loss": 1.7106,
+      "step": 9777
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070514045480604e-06,
+      "loss": 0.8415,
+      "step": 9778
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070418497993503e-06,
+      "loss": 1.1907,
+      "step": 9779
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0703229505064014e-06,
+      "loss": 0.9854,
+      "step": 9780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.070227403019301e-06,
+      "loss": 1.5329,
+      "step": 9781
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0701318555322e-06,
+      "loss": 1.3148,
+      "step": 9782
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0700363080450986e-06,
+      "loss": 1.0095,
+      "step": 9783
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069940760557998e-06,
+      "loss": 0.9487,
+      "step": 9784
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069845213070897e-06,
+      "loss": 1.1126,
+      "step": 9785
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069749665583796e-06,
+      "loss": 0.9092,
+      "step": 9786
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069654118096694e-06,
+      "loss": 0.6939,
+      "step": 9787
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069558570609593e-06,
+      "loss": 0.7253,
+      "step": 9788
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069463023122493e-06,
+      "loss": 1.0914,
+      "step": 9789
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069367475635391e-06,
+      "loss": 1.5276,
+      "step": 9790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06927192814829e-06,
+      "loss": 1.2189,
+      "step": 9791
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069176380661189e-06,
+      "loss": 1.279,
+      "step": 9792
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.069080833174088e-06,
+      "loss": 1.1558,
+      "step": 9793
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068985285686986e-06,
+      "loss": 1.2954,
+      "step": 9794
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068889738199885e-06,
+      "loss": 1.0825,
+      "step": 9795
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068794190712785e-06,
+      "loss": 1.8806,
+      "step": 9796
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068698643225683e-06,
+      "loss": 1.3158,
+      "step": 9797
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0686030957385824e-06,
+      "loss": 1.038,
+      "step": 9798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0685075482514815e-06,
+      "loss": 1.0823,
+      "step": 9799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0684120007643805e-06,
+      "loss": 0.9518,
+      "step": 9800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068316453277279e-06,
+      "loss": 1.1447,
+      "step": 9801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068220905790178e-06,
+      "loss": 0.9635,
+      "step": 9802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068125358303077e-06,
+      "loss": 0.9394,
+      "step": 9803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.068029810815976e-06,
+      "loss": 1.3164,
+      "step": 9804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067934263328875e-06,
+      "loss": 0.7207,
+      "step": 9805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067838715841774e-06,
+      "loss": 1.15,
+      "step": 9806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067743168354673e-06,
+      "loss": 0.7465,
+      "step": 9807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067647620867571e-06,
+      "loss": 1.1801,
+      "step": 9808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06755207338047e-06,
+      "loss": 1.1228,
+      "step": 9809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067456525893369e-06,
+      "loss": 1.1366,
+      "step": 9810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067360978406268e-06,
+      "loss": 1.3742,
+      "step": 9811
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067265430919167e-06,
+      "loss": 1.106,
+      "step": 9812
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067169883432066e-06,
+      "loss": 1.3565,
+      "step": 9813
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0670743359449645e-06,
+      "loss": 0.7822,
+      "step": 9814
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0669787884578636e-06,
+      "loss": 0.7846,
+      "step": 9815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066883240970763e-06,
+      "loss": 0.9451,
+      "step": 9816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066787693483662e-06,
+      "loss": 0.8974,
+      "step": 9817
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066692145996561e-06,
+      "loss": 0.9387,
+      "step": 9818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06659659850946e-06,
+      "loss": 1.3133,
+      "step": 9819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066501051022359e-06,
+      "loss": 1.1124,
+      "step": 9820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.066405503535257e-06,
+      "loss": 1.0206,
+      "step": 9821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066309956048156e-06,
+      "loss": 1.0825,
+      "step": 9822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066214408561055e-06,
+      "loss": 1.2879,
+      "step": 9823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066118861073954e-06,
+      "loss": 1.505,
+      "step": 9824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066023313586853e-06,
+      "loss": 1.157,
+      "step": 9825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065927766099752e-06,
+      "loss": 1.0531,
+      "step": 9826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065832218612651e-06,
+      "loss": 1.0747,
+      "step": 9827
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065736671125549e-06,
+      "loss": 1.2847,
+      "step": 9828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065641123638448e-06,
+      "loss": 1.0444,
+      "step": 9829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0655455761513474e-06,
+      "loss": 1.1725,
+      "step": 9830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0654500286642465e-06,
+      "loss": 1.3157,
+      "step": 9831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0653544811771455e-06,
+      "loss": 1.2911,
+      "step": 9832
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0652589336900446e-06,
+      "loss": 1.252,
+      "step": 9833
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065163386202944e-06,
+      "loss": 0.8655,
+      "step": 9834
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.065067838715842e-06,
+      "loss": 1.0694,
+      "step": 9835
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064972291228741e-06,
+      "loss": 1.0295,
+      "step": 9836
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06487674374164e-06,
+      "loss": 0.7263,
+      "step": 9837
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064781196254539e-06,
+      "loss": 1.2641,
+      "step": 9838
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064685648767438e-06,
+      "loss": 1.0018,
+      "step": 9839
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064590101280337e-06,
+      "loss": 1.3686,
+      "step": 9840
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064494553793236e-06,
+      "loss": 1.2589,
+      "step": 9841
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064399006306134e-06,
+      "loss": 1.3468,
+      "step": 9842
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064303458819033e-06,
+      "loss": 0.9256,
+      "step": 9843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064207911331932e-06,
+      "loss": 1.1055,
+      "step": 9844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.064112363844831e-06,
+      "loss": 1.2095,
+      "step": 9845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06401681635773e-06,
+      "loss": 1.4881,
+      "step": 9846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063921268870629e-06,
+      "loss": 1.3699,
+      "step": 9847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0638257213835284e-06,
+      "loss": 1.0687,
+      "step": 9848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063730173896427e-06,
+      "loss": 0.7284,
+      "step": 9849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063634626409326e-06,
+      "loss": 1.3136,
+      "step": 9850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063539078922225e-06,
+      "loss": 1.0207,
+      "step": 9851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063443531435124e-06,
+      "loss": 0.9174,
+      "step": 9852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063347983948023e-06,
+      "loss": 1.1778,
+      "step": 9853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063252436460922e-06,
+      "loss": 1.0217,
+      "step": 9854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.06315688897382e-06,
+      "loss": 0.9874,
+      "step": 9855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.063061341486719e-06,
+      "loss": 0.9204,
+      "step": 9856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062965793999618e-06,
+      "loss": 0.7397,
+      "step": 9857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062870246512517e-06,
+      "loss": 0.9307,
+      "step": 9858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062774699025416e-06,
+      "loss": 1.1298,
+      "step": 9859
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062679151538314e-06,
+      "loss": 1.1697,
+      "step": 9860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062583604051214e-06,
+      "loss": 1.3388,
+      "step": 9861
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0624880565641124e-06,
+      "loss": 1.6297,
+      "step": 9862
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0623925090770115e-06,
+      "loss": 1.0391,
+      "step": 9863
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0622969615899105e-06,
+      "loss": 0.8873,
+      "step": 9864
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0622014141028096e-06,
+      "loss": 1.179,
+      "step": 9865
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062105866615708e-06,
+      "loss": 0.9414,
+      "step": 9866
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062010319128607e-06,
+      "loss": 1.6052,
+      "step": 9867
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061914771641507e-06,
+      "loss": 0.8153,
+      "step": 9868
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061819224154405e-06,
+      "loss": 1.5258,
+      "step": 9869
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061723676667304e-06,
+      "loss": 1.0881,
+      "step": 9870
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061628129180203e-06,
+      "loss": 1.6345,
+      "step": 9871
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061532581693102e-06,
+      "loss": 1.5311,
+      "step": 9872
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061437034206e-06,
+      "loss": 1.1041,
+      "step": 9873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.061341486718899e-06,
+      "loss": 1.0583,
+      "step": 9874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.061245939231799e-06,
+      "loss": 1.0826,
+      "step": 9875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.061150391744697e-06,
+      "loss": 1.2518,
+      "step": 9876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.061054844257596e-06,
+      "loss": 0.7769,
+      "step": 9877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060959296770495e-06,
+      "loss": 1.1767,
+      "step": 9878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060863749283394e-06,
+      "loss": 0.7857,
+      "step": 9879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060768201796293e-06,
+      "loss": 0.6845,
+      "step": 9880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060672654309192e-06,
+      "loss": 1.2253,
+      "step": 9881
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0605771068220915e-06,
+      "loss": 0.9198,
+      "step": 9882
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.06048155933499e-06,
+      "loss": 1.1171,
+      "step": 9883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060386011847889e-06,
+      "loss": 1.3511,
+      "step": 9884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060290464360788e-06,
+      "loss": 0.9394,
+      "step": 9885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060194916873687e-06,
+      "loss": 0.8502,
+      "step": 9886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060099369386585e-06,
+      "loss": 1.455,
+      "step": 9887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.060003821899484e-06,
+      "loss": 1.0157,
+      "step": 9888
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.059908274412384e-06,
+      "loss": 1.0448,
+      "step": 9889
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.059812726925282e-06,
+      "loss": 0.8782,
+      "step": 9890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.059717179438181e-06,
+      "loss": 1.0152,
+      "step": 9891
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.05962163195108e-06,
+      "loss": 1.5444,
+      "step": 9892
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.059526084463979e-06,
+      "loss": 1.4167,
+      "step": 9893
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0594305369768774e-06,
+      "loss": 1.2263,
+      "step": 9894
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0593349894897765e-06,
+      "loss": 1.3812,
+      "step": 9895
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0592394420026755e-06,
+      "loss": 1.1518,
+      "step": 9896
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0591438945155746e-06,
+      "loss": 1.1554,
+      "step": 9897
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.059048347028474e-06,
+      "loss": 1.0986,
+      "step": 9898
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058952799541373e-06,
+      "loss": 1.8553,
+      "step": 9899
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058857252054272e-06,
+      "loss": 1.072,
+      "step": 9900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.05876170456717e-06,
+      "loss": 1.4816,
+      "step": 9901
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058666157080069e-06,
+      "loss": 0.8569,
+      "step": 9902
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058570609592968e-06,
+      "loss": 1.521,
+      "step": 9903
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058475062105867e-06,
+      "loss": 0.9013,
+      "step": 9904
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058379514618766e-06,
+      "loss": 1.47,
+      "step": 9905
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058283967131665e-06,
+      "loss": 1.0333,
+      "step": 9906
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058188419644563e-06,
+      "loss": 0.6978,
+      "step": 9907
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.058092872157462e-06,
+      "loss": 0.9837,
+      "step": 9908
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057997324670361e-06,
+      "loss": 0.9374,
+      "step": 9909
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.05790177718326e-06,
+      "loss": 0.9369,
+      "step": 9910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057806229696159e-06,
+      "loss": 1.0979,
+      "step": 9911
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0577106822090584e-06,
+      "loss": 0.9626,
+      "step": 9912
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0576151347219575e-06,
+      "loss": 1.3164,
+      "step": 9913
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057519587234856e-06,
+      "loss": 1.5465,
+      "step": 9914
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057424039747755e-06,
+      "loss": 1.2177,
+      "step": 9915
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057328492260654e-06,
+      "loss": 1.343,
+      "step": 9916
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057232944773553e-06,
+      "loss": 0.9579,
+      "step": 9917
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057137397286452e-06,
+      "loss": 1.38,
+      "step": 9918
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.057041849799351e-06,
+      "loss": 0.9168,
+      "step": 9919
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.05694630231225e-06,
+      "loss": 1.6034,
+      "step": 9920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056850754825148e-06,
+      "loss": 1.4746,
+      "step": 9921
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056755207338047e-06,
+      "loss": 1.1452,
+      "step": 9922
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056659659850946e-06,
+      "loss": 1.2861,
+      "step": 9923
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056564112363845e-06,
+      "loss": 1.4459,
+      "step": 9924
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056468564876744e-06,
+      "loss": 1.0315,
+      "step": 9925
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056373017389643e-06,
+      "loss": 1.3204,
+      "step": 9926
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.056277469902542e-06,
+      "loss": 0.9423,
+      "step": 9927
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0561819224154405e-06,
+      "loss": 1.0236,
+      "step": 9928
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0560863749283396e-06,
+      "loss": 1.3812,
+      "step": 9929
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055990827441239e-06,
+      "loss": 0.9619,
+      "step": 9930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055895279954138e-06,
+      "loss": 1.0824,
+      "step": 9931
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055799732467036e-06,
+      "loss": 1.3566,
+      "step": 9932
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055704184979936e-06,
+      "loss": 0.98,
+      "step": 9933
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055608637492835e-06,
+      "loss": 1.0904,
+      "step": 9934
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055513090005733e-06,
+      "loss": 1.2927,
+      "step": 9935
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055417542518632e-06,
+      "loss": 1.087,
+      "step": 9936
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055321995031531e-06,
+      "loss": 1.0659,
+      "step": 9937
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05522644754443e-06,
+      "loss": 1.4318,
+      "step": 9938
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055130900057328e-06,
+      "loss": 1.3358,
+      "step": 9939
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.055035352570228e-06,
+      "loss": 1.2785,
+      "step": 9940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054939805083127e-06,
+      "loss": 1.132,
+      "step": 9941
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054844257596025e-06,
+      "loss": 0.9793,
+      "step": 9942
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054748710108924e-06,
+      "loss": 0.9882,
+      "step": 9943
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0546531626218234e-06,
+      "loss": 0.7574,
+      "step": 9944
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0545576151347225e-06,
+      "loss": 0.6828,
+      "step": 9945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054462067647621e-06,
+      "loss": 0.9231,
+      "step": 9946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0543665201605206e-06,
+      "loss": 0.8636,
+      "step": 9947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05427097267342e-06,
+      "loss": 1.3084,
+      "step": 9948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054175425186318e-06,
+      "loss": 0.7452,
+      "step": 9949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.054079877699217e-06,
+      "loss": 0.8542,
+      "step": 9950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053984330212116e-06,
+      "loss": 0.9174,
+      "step": 9951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053888782725015e-06,
+      "loss": 1.1291,
+      "step": 9952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053793235237913e-06,
+      "loss": 0.9982,
+      "step": 9953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053697687750813e-06,
+      "loss": 0.8263,
+      "step": 9954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053602140263711e-06,
+      "loss": 0.8456,
+      "step": 9955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05350659277661e-06,
+      "loss": 0.9443,
+      "step": 9956
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053411045289509e-06,
+      "loss": 0.9961,
+      "step": 9957
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053315497802408e-06,
+      "loss": 1.1172,
+      "step": 9958
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0532199503153065e-06,
+      "loss": 1.2595,
+      "step": 9959
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0531244028282055e-06,
+      "loss": 0.985,
+      "step": 9960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053028855341105e-06,
+      "loss": 0.9596,
+      "step": 9961
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052933307854004e-06,
+      "loss": 1.0268,
+      "step": 9962
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052837760366903e-06,
+      "loss": 1.2765,
+      "step": 9963
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052742212879802e-06,
+      "loss": 0.8736,
+      "step": 9964
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052646665392701e-06,
+      "loss": 0.9363,
+      "step": 9965
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052551117905599e-06,
+      "loss": 1.1897,
+      "step": 9966
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052455570418498e-06,
+      "loss": 1.1165,
+      "step": 9967
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052360022931397e-06,
+      "loss": 1.2166,
+      "step": 9968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052264475444296e-06,
+      "loss": 0.9926,
+      "step": 9969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052168927957195e-06,
+      "loss": 1.3122,
+      "step": 9970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052073380470094e-06,
+      "loss": 1.2212,
+      "step": 9971
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051977832982993e-06,
+      "loss": 1.0404,
+      "step": 9972
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051882285495891e-06,
+      "loss": 1.0177,
+      "step": 9973
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.05178673800879e-06,
+      "loss": 1.4292,
+      "step": 9974
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051691190521689e-06,
+      "loss": 0.9369,
+      "step": 9975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0515956430345884e-06,
+      "loss": 1.2632,
+      "step": 9976
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0515000955474875e-06,
+      "loss": 1.1044,
+      "step": 9977
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0514045480603865e-06,
+      "loss": 0.838,
+      "step": 9978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0513090005732856e-06,
+      "loss": 1.1027,
+      "step": 9979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.051213453086184e-06,
+      "loss": 1.0805,
+      "step": 9980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.051117905599083e-06,
+      "loss": 1.0332,
+      "step": 9981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.051022358111982e-06,
+      "loss": 0.9005,
+      "step": 9982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050926810624881e-06,
+      "loss": 0.8886,
+      "step": 9983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.05083126313778e-06,
+      "loss": 1.3256,
+      "step": 9984
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050735715650679e-06,
+      "loss": 1.1365,
+      "step": 9985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050640168163578e-06,
+      "loss": 1.083,
+      "step": 9986
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050544620676476e-06,
+      "loss": 1.0532,
+      "step": 9987
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050449073189375e-06,
+      "loss": 1.1358,
+      "step": 9988
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050353525702274e-06,
+      "loss": 0.8112,
+      "step": 9989
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050257978215173e-06,
+      "loss": 1.0997,
+      "step": 9990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050162430728072e-06,
+      "loss": 1.5401,
+      "step": 9991
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.050066883240971e-06,
+      "loss": 0.8318,
+      "step": 9992
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04997133575387e-06,
+      "loss": 1.1278,
+      "step": 9993
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049875788266769e-06,
+      "loss": 0.7246,
+      "step": 9994
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049780240779668e-06,
+      "loss": 1.4476,
+      "step": 9995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049684693292567e-06,
+      "loss": 0.9347,
+      "step": 9996
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049589145805466e-06,
+      "loss": 1.0445,
+      "step": 9997
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049493598318365e-06,
+      "loss": 1.2071,
+      "step": 9998
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049398050831264e-06,
+      "loss": 1.2778,
+      "step": 9999
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049302503344163e-06,
+      "loss": 1.438,
+      "step": 10000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049206955857061e-06,
+      "loss": 0.925,
+      "step": 10001
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04911140836996e-06,
+      "loss": 0.9239,
+      "step": 10002
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.049015860882859e-06,
+      "loss": 0.8795,
+      "step": 10003
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.048920313395758e-06,
+      "loss": 1.0127,
+      "step": 10004
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.048824765908657e-06,
+      "loss": 1.2717,
+      "step": 10005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.048729218421556e-06,
+      "loss": 0.9136,
+      "step": 10006
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.048633670934454e-06,
+      "loss": 0.4609,
+      "step": 10007
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0485381234473534e-06,
+      "loss": 1.2689,
+      "step": 10008
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0484425759602525e-06,
+      "loss": 0.7485,
+      "step": 10009
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0483470284731515e-06,
+      "loss": 1.4364,
+      "step": 10010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0482514809860506e-06,
+      "loss": 1.4303,
+      "step": 10011
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04815593349895e-06,
+      "loss": 0.4542,
+      "step": 10012
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.048060386011849e-06,
+      "loss": 1.3492,
+      "step": 10013
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047964838524747e-06,
+      "loss": 1.7323,
+      "step": 10014
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047869291037646e-06,
+      "loss": 1.1313,
+      "step": 10015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047773743550545e-06,
+      "loss": 0.9268,
+      "step": 10016
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047678196063444e-06,
+      "loss": 1.1845,
+      "step": 10017
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047582648576342e-06,
+      "loss": 0.9043,
+      "step": 10018
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047487101089242e-06,
+      "loss": 1.3938,
+      "step": 10019
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047391553602141e-06,
+      "loss": 0.8454,
+      "step": 10020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047296006115039e-06,
+      "loss": 1.3466,
+      "step": 10021
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047200458627938e-06,
+      "loss": 0.7393,
+      "step": 10022
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047104911140837e-06,
+      "loss": 1.191,
+      "step": 10023
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047009363653736e-06,
+      "loss": 0.9509,
+      "step": 10024
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0469138161666346e-06,
+      "loss": 1.1842,
+      "step": 10025
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0468182686795344e-06,
+      "loss": 0.6243,
+      "step": 10026
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0467227211924335e-06,
+      "loss": 1.3567,
+      "step": 10027
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.046627173705332e-06,
+      "loss": 0.7918,
+      "step": 10028
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.046531626218231e-06,
+      "loss": 0.9301,
+      "step": 10029
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04643607873113e-06,
+      "loss": 1.4478,
+      "step": 10030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.046340531244029e-06,
+      "loss": 1.8253,
+      "step": 10031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.046244983756927e-06,
+      "loss": 1.0336,
+      "step": 10032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.046149436269827e-06,
+      "loss": 0.9763,
+      "step": 10033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.046053888782726e-06,
+      "loss": 1.0926,
+      "step": 10034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045958341295624e-06,
+      "loss": 0.8525,
+      "step": 10035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045862793808523e-06,
+      "loss": 1.1003,
+      "step": 10036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045767246321422e-06,
+      "loss": 0.9071,
+      "step": 10037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045671698834321e-06,
+      "loss": 0.8683,
+      "step": 10038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045576151347219e-06,
+      "loss": 0.9426,
+      "step": 10039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0454806038601184e-06,
+      "loss": 1.1391,
+      "step": 10040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045385056373018e-06,
+      "loss": 1.583,
+      "step": 10041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0452895088859165e-06,
+      "loss": 0.9717,
+      "step": 10042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0451939613988156e-06,
+      "loss": 1.3295,
+      "step": 10043
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045098413911715e-06,
+      "loss": 1.1234,
+      "step": 10044
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.045002866424614e-06,
+      "loss": 0.9942,
+      "step": 10045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044907318937512e-06,
+      "loss": 1.1945,
+      "step": 10046
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044811771450411e-06,
+      "loss": 0.9598,
+      "step": 10047
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04471622396331e-06,
+      "loss": 0.8226,
+      "step": 10048
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044620676476209e-06,
+      "loss": 1.0408,
+      "step": 10049
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044525128989108e-06,
+      "loss": 0.8943,
+      "step": 10050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044429581502007e-06,
+      "loss": 1.3917,
+      "step": 10051
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044334034014906e-06,
+      "loss": 1.3078,
+      "step": 10052
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044238486527804e-06,
+      "loss": 0.8709,
+      "step": 10053
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044142939040703e-06,
+      "loss": 1.5426,
+      "step": 10054
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.044047391553602e-06,
+      "loss": 1.217,
+      "step": 10055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043951844066501e-06,
+      "loss": 1.3352,
+      "step": 10056
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0438562965794e-06,
+      "loss": 0.94,
+      "step": 10057
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0437607490922994e-06,
+      "loss": 1.502,
+      "step": 10058
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043665201605198e-06,
+      "loss": 1.472,
+      "step": 10059
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043569654118097e-06,
+      "loss": 0.99,
+      "step": 10060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043474106630996e-06,
+      "loss": 0.9881,
+      "step": 10061
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043378559143895e-06,
+      "loss": 0.9544,
+      "step": 10062
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043283011656794e-06,
+      "loss": 0.7856,
+      "step": 10063
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043187464169693e-06,
+      "loss": 1.0116,
+      "step": 10064
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.043091916682592e-06,
+      "loss": 1.0289,
+      "step": 10065
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04299636919549e-06,
+      "loss": 0.8632,
+      "step": 10066
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042900821708389e-06,
+      "loss": 0.8517,
+      "step": 10067
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042805274221288e-06,
+      "loss": 1.1746,
+      "step": 10068
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042709726734187e-06,
+      "loss": 0.7172,
+      "step": 10069
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042614179247086e-06,
+      "loss": 1.1849,
+      "step": 10070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042518631759985e-06,
+      "loss": 1.0961,
+      "step": 10071
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.042423084272884e-06,
+      "loss": 0.9168,
+      "step": 10072
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0423275367857825e-06,
+      "loss": 0.7094,
+      "step": 10073
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0422319892986815e-06,
+      "loss": 0.8971,
+      "step": 10074
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0421364418115806e-06,
+      "loss": 1.0097,
+      "step": 10075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.04204089432448e-06,
+      "loss": 1.1034,
+      "step": 10076
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041945346837379e-06,
+      "loss": 1.2698,
+      "step": 10077
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041849799350278e-06,
+      "loss": 1.181,
+      "step": 10078
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041754251863177e-06,
+      "loss": 1.0029,
+      "step": 10079
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041658704376075e-06,
+      "loss": 1.056,
+      "step": 10080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041563156888974e-06,
+      "loss": 1.3356,
+      "step": 10081
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041467609401873e-06,
+      "loss": 0.9982,
+      "step": 10082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041372061914772e-06,
+      "loss": 1.1141,
+      "step": 10083
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.041276514427671e-06,
+      "loss": 0.8101,
+      "step": 10084
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.04118096694057e-06,
+      "loss": 0.8833,
+      "step": 10085
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.041085419453469e-06,
+      "loss": 0.648,
+      "step": 10086
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040989871966367e-06,
+      "loss": 1.6824,
+      "step": 10087
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040894324479266e-06,
+      "loss": 1.1751,
+      "step": 10088
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040798776992165e-06,
+      "loss": 1.193,
+      "step": 10089
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0407032295050644e-06,
+      "loss": 1.1687,
+      "step": 10090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0406076820179635e-06,
+      "loss": 1.409,
+      "step": 10091
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0405121345308625e-06,
+      "loss": 0.9092,
+      "step": 10092
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0404165870437616e-06,
+      "loss": 0.683,
+      "step": 10093
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.04032103955666e-06,
+      "loss": 0.8063,
+      "step": 10094
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040225492069559e-06,
+      "loss": 1.0371,
+      "step": 10095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040129944582458e-06,
+      "loss": 1.1859,
+      "step": 10096
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.040034397095357e-06,
+      "loss": 0.9142,
+      "step": 10097
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039938849608256e-06,
+      "loss": 1.2798,
+      "step": 10098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039843302121155e-06,
+      "loss": 1.2497,
+      "step": 10099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039747754634053e-06,
+      "loss": 0.771,
+      "step": 10100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039652207146952e-06,
+      "loss": 0.9648,
+      "step": 10101
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039556659659851e-06,
+      "loss": 0.6271,
+      "step": 10102
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03946111217275e-06,
+      "loss": 1.0101,
+      "step": 10103
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039365564685649e-06,
+      "loss": 0.8992,
+      "step": 10104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039270017198548e-06,
+      "loss": 0.9896,
+      "step": 10105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.039174469711447e-06,
+      "loss": 0.9938,
+      "step": 10106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0390789222243456e-06,
+      "loss": 1.106,
+      "step": 10107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038983374737245e-06,
+      "loss": 0.9511,
+      "step": 10108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038887827250144e-06,
+      "loss": 0.916,
+      "step": 10109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038792279763043e-06,
+      "loss": 1.1,
+      "step": 10110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038696732275941e-06,
+      "loss": 0.8899,
+      "step": 10111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03860118478884e-06,
+      "loss": 1.0441,
+      "step": 10112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03850563730174e-06,
+      "loss": 0.8891,
+      "step": 10113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038410089814638e-06,
+      "loss": 0.6325,
+      "step": 10114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038314542327537e-06,
+      "loss": 1.1132,
+      "step": 10115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038218994840436e-06,
+      "loss": 1.633,
+      "step": 10116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038123447353335e-06,
+      "loss": 1.2824,
+      "step": 10117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.038027899866233e-06,
+      "loss": 1.4264,
+      "step": 10118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037932352379132e-06,
+      "loss": 1.146,
+      "step": 10119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037836804892032e-06,
+      "loss": 0.8432,
+      "step": 10120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03774125740493e-06,
+      "loss": 0.8406,
+      "step": 10121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0376457099178294e-06,
+      "loss": 0.9628,
+      "step": 10122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0375501624307285e-06,
+      "loss": 0.9032,
+      "step": 10123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0374546149436275e-06,
+      "loss": 1.3911,
+      "step": 10124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037359067456526e-06,
+      "loss": 1.0815,
+      "step": 10125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037263519969425e-06,
+      "loss": 1.2145,
+      "step": 10126
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037167972482325e-06,
+      "loss": 1.2741,
+      "step": 10127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.037072424995223e-06,
+      "loss": 0.7019,
+      "step": 10128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036976877508122e-06,
+      "loss": 1.4255,
+      "step": 10129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036881330021021e-06,
+      "loss": 1.1035,
+      "step": 10130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03678578253392e-06,
+      "loss": 1.3488,
+      "step": 10131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036690235046818e-06,
+      "loss": 1.5218,
+      "step": 10132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036594687559717e-06,
+      "loss": 1.0098,
+      "step": 10133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036499140072617e-06,
+      "loss": 1.1376,
+      "step": 10134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036403592585515e-06,
+      "loss": 1.2856,
+      "step": 10135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.036308045098414e-06,
+      "loss": 1.1906,
+      "step": 10136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.036212497611313e-06,
+      "loss": 0.6465,
+      "step": 10137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.036116950124212e-06,
+      "loss": 1.2354,
+      "step": 10138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0360214026371106e-06,
+      "loss": 0.9503,
+      "step": 10139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03592585515001e-06,
+      "loss": 1.3813,
+      "step": 10140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0358303076629095e-06,
+      "loss": 0.6393,
+      "step": 10141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035734760175808e-06,
+      "loss": 1.4826,
+      "step": 10142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035639212688707e-06,
+      "loss": 1.3531,
+      "step": 10143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035543665201606e-06,
+      "loss": 1.0587,
+      "step": 10144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035448117714505e-06,
+      "loss": 0.8448,
+      "step": 10145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035352570227403e-06,
+      "loss": 0.8992,
+      "step": 10146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035257022740302e-06,
+      "loss": 1.2838,
+      "step": 10147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.035161475253201e-06,
+      "loss": 1.4471,
+      "step": 10148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0350659277661e-06,
+      "loss": 1.2343,
+      "step": 10149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034970380278999e-06,
+      "loss": 0.8179,
+      "step": 10150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034874832791898e-06,
+      "loss": 0.9455,
+      "step": 10151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034779285304797e-06,
+      "loss": 0.8133,
+      "step": 10152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034683737817695e-06,
+      "loss": 1.0559,
+      "step": 10153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0345881903305944e-06,
+      "loss": 1.4671,
+      "step": 10154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0344926428434935e-06,
+      "loss": 0.8971,
+      "step": 10155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0343970953563925e-06,
+      "loss": 1.1727,
+      "step": 10156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0343015478692916e-06,
+      "loss": 1.1209,
+      "step": 10157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034206000382191e-06,
+      "loss": 0.6046,
+      "step": 10158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034110452895089e-06,
+      "loss": 1.031,
+      "step": 10159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.034014905407988e-06,
+      "loss": 1.306,
+      "step": 10160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033919357920887e-06,
+      "loss": 1.1621,
+      "step": 10161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033823810433786e-06,
+      "loss": 0.9891,
+      "step": 10162
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033728262946685e-06,
+      "loss": 1.5619,
+      "step": 10163
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033632715459584e-06,
+      "loss": 1.3876,
+      "step": 10164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033537167972483e-06,
+      "loss": 0.9272,
+      "step": 10165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033441620485381e-06,
+      "loss": 1.6327,
+      "step": 10166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03334607299828e-06,
+      "loss": 0.878,
+      "step": 10167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033250525511179e-06,
+      "loss": 1.7598,
+      "step": 10168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033154978024078e-06,
+      "loss": 0.9714,
+      "step": 10169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.033059430536977e-06,
+      "loss": 1.4127,
+      "step": 10170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032963883049876e-06,
+      "loss": 1.5366,
+      "step": 10171
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0328683355627754e-06,
+      "loss": 1.1103,
+      "step": 10172
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032772788075674e-06,
+      "loss": 1.1796,
+      "step": 10173
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032677240588573e-06,
+      "loss": 0.8163,
+      "step": 10174
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032581693101472e-06,
+      "loss": 0.7531,
+      "step": 10175
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032486145614371e-06,
+      "loss": 0.9814,
+      "step": 10176
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03239059812727e-06,
+      "loss": 0.8057,
+      "step": 10177
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032295050640169e-06,
+      "loss": 1.0653,
+      "step": 10178
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032199503153068e-06,
+      "loss": 0.9154,
+      "step": 10179
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032103955665966e-06,
+      "loss": 0.7067,
+      "step": 10180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032008408178865e-06,
+      "loss": 1.2498,
+      "step": 10181
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.031912860691764e-06,
+      "loss": 1.2342,
+      "step": 10182
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.031817313204663e-06,
+      "loss": 1.1012,
+      "step": 10183
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.031721765717561e-06,
+      "loss": 0.8605,
+      "step": 10184
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.031626218230461e-06,
+      "loss": 1.1865,
+      "step": 10185
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.03153067074336e-06,
+      "loss": 1.0049,
+      "step": 10186
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0314351232562585e-06,
+      "loss": 0.9479,
+      "step": 10187
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0313395757691575e-06,
+      "loss": 1.0376,
+      "step": 10188
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0312440282820566e-06,
+      "loss": 1.8069,
+      "step": 10189
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.031148480794956e-06,
+      "loss": 1.2446,
+      "step": 10190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.031052933307854e-06,
+      "loss": 1.6356,
+      "step": 10191
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030957385820754e-06,
+      "loss": 1.9512,
+      "step": 10192
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030861838333653e-06,
+      "loss": 1.2012,
+      "step": 10193
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030766290846551e-06,
+      "loss": 1.3374,
+      "step": 10194
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.03067074335945e-06,
+      "loss": 0.9069,
+      "step": 10195
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030575195872349e-06,
+      "loss": 1.2725,
+      "step": 10196
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030479648385248e-06,
+      "loss": 1.3253,
+      "step": 10197
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030384100898146e-06,
+      "loss": 0.7631,
+      "step": 10198
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030288553411046e-06,
+      "loss": 1.4954,
+      "step": 10199
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030193005923944e-06,
+      "loss": 1.4514,
+      "step": 10200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030097458436843e-06,
+      "loss": 0.9092,
+      "step": 10201
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.030001910949742e-06,
+      "loss": 1.4524,
+      "step": 10202
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029906363462641e-06,
+      "loss": 1.1001,
+      "step": 10203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0298108159755404e-06,
+      "loss": 0.8171,
+      "step": 10204
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029715268488439e-06,
+      "loss": 1.3689,
+      "step": 10205
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0296197210013385e-06,
+      "loss": 0.8519,
+      "step": 10206
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029524173514237e-06,
+      "loss": 1.4867,
+      "step": 10207
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029428626027136e-06,
+      "loss": 0.6982,
+      "step": 10208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029333078540035e-06,
+      "loss": 0.8422,
+      "step": 10209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029237531052934e-06,
+      "loss": 1.0904,
+      "step": 10210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029141983565832e-06,
+      "loss": 1.2105,
+      "step": 10211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.029046436078731e-06,
+      "loss": 1.4786,
+      "step": 10212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028950888591631e-06,
+      "loss": 1.0153,
+      "step": 10213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028855341104529e-06,
+      "loss": 0.9545,
+      "step": 10214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028759793617428e-06,
+      "loss": 0.9384,
+      "step": 10215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028664246130327e-06,
+      "loss": 1.289,
+      "step": 10216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028568698643226e-06,
+      "loss": 0.8998,
+      "step": 10217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0284731511561244e-06,
+      "loss": 1.0745,
+      "step": 10218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0283776036690235e-06,
+      "loss": 1.7058,
+      "step": 10219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0282820561819225e-06,
+      "loss": 1.1482,
+      "step": 10220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0281865086948216e-06,
+      "loss": 0.8123,
+      "step": 10221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.028090961207721e-06,
+      "loss": 1.6472,
+      "step": 10222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.02799541372062e-06,
+      "loss": 0.4827,
+      "step": 10223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027899866233519e-06,
+      "loss": 0.8992,
+      "step": 10224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027804318746417e-06,
+      "loss": 1.2172,
+      "step": 10225
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027708771259316e-06,
+      "loss": 1.1399,
+      "step": 10226
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027613223772215e-06,
+      "loss": 1.1213,
+      "step": 10227
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027517676285114e-06,
+      "loss": 0.7203,
+      "step": 10228
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027422128798013e-06,
+      "loss": 1.0724,
+      "step": 10229
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027326581310912e-06,
+      "loss": 1.423,
+      "step": 10230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027231033823811e-06,
+      "loss": 1.1164,
+      "step": 10231
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027135486336709e-06,
+      "loss": 1.0992,
+      "step": 10232
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.027039938849608e-06,
+      "loss": 1.1365,
+      "step": 10233
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026944391362507e-06,
+      "loss": 0.992,
+      "step": 10234
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026848843875406e-06,
+      "loss": 0.9582,
+      "step": 10235
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0267532963883054e-06,
+      "loss": 1.1837,
+      "step": 10236
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0266577489012045e-06,
+      "loss": 1.0914,
+      "step": 10237
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0265622014141035e-06,
+      "loss": 0.892,
+      "step": 10238
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026466653927002e-06,
+      "loss": 1.6771,
+      "step": 10239
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026371106439901e-06,
+      "loss": 1.117,
+      "step": 10240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0262755589528e-06,
+      "loss": 0.7046,
+      "step": 10241
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.026180011465699e-06,
+      "loss": 1.1342,
+      "step": 10242
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.026084463978598e-06,
+      "loss": 1.2127,
+      "step": 10243
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025988916491497e-06,
+      "loss": 0.7727,
+      "step": 10244
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025893369004396e-06,
+      "loss": 0.9024,
+      "step": 10245
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025797821517294e-06,
+      "loss": 0.9768,
+      "step": 10246
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025702274030193e-06,
+      "loss": 0.9629,
+      "step": 10247
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025606726543092e-06,
+      "loss": 0.8361,
+      "step": 10248
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025511179055991e-06,
+      "loss": 0.8837,
+      "step": 10249
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02541563156889e-06,
+      "loss": 1.2299,
+      "step": 10250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025320084081789e-06,
+      "loss": 1.4143,
+      "step": 10251
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0252245365946875e-06,
+      "loss": 1.409,
+      "step": 10252
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0251289891075866e-06,
+      "loss": 0.8608,
+      "step": 10253
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.025033441620486e-06,
+      "loss": 1.1334,
+      "step": 10254
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024937894133385e-06,
+      "loss": 0.8225,
+      "step": 10255
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024842346646284e-06,
+      "loss": 1.0247,
+      "step": 10256
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024746799159183e-06,
+      "loss": 0.823,
+      "step": 10257
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024651251672082e-06,
+      "loss": 1.024,
+      "step": 10258
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02455570418498e-06,
+      "loss": 0.9481,
+      "step": 10259
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024460156697879e-06,
+      "loss": 0.8309,
+      "step": 10260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024364609210778e-06,
+      "loss": 0.9315,
+      "step": 10261
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024269061723677e-06,
+      "loss": 0.8064,
+      "step": 10262
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024173514236575e-06,
+      "loss": 1.3305,
+      "step": 10263
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.024077966749475e-06,
+      "loss": 1.022,
+      "step": 10264
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023982419262374e-06,
+      "loss": 1.4436,
+      "step": 10265
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023886871775272e-06,
+      "loss": 1.2359,
+      "step": 10266
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023791324288171e-06,
+      "loss": 1.1276,
+      "step": 10267
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0236957768010704e-06,
+      "loss": 0.8546,
+      "step": 10268
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0236002293139695e-06,
+      "loss": 0.8403,
+      "step": 10269
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023504681826868e-06,
+      "loss": 1.4312,
+      "step": 10270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0234091343397676e-06,
+      "loss": 1.1334,
+      "step": 10271
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023313586852667e-06,
+      "loss": 1.1253,
+      "step": 10272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023218039365565e-06,
+      "loss": 1.2763,
+      "step": 10273
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023122491878464e-06,
+      "loss": 1.0358,
+      "step": 10274
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.023026944391363e-06,
+      "loss": 1.0941,
+      "step": 10275
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022931396904262e-06,
+      "loss": 1.0945,
+      "step": 10276
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02283584941716e-06,
+      "loss": 0.9947,
+      "step": 10277
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02274030193006e-06,
+      "loss": 1.1731,
+      "step": 10278
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022644754442959e-06,
+      "loss": 1.15,
+      "step": 10279
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022549206955857e-06,
+      "loss": 1.0263,
+      "step": 10280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022453659468756e-06,
+      "loss": 1.0684,
+      "step": 10281
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022358111981655e-06,
+      "loss": 1.5452,
+      "step": 10282
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022262564494554e-06,
+      "loss": 0.9068,
+      "step": 10283
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0221670170074525e-06,
+      "loss": 1.4969,
+      "step": 10284
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022071469520352e-06,
+      "loss": 1.4911,
+      "step": 10285
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0219759220332514e-06,
+      "loss": 1.1902,
+      "step": 10286
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.02188037454615e-06,
+      "loss": 1.126,
+      "step": 10287
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021784827059049e-06,
+      "loss": 1.0423,
+      "step": 10288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021689279571948e-06,
+      "loss": 1.0013,
+      "step": 10289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021593732084847e-06,
+      "loss": 1.1429,
+      "step": 10290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021498184597745e-06,
+      "loss": 0.8347,
+      "step": 10291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021402637110644e-06,
+      "loss": 1.2483,
+      "step": 10292
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.021307089623543e-06,
+      "loss": 1.1151,
+      "step": 10293
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.021211542136442e-06,
+      "loss": 0.8281,
+      "step": 10294
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.021115994649341e-06,
+      "loss": 1.0758,
+      "step": 10295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.02102044716224e-06,
+      "loss": 1.0123,
+      "step": 10296
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.020924899675139e-06,
+      "loss": 1.0192,
+      "step": 10297
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.020829352188037e-06,
+      "loss": 1.0153,
+      "step": 10298
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.020733804700936e-06,
+      "loss": 1.0862,
+      "step": 10299
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0206382572138354e-06,
+      "loss": 1.1087,
+      "step": 10300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0205427097267345e-06,
+      "loss": 1.3538,
+      "step": 10301
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0204471622396335e-06,
+      "loss": 1.2847,
+      "step": 10302
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0203516147525326e-06,
+      "loss": 1.4201,
+      "step": 10303
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.020256067265431e-06,
+      "loss": 0.9381,
+      "step": 10304
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.02016051977833e-06,
+      "loss": 0.969,
+      "step": 10305
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.020064972291229e-06,
+      "loss": 0.7614,
+      "step": 10306
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019969424804128e-06,
+      "loss": 1.1361,
+      "step": 10307
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019873877317027e-06,
+      "loss": 1.0404,
+      "step": 10308
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019778329829926e-06,
+      "loss": 1.4522,
+      "step": 10309
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019682782342825e-06,
+      "loss": 0.9946,
+      "step": 10310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019587234855723e-06,
+      "loss": 1.0471,
+      "step": 10311
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019491687368622e-06,
+      "loss": 1.1103,
+      "step": 10312
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019396139881521e-06,
+      "loss": 0.9672,
+      "step": 10313
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.01930059239442e-06,
+      "loss": 0.8875,
+      "step": 10314
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019205044907319e-06,
+      "loss": 1.1614,
+      "step": 10315
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019109497420218e-06,
+      "loss": 0.8767,
+      "step": 10316
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.019013949933117e-06,
+      "loss": 1.6292,
+      "step": 10317
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018918402446016e-06,
+      "loss": 1.1423,
+      "step": 10318
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018822854958915e-06,
+      "loss": 0.9458,
+      "step": 10319
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018727307471814e-06,
+      "loss": 1.3093,
+      "step": 10320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018631759984713e-06,
+      "loss": 0.8381,
+      "step": 10321
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018536212497612e-06,
+      "loss": 1.0112,
+      "step": 10322
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018440665010511e-06,
+      "loss": 1.0507,
+      "step": 10323
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.01834511752341e-06,
+      "loss": 1.3679,
+      "step": 10324
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018249570036308e-06,
+      "loss": 1.3191,
+      "step": 10325
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018154022549207e-06,
+      "loss": 0.881,
+      "step": 10326
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.018058475062106e-06,
+      "loss": 1.3784,
+      "step": 10327
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017962927575005e-06,
+      "loss": 1.3065,
+      "step": 10328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017867380087904e-06,
+      "loss": 1.653,
+      "step": 10329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017771832600803e-06,
+      "loss": 1.2062,
+      "step": 10330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017676285113702e-06,
+      "loss": 1.1759,
+      "step": 10331
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0175807376266004e-06,
+      "loss": 1.1524,
+      "step": 10332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0174851901394995e-06,
+      "loss": 1.4982,
+      "step": 10333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0173896426523985e-06,
+      "loss": 0.5578,
+      "step": 10334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0172940951652976e-06,
+      "loss": 1.126,
+      "step": 10335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017198547678197e-06,
+      "loss": 1.054,
+      "step": 10336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017103000191096e-06,
+      "loss": 0.8966,
+      "step": 10337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.017007452703995e-06,
+      "loss": 1.3308,
+      "step": 10338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016911905216893e-06,
+      "loss": 0.805,
+      "step": 10339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016816357729792e-06,
+      "loss": 1.0038,
+      "step": 10340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016720810242691e-06,
+      "loss": 1.0671,
+      "step": 10341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.01662526275559e-06,
+      "loss": 0.9372,
+      "step": 10342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016529715268489e-06,
+      "loss": 1.3517,
+      "step": 10343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016434167781388e-06,
+      "loss": 0.8333,
+      "step": 10344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016338620294287e-06,
+      "loss": 1.6417,
+      "step": 10345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.016243072807185e-06,
+      "loss": 1.26,
+      "step": 10346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.016147525320084e-06,
+      "loss": 1.1435,
+      "step": 10347
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.016051977832983e-06,
+      "loss": 1.2072,
+      "step": 10348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015956430345882e-06,
+      "loss": 1.1464,
+      "step": 10349
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0158608828587814e-06,
+      "loss": 0.9457,
+      "step": 10350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0157653353716805e-06,
+      "loss": 1.0812,
+      "step": 10351
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015669787884579e-06,
+      "loss": 1.0342,
+      "step": 10352
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015574240397478e-06,
+      "loss": 1.4137,
+      "step": 10353
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015478692910377e-06,
+      "loss": 1.3599,
+      "step": 10354
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015383145423276e-06,
+      "loss": 0.9898,
+      "step": 10355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015287597936174e-06,
+      "loss": 1.2911,
+      "step": 10356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015192050449074e-06,
+      "loss": 0.8273,
+      "step": 10357
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015096502961973e-06,
+      "loss": 1.2818,
+      "step": 10358
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.015000955474871e-06,
+      "loss": 1.4924,
+      "step": 10359
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01490540798777e-06,
+      "loss": 1.0224,
+      "step": 10360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014809860500669e-06,
+      "loss": 1.1996,
+      "step": 10361
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014714313013568e-06,
+      "loss": 1.0748,
+      "step": 10362
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014618765526466e-06,
+      "loss": 1.5229,
+      "step": 10363
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0145232180393654e-06,
+      "loss": 1.3453,
+      "step": 10364
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014427670552265e-06,
+      "loss": 1.1214,
+      "step": 10365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0143321230651635e-06,
+      "loss": 1.2364,
+      "step": 10366
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0142365755780626e-06,
+      "loss": 0.9407,
+      "step": 10367
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014141028090962e-06,
+      "loss": 1.0033,
+      "step": 10368
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.014045480603861e-06,
+      "loss": 0.8251,
+      "step": 10369
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013949933116759e-06,
+      "loss": 1.0689,
+      "step": 10370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013854385629658e-06,
+      "loss": 0.7648,
+      "step": 10371
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013758838142558e-06,
+      "loss": 1.7805,
+      "step": 10372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013663290655456e-06,
+      "loss": 1.4083,
+      "step": 10373
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013567743168355e-06,
+      "loss": 0.8236,
+      "step": 10374
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013472195681254e-06,
+      "loss": 1.3978,
+      "step": 10375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013376648194153e-06,
+      "loss": 1.2657,
+      "step": 10376
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.013281100707051e-06,
+      "loss": 0.8704,
+      "step": 10377
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01318555321995e-06,
+      "loss": 1.6999,
+      "step": 10378
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01309000573285e-06,
+      "loss": 0.9999,
+      "step": 10379
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012994458245748e-06,
+      "loss": 0.8855,
+      "step": 10380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012898910758647e-06,
+      "loss": 1.3881,
+      "step": 10381
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0128033632715464e-06,
+      "loss": 1.1409,
+      "step": 10382
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0127078157844455e-06,
+      "loss": 1.6307,
+      "step": 10383
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012612268297344e-06,
+      "loss": 1.3928,
+      "step": 10384
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012516720810243e-06,
+      "loss": 1.1966,
+      "step": 10385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012421173323143e-06,
+      "loss": 1.1434,
+      "step": 10386
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012325625836041e-06,
+      "loss": 1.1445,
+      "step": 10387
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01223007834894e-06,
+      "loss": 1.0568,
+      "step": 10388
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012134530861839e-06,
+      "loss": 0.7197,
+      "step": 10389
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012038983374738e-06,
+      "loss": 1.0359,
+      "step": 10390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011943435887636e-06,
+      "loss": 1.1154,
+      "step": 10391
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011847888400535e-06,
+      "loss": 1.4571,
+      "step": 10392
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011752340913434e-06,
+      "loss": 1.2617,
+      "step": 10393
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011656793426333e-06,
+      "loss": 0.8449,
+      "step": 10394
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011561245939232e-06,
+      "loss": 1.4705,
+      "step": 10395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.011465698452131e-06,
+      "loss": 1.3937,
+      "step": 10396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01137015096503e-06,
+      "loss": 1.1622,
+      "step": 10397
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0112746034779285e-06,
+      "loss": 0.9029,
+      "step": 10398
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0111790559908276e-06,
+      "loss": 0.9328,
+      "step": 10399
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.011083508503727e-06,
+      "loss": 1.0706,
+      "step": 10400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010987961016626e-06,
+      "loss": 1.0808,
+      "step": 10401
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010892413529525e-06,
+      "loss": 1.0603,
+      "step": 10402
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010796866042424e-06,
+      "loss": 0.6042,
+      "step": 10403
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010701318555322e-06,
+      "loss": 1.4022,
+      "step": 10404
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010605771068221e-06,
+      "loss": 1.0249,
+      "step": 10405
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.01051022358112e-06,
+      "loss": 1.0618,
+      "step": 10406
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010414676094019e-06,
+      "loss": 1.3091,
+      "step": 10407
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010319128606918e-06,
+      "loss": 0.8393,
+      "step": 10408
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010223581119817e-06,
+      "loss": 0.8348,
+      "step": 10409
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010128033632716e-06,
+      "loss": 1.1667,
+      "step": 10410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.010032486145614e-06,
+      "loss": 1.5003,
+      "step": 10411
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009936938658513e-06,
+      "loss": 0.794,
+      "step": 10412
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009841391171412e-06,
+      "loss": 0.9721,
+      "step": 10413
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0097458436843114e-06,
+      "loss": 0.9732,
+      "step": 10414
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0096502961972105e-06,
+      "loss": 1.0903,
+      "step": 10415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0095547487101095e-06,
+      "loss": 0.9548,
+      "step": 10416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0094592012230086e-06,
+      "loss": 1.2049,
+      "step": 10417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009363653735907e-06,
+      "loss": 0.8627,
+      "step": 10418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009268106248806e-06,
+      "loss": 0.717,
+      "step": 10419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009172558761705e-06,
+      "loss": 0.8723,
+      "step": 10420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.009077011274604e-06,
+      "loss": 0.7346,
+      "step": 10421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008981463787503e-06,
+      "loss": 1.3906,
+      "step": 10422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008885916300402e-06,
+      "loss": 1.1556,
+      "step": 10423
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008790368813301e-06,
+      "loss": 1.1484,
+      "step": 10424
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008694821326199e-06,
+      "loss": 1.3421,
+      "step": 10425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008599273839098e-06,
+      "loss": 1.2526,
+      "step": 10426
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008503726351997e-06,
+      "loss": 1.3944,
+      "step": 10427
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008408178864896e-06,
+      "loss": 0.8601,
+      "step": 10428
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008312631377795e-06,
+      "loss": 1.5407,
+      "step": 10429
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008217083890694e-06,
+      "loss": 0.6789,
+      "step": 10430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008121536403593e-06,
+      "loss": 1.2335,
+      "step": 10431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.008025988916492e-06,
+      "loss": 0.9357,
+      "step": 10432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007930441429391e-06,
+      "loss": 0.9139,
+      "step": 10433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.00783489394229e-06,
+      "loss": 0.9154,
+      "step": 10434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007739346455189e-06,
+      "loss": 0.9662,
+      "step": 10435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007643798968087e-06,
+      "loss": 0.8713,
+      "step": 10436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007548251480987e-06,
+      "loss": 1.2619,
+      "step": 10437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007452703993886e-06,
+      "loss": 0.9847,
+      "step": 10438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007357156506784e-06,
+      "loss": 1.2344,
+      "step": 10439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007261609019683e-06,
+      "loss": 1.3179,
+      "step": 10440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007166061532582e-06,
+      "loss": 1.2808,
+      "step": 10441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007070514045481e-06,
+      "loss": 1.2121,
+      "step": 10442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006974966558379e-06,
+      "loss": 0.9886,
+      "step": 10443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006879419071279e-06,
+      "loss": 1.219,
+      "step": 10444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006783871584177e-06,
+      "loss": 0.9687,
+      "step": 10445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0066883240970764e-06,
+      "loss": 0.9402,
+      "step": 10446
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0065927766099755e-06,
+      "loss": 1.1166,
+      "step": 10447
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0064972291228745e-06,
+      "loss": 1.514,
+      "step": 10448
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0064016816357736e-06,
+      "loss": 1.2863,
+      "step": 10449
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.006306134148672e-06,
+      "loss": 1.1215,
+      "step": 10450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.006210586661572e-06,
+      "loss": 1.3715,
+      "step": 10451
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00611503917447e-06,
+      "loss": 1.3721,
+      "step": 10452
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.006019491687369e-06,
+      "loss": 0.9718,
+      "step": 10453
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005923944200268e-06,
+      "loss": 1.1804,
+      "step": 10454
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005828396713167e-06,
+      "loss": 1.1917,
+      "step": 10455
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005732849226065e-06,
+      "loss": 1.1252,
+      "step": 10456
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005637301738964e-06,
+      "loss": 0.7257,
+      "step": 10457
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005541754251864e-06,
+      "loss": 1.4419,
+      "step": 10458
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005446206764762e-06,
+      "loss": 0.9432,
+      "step": 10459
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005350659277661e-06,
+      "loss": 1.1717,
+      "step": 10460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00525511179056e-06,
+      "loss": 1.2732,
+      "step": 10461
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.005159564303459e-06,
+      "loss": 1.2459,
+      "step": 10462
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0050640168163576e-06,
+      "loss": 0.843,
+      "step": 10463
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004968469329257e-06,
+      "loss": 1.1904,
+      "step": 10464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0048729218421565e-06,
+      "loss": 0.969,
+      "step": 10465
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004777374355055e-06,
+      "loss": 1.1564,
+      "step": 10466
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004681826867954e-06,
+      "loss": 1.121,
+      "step": 10467
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004586279380853e-06,
+      "loss": 1.0852,
+      "step": 10468
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004490731893752e-06,
+      "loss": 1.0454,
+      "step": 10469
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00439518440665e-06,
+      "loss": 1.0348,
+      "step": 10470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004299636919549e-06,
+      "loss": 1.1976,
+      "step": 10471
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004204089432448e-06,
+      "loss": 0.5494,
+      "step": 10472
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004108541945347e-06,
+      "loss": 1.0393,
+      "step": 10473
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.004012994458246e-06,
+      "loss": 1.1825,
+      "step": 10474
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003917446971145e-06,
+      "loss": 1.0053,
+      "step": 10475
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003821899484044e-06,
+      "loss": 0.9506,
+      "step": 10476
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6868720616662292,
+      "eval_loss": 1.1045219898223877,
+      "eval_runtime": 141.0045,
+      "eval_samples_per_second": 297.175,
+      "eval_steps_per_second": 2.326,
+      "step": 10476
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003726351996942e-06,
+      "loss": 1.5563,
+      "step": 10477
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0036308045098414e-06,
+      "loss": 1.0491,
+      "step": 10478
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0035352570227405e-06,
+      "loss": 0.856,
+      "step": 10479
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0034397095356395e-06,
+      "loss": 0.7296,
+      "step": 10480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0033441620485386e-06,
+      "loss": 0.8767,
+      "step": 10481
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003248614561438e-06,
+      "loss": 0.8899,
+      "step": 10482
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003153067074337e-06,
+      "loss": 0.884,
+      "step": 10483
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.003057519587235e-06,
+      "loss": 0.9152,
+      "step": 10484
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002961972100134e-06,
+      "loss": 0.6551,
+      "step": 10485
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002866424613033e-06,
+      "loss": 1.4204,
+      "step": 10486
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002770877125932e-06,
+      "loss": 0.9356,
+      "step": 10487
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002675329638831e-06,
+      "loss": 0.8019,
+      "step": 10488
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00257978215173e-06,
+      "loss": 1.0693,
+      "step": 10489
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002484234664629e-06,
+      "loss": 0.8935,
+      "step": 10490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002388687177527e-06,
+      "loss": 0.9722,
+      "step": 10491
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002293139690426e-06,
+      "loss": 0.7229,
+      "step": 10492
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002197592203325e-06,
+      "loss": 0.8974,
+      "step": 10493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002102044716224e-06,
+      "loss": 1.041,
+      "step": 10494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002006497229123e-06,
+      "loss": 0.8154,
+      "step": 10495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0019109497420224e-06,
+      "loss": 0.9668,
+      "step": 10496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001815402254921e-06,
+      "loss": 0.8639,
+      "step": 10497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.00171985476782e-06,
+      "loss": 1.3934,
+      "step": 10498
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001624307280719e-06,
+      "loss": 1.0312,
+      "step": 10499
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001528759793618e-06,
+      "loss": 1.0478,
+      "step": 10500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001433212306517e-06,
+      "loss": 0.641,
+      "step": 10501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001337664819416e-06,
+      "loss": 0.7224,
+      "step": 10502
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.001242117332315e-06,
+      "loss": 0.9283,
+      "step": 10503
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.001146569845213e-06,
+      "loss": 0.9681,
+      "step": 10504
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.001051022358112e-06,
+      "loss": 0.5965,
+      "step": 10505
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000955474871011e-06,
+      "loss": 0.9705,
+      "step": 10506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.00085992738391e-06,
+      "loss": 1.2166,
+      "step": 10507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000764379896808e-06,
+      "loss": 0.8231,
+      "step": 10508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000668832409708e-06,
+      "loss": 1.0086,
+      "step": 10509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000573284922607e-06,
+      "loss": 0.7517,
+      "step": 10510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0004777374355055e-06,
+      "loss": 0.9477,
+      "step": 10511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0003821899484045e-06,
+      "loss": 1.4981,
+      "step": 10512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0002866424613036e-06,
+      "loss": 1.222,
+      "step": 10513
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000191094974203e-06,
+      "loss": 1.2286,
+      "step": 10514
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000095547487101e-06,
+      "loss": 1.236,
+      "step": 10515
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.783,
+      "step": 10516
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9999044525129e-06,
+      "loss": 1.1703,
+      "step": 10517
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999808905025798e-06,
+      "loss": 0.8008,
+      "step": 10518
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999713357538697e-06,
+      "loss": 1.1747,
+      "step": 10519
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999617810051596e-06,
+      "loss": 0.9992,
+      "step": 10520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999522262564495e-06,
+      "loss": 1.228,
+      "step": 10521
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999426715077393e-06,
+      "loss": 1.0856,
+      "step": 10522
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999331167590293e-06,
+      "loss": 1.026,
+      "step": 10523
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999235620103192e-06,
+      "loss": 0.9424,
+      "step": 10524
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.99914007261609e-06,
+      "loss": 0.5734,
+      "step": 10525
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.999044525128989e-06,
+      "loss": 1.0677,
+      "step": 10526
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998948977641888e-06,
+      "loss": 1.2523,
+      "step": 10527
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9988534301547874e-06,
+      "loss": 1.1781,
+      "step": 10528
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998757882667686e-06,
+      "loss": 0.945,
+      "step": 10529
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9986623351805855e-06,
+      "loss": 0.8308,
+      "step": 10530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9985667876934846e-06,
+      "loss": 1.0804,
+      "step": 10531
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998471240206383e-06,
+      "loss": 0.8203,
+      "step": 10532
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998375692719282e-06,
+      "loss": 0.7471,
+      "step": 10533
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998280145232181e-06,
+      "loss": 1.5695,
+      "step": 10534
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.99818459774508e-06,
+      "loss": 1.0456,
+      "step": 10535
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.998089050257978e-06,
+      "loss": 1.271,
+      "step": 10536
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997993502770878e-06,
+      "loss": 1.074,
+      "step": 10537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997897955283777e-06,
+      "loss": 1.0635,
+      "step": 10538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997802407796675e-06,
+      "loss": 0.9562,
+      "step": 10539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997706860309574e-06,
+      "loss": 0.9022,
+      "step": 10540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997611312822473e-06,
+      "loss": 0.7061,
+      "step": 10541
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997515765335372e-06,
+      "loss": 1.0331,
+      "step": 10542
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9974202178482705e-06,
+      "loss": 1.0569,
+      "step": 10543
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9973246703611695e-06,
+      "loss": 0.824,
+      "step": 10544
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9972291228740685e-06,
+      "loss": 1.3172,
+      "step": 10545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997133575386968e-06,
+      "loss": 0.8125,
+      "step": 10546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.997038027899867e-06,
+      "loss": 0.7518,
+      "step": 10547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996942480412766e-06,
+      "loss": 1.0344,
+      "step": 10548
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996846932925664e-06,
+      "loss": 0.8322,
+      "step": 10549
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996751385438563e-06,
+      "loss": 1.2307,
+      "step": 10550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996655837951462e-06,
+      "loss": 1.1327,
+      "step": 10551
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996560290464361e-06,
+      "loss": 1.0074,
+      "step": 10552
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.99646474297726e-06,
+      "loss": 1.1677,
+      "step": 10553
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996369195490159e-06,
+      "loss": 0.5899,
+      "step": 10554
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.996273648003058e-06,
+      "loss": 0.7717,
+      "step": 10555
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.996178100515956e-06,
+      "loss": 1.5489,
+      "step": 10556
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.996082553028855e-06,
+      "loss": 0.8076,
+      "step": 10557
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995987005541754e-06,
+      "loss": 1.0483,
+      "step": 10558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995891458054653e-06,
+      "loss": 1.0017,
+      "step": 10559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9957959105675524e-06,
+      "loss": 0.9167,
+      "step": 10560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9957003630804515e-06,
+      "loss": 1.1433,
+      "step": 10561
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9956048155933505e-06,
+      "loss": 0.8306,
+      "step": 10562
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995509268106249e-06,
+      "loss": 1.0846,
+      "step": 10563
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995413720619148e-06,
+      "loss": 0.9424,
+      "step": 10564
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995318173132047e-06,
+      "loss": 0.9011,
+      "step": 10565
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995222625644946e-06,
+      "loss": 1.4879,
+      "step": 10566
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995127078157845e-06,
+      "loss": 1.0641,
+      "step": 10567
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.995031530670744e-06,
+      "loss": 0.948,
+      "step": 10568
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994935983183643e-06,
+      "loss": 0.7693,
+      "step": 10569
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994840435696541e-06,
+      "loss": 1.0276,
+      "step": 10570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99474488820944e-06,
+      "loss": 1.0388,
+      "step": 10571
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994649340722339e-06,
+      "loss": 1.0735,
+      "step": 10572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994553793235238e-06,
+      "loss": 1.5222,
+      "step": 10573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994458245748137e-06,
+      "loss": 1.2979,
+      "step": 10574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994362698261036e-06,
+      "loss": 1.0569,
+      "step": 10575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994267150773935e-06,
+      "loss": 1.3509,
+      "step": 10576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9941716032868335e-06,
+      "loss": 1.1719,
+      "step": 10577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.994076055799733e-06,
+      "loss": 0.8176,
+      "step": 10578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993980508312632e-06,
+      "loss": 0.7583,
+      "step": 10579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993884960825531e-06,
+      "loss": 1.312,
+      "step": 10580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99378941333843e-06,
+      "loss": 0.9605,
+      "step": 10581
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993693865851329e-06,
+      "loss": 1.3001,
+      "step": 10582
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993598318364228e-06,
+      "loss": 0.9571,
+      "step": 10583
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993502770877126e-06,
+      "loss": 0.8563,
+      "step": 10584
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993407223390025e-06,
+      "loss": 0.9482,
+      "step": 10585
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993311675902924e-06,
+      "loss": 0.6356,
+      "step": 10586
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993216128415823e-06,
+      "loss": 1.1578,
+      "step": 10587
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993120580928722e-06,
+      "loss": 1.2685,
+      "step": 10588
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.993025033441621e-06,
+      "loss": 1.1483,
+      "step": 10589
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99292948595452e-06,
+      "loss": 0.99,
+      "step": 10590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992833938467418e-06,
+      "loss": 0.9661,
+      "step": 10591
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9927383909803174e-06,
+      "loss": 0.9471,
+      "step": 10592
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9926428434932165e-06,
+      "loss": 1.1727,
+      "step": 10593
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9925472960061155e-06,
+      "loss": 1.0035,
+      "step": 10594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9924517485190145e-06,
+      "loss": 1.143,
+      "step": 10595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992356201031914e-06,
+      "loss": 0.9745,
+      "step": 10596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992260653544812e-06,
+      "loss": 1.3535,
+      "step": 10597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992165106057711e-06,
+      "loss": 0.9994,
+      "step": 10598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99206955857061e-06,
+      "loss": 1.0311,
+      "step": 10599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991974011083509e-06,
+      "loss": 0.6429,
+      "step": 10600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991878463596408e-06,
+      "loss": 1.2531,
+      "step": 10601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991782916109307e-06,
+      "loss": 0.6915,
+      "step": 10602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991687368622206e-06,
+      "loss": 1.0786,
+      "step": 10603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991591821135104e-06,
+      "loss": 1.8617,
+      "step": 10604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991496273648003e-06,
+      "loss": 0.9791,
+      "step": 10605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991400726160902e-06,
+      "loss": 1.2213,
+      "step": 10606
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.991305178673801e-06,
+      "loss": 1.0109,
+      "step": 10607
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9912096311866995e-06,
+      "loss": 0.9768,
+      "step": 10608
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.991114083699599e-06,
+      "loss": 1.0759,
+      "step": 10609
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9910185362124984e-06,
+      "loss": 0.9974,
+      "step": 10610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990922988725397e-06,
+      "loss": 1.3125,
+      "step": 10611
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990827441238296e-06,
+      "loss": 0.9734,
+      "step": 10612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990731893751195e-06,
+      "loss": 1.3133,
+      "step": 10613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990636346264094e-06,
+      "loss": 0.7497,
+      "step": 10614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990540798776992e-06,
+      "loss": 1.5539,
+      "step": 10615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990445251289891e-06,
+      "loss": 1.2611,
+      "step": 10616
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990349703802791e-06,
+      "loss": 1.0785,
+      "step": 10617
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990254156315689e-06,
+      "loss": 0.9413,
+      "step": 10618
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990158608828588e-06,
+      "loss": 1.3739,
+      "step": 10619
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.990063061341487e-06,
+      "loss": 0.8135,
+      "step": 10620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989967513854386e-06,
+      "loss": 1.2107,
+      "step": 10621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989871966367284e-06,
+      "loss": 1.2315,
+      "step": 10622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989776418880183e-06,
+      "loss": 1.1358,
+      "step": 10623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989680871393083e-06,
+      "loss": 1.462,
+      "step": 10624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9895853239059815e-06,
+      "loss": 1.347,
+      "step": 10625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9894897764188805e-06,
+      "loss": 0.9509,
+      "step": 10626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9893942289317795e-06,
+      "loss": 0.6741,
+      "step": 10627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989298681444679e-06,
+      "loss": 1.212,
+      "step": 10628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989203133957577e-06,
+      "loss": 0.6827,
+      "step": 10629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989107586470476e-06,
+      "loss": 1.1444,
+      "step": 10630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.989012038983376e-06,
+      "loss": 1.2999,
+      "step": 10631
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988916491496274e-06,
+      "loss": 1.0867,
+      "step": 10632
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988820944009173e-06,
+      "loss": 1.4185,
+      "step": 10633
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988725396522072e-06,
+      "loss": 0.9776,
+      "step": 10634
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988629849034971e-06,
+      "loss": 1.3119,
+      "step": 10635
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988534301547869e-06,
+      "loss": 1.0884,
+      "step": 10636
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988438754060768e-06,
+      "loss": 0.9498,
+      "step": 10637
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988343206573667e-06,
+      "loss": 1.1973,
+      "step": 10638
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988247659086566e-06,
+      "loss": 0.9176,
+      "step": 10639
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988152111599465e-06,
+      "loss": 0.7704,
+      "step": 10640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.988056564112364e-06,
+      "loss": 1.3415,
+      "step": 10641
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9879610166252634e-06,
+      "loss": 0.9154,
+      "step": 10642
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987865469138162e-06,
+      "loss": 0.8827,
+      "step": 10643
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987769921651061e-06,
+      "loss": 0.7804,
+      "step": 10644
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.98767437416396e-06,
+      "loss": 1.042,
+      "step": 10645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987578826676859e-06,
+      "loss": 1.0166,
+      "step": 10646
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987483279189758e-06,
+      "loss": 1.0017,
+      "step": 10647
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987387731702657e-06,
+      "loss": 0.9261,
+      "step": 10648
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987292184215555e-06,
+      "loss": 1.1081,
+      "step": 10649
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987196636728454e-06,
+      "loss": 1.209,
+      "step": 10650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987101089241353e-06,
+      "loss": 1.184,
+      "step": 10651
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987005541754252e-06,
+      "loss": 0.6148,
+      "step": 10652
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986909994267151e-06,
+      "loss": 1.2781,
+      "step": 10653
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.98681444678005e-06,
+      "loss": 1.1996,
+      "step": 10654
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.986718899292949e-06,
+      "loss": 0.6481,
+      "step": 10655
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9866233518058474e-06,
+      "loss": 1.1552,
+      "step": 10656
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9865278043187465e-06,
+      "loss": 1.3798,
+      "step": 10657
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9864322568316455e-06,
+      "loss": 0.5106,
+      "step": 10658
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9863367093445445e-06,
+      "loss": 0.798,
+      "step": 10659
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.986241161857444e-06,
+      "loss": 1.1284,
+      "step": 10660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.986145614370343e-06,
+      "loss": 1.0522,
+      "step": 10661
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.986050066883242e-06,
+      "loss": 0.8191,
+      "step": 10662
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98595451939614e-06,
+      "loss": 1.5416,
+      "step": 10663
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985858971909039e-06,
+      "loss": 0.784,
+      "step": 10664
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985763424421938e-06,
+      "loss": 1.0319,
+      "step": 10665
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985667876934837e-06,
+      "loss": 1.5302,
+      "step": 10666
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985572329447736e-06,
+      "loss": 0.7656,
+      "step": 10667
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985476781960635e-06,
+      "loss": 1.0898,
+      "step": 10668
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985381234473534e-06,
+      "loss": 1.0713,
+      "step": 10669
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985285686986432e-06,
+      "loss": 0.7659,
+      "step": 10670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.985190139499331e-06,
+      "loss": 0.942,
+      "step": 10671
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98509459201223e-06,
+      "loss": 0.9263,
+      "step": 10672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984999044525129e-06,
+      "loss": 0.9359,
+      "step": 10673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9849034970380284e-06,
+      "loss": 1.5231,
+      "step": 10674
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9848079495509275e-06,
+      "loss": 0.9537,
+      "step": 10675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9847124020638265e-06,
+      "loss": 1.2731,
+      "step": 10676
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984616854576725e-06,
+      "loss": 1.6479,
+      "step": 10677
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984521307089624e-06,
+      "loss": 0.7235,
+      "step": 10678
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984425759602523e-06,
+      "loss": 0.728,
+      "step": 10679
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984330212115422e-06,
+      "loss": 0.8465,
+      "step": 10680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984234664628321e-06,
+      "loss": 1.3353,
+      "step": 10681
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98413911714122e-06,
+      "loss": 1.199,
+      "step": 10682
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.984043569654119e-06,
+      "loss": 0.8037,
+      "step": 10683
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983948022167017e-06,
+      "loss": 1.1987,
+      "step": 10684
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983852474679916e-06,
+      "loss": 1.1018,
+      "step": 10685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983756927192815e-06,
+      "loss": 1.2694,
+      "step": 10686
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983661379705714e-06,
+      "loss": 1.8945,
+      "step": 10687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9835658322186124e-06,
+      "loss": 1.3734,
+      "step": 10688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983470284731512e-06,
+      "loss": 1.2523,
+      "step": 10689
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9833747372444105e-06,
+      "loss": 1.6098,
+      "step": 10690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9832791897573095e-06,
+      "loss": 1.207,
+      "step": 10691
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983183642270209e-06,
+      "loss": 0.6863,
+      "step": 10692
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.983088094783108e-06,
+      "loss": 1.192,
+      "step": 10693
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982992547296007e-06,
+      "loss": 0.7844,
+      "step": 10694
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982896999808905e-06,
+      "loss": 1.1228,
+      "step": 10695
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982801452321805e-06,
+      "loss": 0.7174,
+      "step": 10696
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982705904834703e-06,
+      "loss": 1.1013,
+      "step": 10697
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982610357347602e-06,
+      "loss": 1.1255,
+      "step": 10698
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982514809860501e-06,
+      "loss": 0.9925,
+      "step": 10699
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9824192623734e-06,
+      "loss": 1.2386,
+      "step": 10700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982323714886298e-06,
+      "loss": 0.5395,
+      "step": 10701
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982228167399197e-06,
+      "loss": 1.1386,
+      "step": 10702
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982132619912097e-06,
+      "loss": 1.1874,
+      "step": 10703
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982037072424995e-06,
+      "loss": 1.1175,
+      "step": 10704
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981941524937894e-06,
+      "loss": 0.9246,
+      "step": 10705
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9818459774507934e-06,
+      "loss": 0.8267,
+      "step": 10706
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9817504299636925e-06,
+      "loss": 1.4719,
+      "step": 10707
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981654882476591e-06,
+      "loss": 1.0523,
+      "step": 10708
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98155933498949e-06,
+      "loss": 0.7599,
+      "step": 10709
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.98146378750239e-06,
+      "loss": 0.9279,
+      "step": 10710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981368240015288e-06,
+      "loss": 0.993,
+      "step": 10711
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981272692528187e-06,
+      "loss": 0.7053,
+      "step": 10712
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981177145041086e-06,
+      "loss": 1.5645,
+      "step": 10713
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.981081597553985e-06,
+      "loss": 1.0542,
+      "step": 10714
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980986050066883e-06,
+      "loss": 0.9846,
+      "step": 10715
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980890502579782e-06,
+      "loss": 0.9783,
+      "step": 10716
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980794955092682e-06,
+      "loss": 1.2021,
+      "step": 10717
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.98069940760558e-06,
+      "loss": 0.9336,
+      "step": 10718
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980603860118479e-06,
+      "loss": 0.5046,
+      "step": 10719
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980508312631378e-06,
+      "loss": 1.0601,
+      "step": 10720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980412765144277e-06,
+      "loss": 1.141,
+      "step": 10721
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9803172176571755e-06,
+      "loss": 1.0833,
+      "step": 10722
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9802216701700745e-06,
+      "loss": 0.7797,
+      "step": 10723
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980126122682974e-06,
+      "loss": 1.4556,
+      "step": 10724
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.980030575195873e-06,
+      "loss": 1.5619,
+      "step": 10725
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979935027708772e-06,
+      "loss": 0.8495,
+      "step": 10726
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979839480221671e-06,
+      "loss": 1.4567,
+      "step": 10727
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97974393273457e-06,
+      "loss": 0.8959,
+      "step": 10728
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979648385247468e-06,
+      "loss": 1.2617,
+      "step": 10729
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979552837760367e-06,
+      "loss": 0.7689,
+      "step": 10730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979457290273266e-06,
+      "loss": 1.1678,
+      "step": 10731
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979361742786165e-06,
+      "loss": 0.6864,
+      "step": 10732
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979266195299064e-06,
+      "loss": 0.7872,
+      "step": 10733
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979170647811963e-06,
+      "loss": 1.3079,
+      "step": 10734
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.979075100324862e-06,
+      "loss": 0.5537,
+      "step": 10735
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97897955283776e-06,
+      "loss": 1.0904,
+      "step": 10736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.978884005350659e-06,
+      "loss": 0.663,
+      "step": 10737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9787884578635584e-06,
+      "loss": 1.0097,
+      "step": 10738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9786929103764575e-06,
+      "loss": 1.2655,
+      "step": 10739
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9785973628893565e-06,
+      "loss": 0.812,
+      "step": 10740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9785018154022555e-06,
+      "loss": 1.3144,
+      "step": 10741
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.978406267915155e-06,
+      "loss": 1.2732,
+      "step": 10742
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.978310720428053e-06,
+      "loss": 1.086,
+      "step": 10743
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.978215172940952e-06,
+      "loss": 0.9166,
+      "step": 10744
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.978119625453851e-06,
+      "loss": 1.1517,
+      "step": 10745
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97802407796675e-06,
+      "loss": 1.3301,
+      "step": 10746
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977928530479649e-06,
+      "loss": 1.1053,
+      "step": 10747
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977832982992548e-06,
+      "loss": 1.2008,
+      "step": 10748
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977737435505446e-06,
+      "loss": 1.1178,
+      "step": 10749
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977641888018345e-06,
+      "loss": 1.0634,
+      "step": 10750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977546340531244e-06,
+      "loss": 1.2431,
+      "step": 10751
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977450793044143e-06,
+      "loss": 0.7889,
+      "step": 10752
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977355245557042e-06,
+      "loss": 0.8215,
+      "step": 10753
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977259698069941e-06,
+      "loss": 0.7229,
+      "step": 10754
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97716415058284e-06,
+      "loss": 1.266,
+      "step": 10755
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977068603095739e-06,
+      "loss": 0.5638,
+      "step": 10756
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976973055608638e-06,
+      "loss": 1.1617,
+      "step": 10757
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976877508121537e-06,
+      "loss": 0.8975,
+      "step": 10758
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976781960634436e-06,
+      "loss": 0.992,
+      "step": 10759
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976686413147334e-06,
+      "loss": 1.1629,
+      "step": 10760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976590865660234e-06,
+      "loss": 1.3059,
+      "step": 10761
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976495318173133e-06,
+      "loss": 1.7425,
+      "step": 10762
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976399770686031e-06,
+      "loss": 1.0239,
+      "step": 10763
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.97630422319893e-06,
+      "loss": 1.423,
+      "step": 10764
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.976208675711829e-06,
+      "loss": 1.4089,
+      "step": 10765
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.976113128224728e-06,
+      "loss": 0.7578,
+      "step": 10766
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.976017580737626e-06,
+      "loss": 1.1355,
+      "step": 10767
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975922033250526e-06,
+      "loss": 1.2827,
+      "step": 10768
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975826485763425e-06,
+      "loss": 1.2742,
+      "step": 10769
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9757309382763234e-06,
+      "loss": 0.9989,
+      "step": 10770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9756353907892225e-06,
+      "loss": 1.3373,
+      "step": 10771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9755398433021215e-06,
+      "loss": 1.243,
+      "step": 10772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9754442958150205e-06,
+      "loss": 1.1881,
+      "step": 10773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975348748327919e-06,
+      "loss": 1.2126,
+      "step": 10774
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975253200840819e-06,
+      "loss": 0.9614,
+      "step": 10775
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975157653353718e-06,
+      "loss": 1.1708,
+      "step": 10776
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.975062105866616e-06,
+      "loss": 1.5506,
+      "step": 10777
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974966558379515e-06,
+      "loss": 1.0372,
+      "step": 10778
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974871010892414e-06,
+      "loss": 0.828,
+      "step": 10779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974775463405313e-06,
+      "loss": 1.3529,
+      "step": 10780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974679915918211e-06,
+      "loss": 1.5529,
+      "step": 10781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974584368431111e-06,
+      "loss": 0.9259,
+      "step": 10782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.97448882094401e-06,
+      "loss": 0.9468,
+      "step": 10783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974393273456908e-06,
+      "loss": 1.0062,
+      "step": 10784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974297725969807e-06,
+      "loss": 0.8554,
+      "step": 10785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974202178482706e-06,
+      "loss": 0.7717,
+      "step": 10786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974106630995605e-06,
+      "loss": 0.5504,
+      "step": 10787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.974011083508504e-06,
+      "loss": 0.8256,
+      "step": 10788
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9739155360214035e-06,
+      "loss": 1.0378,
+      "step": 10789
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973819988534302e-06,
+      "loss": 0.7178,
+      "step": 10790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973724441047201e-06,
+      "loss": 1.2505,
+      "step": 10791
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9736288935601e-06,
+      "loss": 1.1421,
+      "step": 10792
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973533346072999e-06,
+      "loss": 1.1581,
+      "step": 10793
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973437798585898e-06,
+      "loss": 1.019,
+      "step": 10794
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973342251098796e-06,
+      "loss": 0.8678,
+      "step": 10795
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973246703611695e-06,
+      "loss": 0.7772,
+      "step": 10796
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973151156124594e-06,
+      "loss": 0.9193,
+      "step": 10797
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.973055608637493e-06,
+      "loss": 1.0616,
+      "step": 10798
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972960061150392e-06,
+      "loss": 0.8891,
+      "step": 10799
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972864513663291e-06,
+      "loss": 2.0096,
+      "step": 10800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972768966176189e-06,
+      "loss": 0.737,
+      "step": 10801
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9726734186890884e-06,
+      "loss": 0.766,
+      "step": 10802
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9725778712019875e-06,
+      "loss": 0.9746,
+      "step": 10803
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9724823237148865e-06,
+      "loss": 1.0257,
+      "step": 10804
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9723867762277855e-06,
+      "loss": 0.9195,
+      "step": 10805
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972291228740685e-06,
+      "loss": 1.4752,
+      "step": 10806
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972195681253584e-06,
+      "loss": 1.3671,
+      "step": 10807
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972100133766482e-06,
+      "loss": 1.161,
+      "step": 10808
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.972004586279381e-06,
+      "loss": 0.9793,
+      "step": 10809
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.97190903879228e-06,
+      "loss": 0.5299,
+      "step": 10810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971813491305179e-06,
+      "loss": 0.8746,
+      "step": 10811
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971717943818078e-06,
+      "loss": 0.9115,
+      "step": 10812
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971622396330977e-06,
+      "loss": 0.6178,
+      "step": 10813
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971526848843876e-06,
+      "loss": 1.3289,
+      "step": 10814
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971431301356774e-06,
+      "loss": 0.9807,
+      "step": 10815
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971335753869673e-06,
+      "loss": 1.1145,
+      "step": 10816
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.971240206382572e-06,
+      "loss": 0.7213,
+      "step": 10817
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.971144658895471e-06,
+      "loss": 1.3627,
+      "step": 10818
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.97104911140837e-06,
+      "loss": 1.3662,
+      "step": 10819
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9709535639212694e-06,
+      "loss": 1.3398,
+      "step": 10820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9708580164341685e-06,
+      "loss": 1.039,
+      "step": 10821
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970762468947067e-06,
+      "loss": 1.3275,
+      "step": 10822
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970666921459966e-06,
+      "loss": 1.1476,
+      "step": 10823
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970571373972865e-06,
+      "loss": 1.0273,
+      "step": 10824
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970475826485764e-06,
+      "loss": 1.4732,
+      "step": 10825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970380278998663e-06,
+      "loss": 0.9352,
+      "step": 10826
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970284731511562e-06,
+      "loss": 0.9996,
+      "step": 10827
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970189184024461e-06,
+      "loss": 1.2037,
+      "step": 10828
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.970093636537359e-06,
+      "loss": 0.9772,
+      "step": 10829
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969998089050258e-06,
+      "loss": 1.0797,
+      "step": 10830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969902541563157e-06,
+      "loss": 0.804,
+      "step": 10831
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969806994076056e-06,
+      "loss": 0.9393,
+      "step": 10832
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969711446588955e-06,
+      "loss": 0.9137,
+      "step": 10833
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969615899101854e-06,
+      "loss": 1.1247,
+      "step": 10834
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969520351614753e-06,
+      "loss": 0.838,
+      "step": 10835
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9694248041276515e-06,
+      "loss": 0.7852,
+      "step": 10836
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9693292566405505e-06,
+      "loss": 1.2684,
+      "step": 10837
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.96923370915345e-06,
+      "loss": 1.0076,
+      "step": 10838
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969138161666349e-06,
+      "loss": 1.1278,
+      "step": 10839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.969042614179248e-06,
+      "loss": 1.1481,
+      "step": 10840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968947066692147e-06,
+      "loss": 0.9602,
+      "step": 10841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968851519205045e-06,
+      "loss": 0.7592,
+      "step": 10842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968755971717944e-06,
+      "loss": 0.7985,
+      "step": 10843
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968660424230843e-06,
+      "loss": 1.0573,
+      "step": 10844
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968564876743742e-06,
+      "loss": 1.3215,
+      "step": 10845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968469329256641e-06,
+      "loss": 1.1864,
+      "step": 10846
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.96837378176954e-06,
+      "loss": 0.697,
+      "step": 10847
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968278234282439e-06,
+      "loss": 1.3255,
+      "step": 10848
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968182686795337e-06,
+      "loss": 1.2321,
+      "step": 10849
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.968087139308236e-06,
+      "loss": 0.8637,
+      "step": 10850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967991591821135e-06,
+      "loss": 1.3656,
+      "step": 10851
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9678960443340344e-06,
+      "loss": 0.6032,
+      "step": 10852
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967800496846933e-06,
+      "loss": 0.9206,
+      "step": 10853
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9677049493598325e-06,
+      "loss": 1.5005,
+      "step": 10854
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9676094018727315e-06,
+      "loss": 1.2377,
+      "step": 10855
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.96751385438563e-06,
+      "loss": 1.0586,
+      "step": 10856
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967418306898529e-06,
+      "loss": 1.1304,
+      "step": 10857
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967322759411428e-06,
+      "loss": 1.163,
+      "step": 10858
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967227211924327e-06,
+      "loss": 1.1011,
+      "step": 10859
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967131664437225e-06,
+      "loss": 1.4923,
+      "step": 10860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.967036116950125e-06,
+      "loss": 1.1037,
+      "step": 10861
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966940569463024e-06,
+      "loss": 1.2892,
+      "step": 10862
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966845021975922e-06,
+      "loss": 0.6073,
+      "step": 10863
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966749474488821e-06,
+      "loss": 0.875,
+      "step": 10864
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.96665392700172e-06,
+      "loss": 1.0887,
+      "step": 10865
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966558379514619e-06,
+      "loss": 1.3343,
+      "step": 10866
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9664628320275175e-06,
+      "loss": 0.8608,
+      "step": 10867
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9663672845404165e-06,
+      "loss": 1.0831,
+      "step": 10868
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.966271737053316e-06,
+      "loss": 0.7335,
+      "step": 10869
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.966176189566215e-06,
+      "loss": 0.9791,
+      "step": 10870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.966080642079114e-06,
+      "loss": 0.8955,
+      "step": 10871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965985094592013e-06,
+      "loss": 1.1851,
+      "step": 10872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965889547104912e-06,
+      "loss": 0.6842,
+      "step": 10873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96579399961781e-06,
+      "loss": 0.9145,
+      "step": 10874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965698452130709e-06,
+      "loss": 1.0902,
+      "step": 10875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965602904643609e-06,
+      "loss": 0.9978,
+      "step": 10876
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965507357156507e-06,
+      "loss": 0.9744,
+      "step": 10877
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965411809669406e-06,
+      "loss": 1.0546,
+      "step": 10878
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965316262182305e-06,
+      "loss": 0.9535,
+      "step": 10879
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965220714695204e-06,
+      "loss": 0.8564,
+      "step": 10880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965125167208102e-06,
+      "loss": 0.965,
+      "step": 10881
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.965029619721001e-06,
+      "loss": 0.8514,
+      "step": 10882
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9649340722339e-06,
+      "loss": 1.5582,
+      "step": 10883
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9648385247467994e-06,
+      "loss": 1.3769,
+      "step": 10884
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9647429772596985e-06,
+      "loss": 1.1243,
+      "step": 10885
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9646474297725975e-06,
+      "loss": 0.8013,
+      "step": 10886
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9645518822854965e-06,
+      "loss": 1.2995,
+      "step": 10887
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.964456334798395e-06,
+      "loss": 1.0829,
+      "step": 10888
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.964360787311294e-06,
+      "loss": 0.9513,
+      "step": 10889
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.964265239824193e-06,
+      "loss": 1.4975,
+      "step": 10890
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.964169692337092e-06,
+      "loss": 1.4484,
+      "step": 10891
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.964074144849991e-06,
+      "loss": 1.2779,
+      "step": 10892
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96397859736289e-06,
+      "loss": 1.4855,
+      "step": 10893
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963883049875788e-06,
+      "loss": 1.0983,
+      "step": 10894
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963787502388687e-06,
+      "loss": 1.5016,
+      "step": 10895
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963691954901586e-06,
+      "loss": 1.3979,
+      "step": 10896
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963596407414485e-06,
+      "loss": 0.8654,
+      "step": 10897
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963500859927384e-06,
+      "loss": 1.0718,
+      "step": 10898
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963405312440283e-06,
+      "loss": 1.0672,
+      "step": 10899
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963309764953182e-06,
+      "loss": 1.1393,
+      "step": 10900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9632142174660805e-06,
+      "loss": 1.0391,
+      "step": 10901
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96311866997898e-06,
+      "loss": 0.9518,
+      "step": 10902
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963023122491879e-06,
+      "loss": 1.1551,
+      "step": 10903
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962927575004778e-06,
+      "loss": 0.9207,
+      "step": 10904
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962832027517677e-06,
+      "loss": 1.0257,
+      "step": 10905
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962736480030576e-06,
+      "loss": 1.1265,
+      "step": 10906
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962640932543475e-06,
+      "loss": 0.9283,
+      "step": 10907
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962545385056373e-06,
+      "loss": 1.0482,
+      "step": 10908
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962449837569272e-06,
+      "loss": 0.7133,
+      "step": 10909
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962354290082171e-06,
+      "loss": 1.3264,
+      "step": 10910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96225874259507e-06,
+      "loss": 0.7584,
+      "step": 10911
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962163195107969e-06,
+      "loss": 0.7284,
+      "step": 10912
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962067647620868e-06,
+      "loss": 0.8903,
+      "step": 10913
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.961972100133767e-06,
+      "loss": 1.1265,
+      "step": 10914
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.961876552646665e-06,
+      "loss": 1.357,
+      "step": 10915
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9617810051595644e-06,
+      "loss": 0.5675,
+      "step": 10916
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9616854576724635e-06,
+      "loss": 1.2875,
+      "step": 10917
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9615899101853625e-06,
+      "loss": 0.8573,
+      "step": 10918
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9614943626982615e-06,
+      "loss": 1.2204,
+      "step": 10919
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.961398815211161e-06,
+      "loss": 1.4513,
+      "step": 10920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96130326772406e-06,
+      "loss": 1.1925,
+      "step": 10921
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961207720236958e-06,
+      "loss": 0.9134,
+      "step": 10922
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961112172749857e-06,
+      "loss": 1.1859,
+      "step": 10923
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.961016625262756e-06,
+      "loss": 0.9169,
+      "step": 10924
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960921077775655e-06,
+      "loss": 0.9362,
+      "step": 10925
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960825530288554e-06,
+      "loss": 1.1262,
+      "step": 10926
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960729982801453e-06,
+      "loss": 1.3536,
+      "step": 10927
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960634435314352e-06,
+      "loss": 0.8721,
+      "step": 10928
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.96053888782725e-06,
+      "loss": 0.9918,
+      "step": 10929
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960443340340149e-06,
+      "loss": 1.0449,
+      "step": 10930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960347792853048e-06,
+      "loss": 0.7323,
+      "step": 10931
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960252245365947e-06,
+      "loss": 1.2121,
+      "step": 10932
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.960156697878846e-06,
+      "loss": 0.642,
+      "step": 10933
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9600611503917454e-06,
+      "loss": 1.0943,
+      "step": 10934
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9599656029046445e-06,
+      "loss": 1.3189,
+      "step": 10935
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959870055417543e-06,
+      "loss": 0.7256,
+      "step": 10936
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959774507930442e-06,
+      "loss": 1.4032,
+      "step": 10937
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959678960443341e-06,
+      "loss": 0.9345,
+      "step": 10938
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.95958341295624e-06,
+      "loss": 1.1324,
+      "step": 10939
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959487865469138e-06,
+      "loss": 1.0819,
+      "step": 10940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959392317982038e-06,
+      "loss": 0.6528,
+      "step": 10941
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959296770494936e-06,
+      "loss": 1.3588,
+      "step": 10942
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959201223007835e-06,
+      "loss": 0.6685,
+      "step": 10943
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959105675520734e-06,
+      "loss": 0.5853,
+      "step": 10944
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.959010128033633e-06,
+      "loss": 1.2931,
+      "step": 10945
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958914580546531e-06,
+      "loss": 0.9289,
+      "step": 10946
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.95881903305943e-06,
+      "loss": 0.6579,
+      "step": 10947
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.95872348557233e-06,
+      "loss": 0.9652,
+      "step": 10948
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9586279380852285e-06,
+      "loss": 1.0585,
+      "step": 10949
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9585323905981275e-06,
+      "loss": 1.2565,
+      "step": 10950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9584368431110265e-06,
+      "loss": 0.9317,
+      "step": 10951
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958341295623926e-06,
+      "loss": 0.8605,
+      "step": 10952
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958245748136824e-06,
+      "loss": 0.8609,
+      "step": 10953
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958150200649723e-06,
+      "loss": 0.753,
+      "step": 10954
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.958054653162623e-06,
+      "loss": 1.388,
+      "step": 10955
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957959105675521e-06,
+      "loss": 1.4615,
+      "step": 10956
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.95786355818842e-06,
+      "loss": 0.9731,
+      "step": 10957
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957768010701319e-06,
+      "loss": 1.4174,
+      "step": 10958
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957672463214218e-06,
+      "loss": 0.7769,
+      "step": 10959
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957576915727116e-06,
+      "loss": 0.8057,
+      "step": 10960
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957481368240015e-06,
+      "loss": 0.7444,
+      "step": 10961
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957385820752915e-06,
+      "loss": 1.4623,
+      "step": 10962
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957290273265813e-06,
+      "loss": 1.326,
+      "step": 10963
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957194725778712e-06,
+      "loss": 1.2867,
+      "step": 10964
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957099178291611e-06,
+      "loss": 0.6597,
+      "step": 10965
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9570036308045104e-06,
+      "loss": 0.8979,
+      "step": 10966
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956908083317409e-06,
+      "loss": 1.2981,
+      "step": 10967
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956812535830308e-06,
+      "loss": 0.9961,
+      "step": 10968
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9567169883432075e-06,
+      "loss": 1.1373,
+      "step": 10969
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956621440856106e-06,
+      "loss": 1.1535,
+      "step": 10970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956525893369005e-06,
+      "loss": 1.0312,
+      "step": 10971
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956430345881904e-06,
+      "loss": 1.118,
+      "step": 10972
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956334798394803e-06,
+      "loss": 0.9252,
+      "step": 10973
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.956239250907701e-06,
+      "loss": 1.3936,
+      "step": 10974
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9561437034206e-06,
+      "loss": 0.9791,
+      "step": 10975
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.956048155933499e-06,
+      "loss": 1.1014,
+      "step": 10976
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955952608446398e-06,
+      "loss": 0.7411,
+      "step": 10977
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955857060959297e-06,
+      "loss": 0.9695,
+      "step": 10978
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955761513472196e-06,
+      "loss": 0.6839,
+      "step": 10979
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955665965985095e-06,
+      "loss": 1.3653,
+      "step": 10980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9555704184979935e-06,
+      "loss": 0.9032,
+      "step": 10981
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9554748710108925e-06,
+      "loss": 1.1927,
+      "step": 10982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9553793235237915e-06,
+      "loss": 1.1927,
+      "step": 10983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955283776036691e-06,
+      "loss": 0.6578,
+      "step": 10984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.95518822854959e-06,
+      "loss": 1.3181,
+      "step": 10985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.955092681062489e-06,
+      "loss": 1.1277,
+      "step": 10986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954997133575388e-06,
+      "loss": 1.0284,
+      "step": 10987
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954901586088286e-06,
+      "loss": 1.0463,
+      "step": 10988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954806038601185e-06,
+      "loss": 1.1305,
+      "step": 10989
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954710491114084e-06,
+      "loss": 0.7863,
+      "step": 10990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954614943626983e-06,
+      "loss": 1.1294,
+      "step": 10991
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954519396139882e-06,
+      "loss": 0.6332,
+      "step": 10992
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954423848652781e-06,
+      "loss": 0.874,
+      "step": 10993
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954328301165679e-06,
+      "loss": 1.0345,
+      "step": 10994
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954232753678578e-06,
+      "loss": 1.2678,
+      "step": 10995
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954137206191477e-06,
+      "loss": 1.1169,
+      "step": 10996
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.954041658704376e-06,
+      "loss": 0.7393,
+      "step": 10997
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9539461112172754e-06,
+      "loss": 1.0023,
+      "step": 10998
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9538505637301745e-06,
+      "loss": 1.2506,
+      "step": 10999
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9537550162430735e-06,
+      "loss": 0.6995,
+      "step": 11000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953659468755972e-06,
+      "loss": 0.9928,
+      "step": 11001
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953563921268871e-06,
+      "loss": 0.7374,
+      "step": 11002
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.95346837378177e-06,
+      "loss": 0.9669,
+      "step": 11003
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953372826294669e-06,
+      "loss": 1.0238,
+      "step": 11004
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953277278807568e-06,
+      "loss": 1.4882,
+      "step": 11005
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953181731320467e-06,
+      "loss": 0.8641,
+      "step": 11006
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.953086183833366e-06,
+      "loss": 0.9847,
+      "step": 11007
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952990636346264e-06,
+      "loss": 1.086,
+      "step": 11008
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952895088859163e-06,
+      "loss": 0.6419,
+      "step": 11009
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952799541372062e-06,
+      "loss": 0.9146,
+      "step": 11010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952703993884961e-06,
+      "loss": 1.0697,
+      "step": 11011
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9526084463978594e-06,
+      "loss": 1.1219,
+      "step": 11012
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952512898910759e-06,
+      "loss": 1.06,
+      "step": 11013
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952417351423658e-06,
+      "loss": 1.2306,
+      "step": 11014
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9523218039365565e-06,
+      "loss": 1.5329,
+      "step": 11015
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952226256449456e-06,
+      "loss": 0.7289,
+      "step": 11016
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952130708962355e-06,
+      "loss": 1.3245,
+      "step": 11017
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.952035161475254e-06,
+      "loss": 1.2092,
+      "step": 11018
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951939613988152e-06,
+      "loss": 1.2263,
+      "step": 11019
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951844066501052e-06,
+      "loss": 0.8715,
+      "step": 11020
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951748519013951e-06,
+      "loss": 1.0341,
+      "step": 11021
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951652971526849e-06,
+      "loss": 0.8073,
+      "step": 11022
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951557424039748e-06,
+      "loss": 1.0632,
+      "step": 11023
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951461876552647e-06,
+      "loss": 0.7077,
+      "step": 11024
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951366329065546e-06,
+      "loss": 1.2739,
+      "step": 11025
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.951270781578444e-06,
+      "loss": 0.6883,
+      "step": 11026
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.951175234091344e-06,
+      "loss": 0.9337,
+      "step": 11027
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.951079686604243e-06,
+      "loss": 1.0846,
+      "step": 11028
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950984139117141e-06,
+      "loss": 0.8518,
+      "step": 11029
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9508885916300404e-06,
+      "loss": 0.9318,
+      "step": 11030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9507930441429395e-06,
+      "loss": 1.1573,
+      "step": 11031
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9506974966558385e-06,
+      "loss": 0.8726,
+      "step": 11032
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950601949168737e-06,
+      "loss": 0.9344,
+      "step": 11033
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950506401681637e-06,
+      "loss": 0.9973,
+      "step": 11034
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950410854194535e-06,
+      "loss": 1.2085,
+      "step": 11035
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950315306707434e-06,
+      "loss": 0.8008,
+      "step": 11036
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950219759220333e-06,
+      "loss": 0.9628,
+      "step": 11037
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950124211733232e-06,
+      "loss": 0.8761,
+      "step": 11038
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.950028664246131e-06,
+      "loss": 0.9944,
+      "step": 11039
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949933116759029e-06,
+      "loss": 0.7725,
+      "step": 11040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949837569271929e-06,
+      "loss": 1.0753,
+      "step": 11041
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949742021784827e-06,
+      "loss": 0.6781,
+      "step": 11042
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949646474297726e-06,
+      "loss": 1.1913,
+      "step": 11043
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949550926810625e-06,
+      "loss": 0.9555,
+      "step": 11044
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949455379323524e-06,
+      "loss": 1.4107,
+      "step": 11045
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9493598318364225e-06,
+      "loss": 0.7474,
+      "step": 11046
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9492642843493215e-06,
+      "loss": 0.9745,
+      "step": 11047
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.949168736862221e-06,
+      "loss": 1.4854,
+      "step": 11048
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94907318937512e-06,
+      "loss": 0.7469,
+      "step": 11049
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948977641888019e-06,
+      "loss": 1.4201,
+      "step": 11050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948882094400918e-06,
+      "loss": 0.9427,
+      "step": 11051
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948786546913817e-06,
+      "loss": 0.9632,
+      "step": 11052
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948690999426715e-06,
+      "loss": 1.276,
+      "step": 11053
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948595451939614e-06,
+      "loss": 1.0802,
+      "step": 11054
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948499904452513e-06,
+      "loss": 1.0182,
+      "step": 11055
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948404356965412e-06,
+      "loss": 1.0654,
+      "step": 11056
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948308809478311e-06,
+      "loss": 1.3209,
+      "step": 11057
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94821326199121e-06,
+      "loss": 0.8816,
+      "step": 11058
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948117714504109e-06,
+      "loss": 0.7716,
+      "step": 11059
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.948022167017007e-06,
+      "loss": 0.7088,
+      "step": 11060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947926619529906e-06,
+      "loss": 0.6771,
+      "step": 11061
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9478310720428054e-06,
+      "loss": 1.1351,
+      "step": 11062
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9477355245557045e-06,
+      "loss": 0.8515,
+      "step": 11063
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9476399770686035e-06,
+      "loss": 0.9073,
+      "step": 11064
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9475444295815025e-06,
+      "loss": 1.1118,
+      "step": 11065
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947448882094402e-06,
+      "loss": 1.3355,
+      "step": 11066
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9473533346073e-06,
+      "loss": 1.2199,
+      "step": 11067
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947257787120199e-06,
+      "loss": 1.1291,
+      "step": 11068
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947162239633098e-06,
+      "loss": 1.7701,
+      "step": 11069
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.947066692145997e-06,
+      "loss": 0.9877,
+      "step": 11070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946971144658896e-06,
+      "loss": 0.8236,
+      "step": 11071
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946875597171795e-06,
+      "loss": 0.729,
+      "step": 11072
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946780049684694e-06,
+      "loss": 1.6092,
+      "step": 11073
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946684502197592e-06,
+      "loss": 0.9137,
+      "step": 11074
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946588954710491e-06,
+      "loss": 0.9264,
+      "step": 11075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94649340722339e-06,
+      "loss": 0.9825,
+      "step": 11076
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946397859736289e-06,
+      "loss": 1.2316,
+      "step": 11077
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946302312249188e-06,
+      "loss": 0.8187,
+      "step": 11078
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.946206764762087e-06,
+      "loss": 0.9858,
+      "step": 11079
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9461112172749864e-06,
+      "loss": 0.5691,
+      "step": 11080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.946015669787885e-06,
+      "loss": 1.032,
+      "step": 11081
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945920122300784e-06,
+      "loss": 0.9075,
+      "step": 11082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945824574813683e-06,
+      "loss": 0.9063,
+      "step": 11083
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945729027326582e-06,
+      "loss": 1.1825,
+      "step": 11084
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945633479839481e-06,
+      "loss": 0.95,
+      "step": 11085
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94553793235238e-06,
+      "loss": 1.4702,
+      "step": 11086
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945442384865278e-06,
+      "loss": 1.4568,
+      "step": 11087
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945346837378177e-06,
+      "loss": 0.9093,
+      "step": 11088
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945251289891076e-06,
+      "loss": 0.9997,
+      "step": 11089
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945155742403975e-06,
+      "loss": 0.9869,
+      "step": 11090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.945060194916874e-06,
+      "loss": 1.3598,
+      "step": 11091
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944964647429773e-06,
+      "loss": 1.2297,
+      "step": 11092
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944869099942672e-06,
+      "loss": 1.2527,
+      "step": 11093
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94477355245557e-06,
+      "loss": 1.3764,
+      "step": 11094
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9446780049684695e-06,
+      "loss": 1.1949,
+      "step": 11095
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9445824574813685e-06,
+      "loss": 1.7714,
+      "step": 11096
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9444869099942675e-06,
+      "loss": 1.2307,
+      "step": 11097
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944391362507166e-06,
+      "loss": 0.946,
+      "step": 11098
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944295815020066e-06,
+      "loss": 1.1662,
+      "step": 11099
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944200267532965e-06,
+      "loss": 0.8259,
+      "step": 11100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944104720045863e-06,
+      "loss": 1.1969,
+      "step": 11101
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.944009172558762e-06,
+      "loss": 0.5613,
+      "step": 11102
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943913625071661e-06,
+      "loss": 1.081,
+      "step": 11103
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94381807758456e-06,
+      "loss": 1.1682,
+      "step": 11104
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943722530097458e-06,
+      "loss": 0.9564,
+      "step": 11105
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943626982610358e-06,
+      "loss": 1.0115,
+      "step": 11106
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943531435123257e-06,
+      "loss": 0.8646,
+      "step": 11107
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943435887636155e-06,
+      "loss": 0.9544,
+      "step": 11108
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943340340149054e-06,
+      "loss": 1.1948,
+      "step": 11109
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943244792661953e-06,
+      "loss": 1.1246,
+      "step": 11110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943149245174852e-06,
+      "loss": 1.3584,
+      "step": 11111
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.943053697687751e-06,
+      "loss": 1.0108,
+      "step": 11112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9429581502006505e-06,
+      "loss": 1.233,
+      "step": 11113
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9428626027135495e-06,
+      "loss": 1.1611,
+      "step": 11114
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942767055226448e-06,
+      "loss": 1.1697,
+      "step": 11115
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942671507739347e-06,
+      "loss": 1.0964,
+      "step": 11116
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942575960252246e-06,
+      "loss": 1.0775,
+      "step": 11117
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942480412765145e-06,
+      "loss": 0.8889,
+      "step": 11118
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942384865278043e-06,
+      "loss": 1.022,
+      "step": 11119
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942289317790942e-06,
+      "loss": 0.604,
+      "step": 11120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942193770303842e-06,
+      "loss": 0.8595,
+      "step": 11121
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.94209822281674e-06,
+      "loss": 0.787,
+      "step": 11122
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942002675329639e-06,
+      "loss": 0.9432,
+      "step": 11123
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941907127842538e-06,
+      "loss": 1.0135,
+      "step": 11124
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941811580355437e-06,
+      "loss": 0.6409,
+      "step": 11125
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941716032868335e-06,
+      "loss": 1.0374,
+      "step": 11126
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9416204853812345e-06,
+      "loss": 1.2576,
+      "step": 11127
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941524937894134e-06,
+      "loss": 1.1055,
+      "step": 11128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9414293904070325e-06,
+      "loss": 1.5658,
+      "step": 11129
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941333842919932e-06,
+      "loss": 0.8673,
+      "step": 11130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.941238295432831e-06,
+      "loss": 1.0582,
+      "step": 11131
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94114274794573e-06,
+      "loss": 0.7804,
+      "step": 11132
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.941047200458628e-06,
+      "loss": 1.6227,
+      "step": 11133
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940951652971527e-06,
+      "loss": 0.9441,
+      "step": 11134
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940856105484426e-06,
+      "loss": 0.9711,
+      "step": 11135
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940760557997325e-06,
+      "loss": 0.986,
+      "step": 11136
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940665010510224e-06,
+      "loss": 1.5139,
+      "step": 11137
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940569463023123e-06,
+      "loss": 0.9273,
+      "step": 11138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940473915536022e-06,
+      "loss": 1.1317,
+      "step": 11139
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94037836804892e-06,
+      "loss": 1.3337,
+      "step": 11140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940282820561819e-06,
+      "loss": 1.0974,
+      "step": 11141
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940187273074718e-06,
+      "loss": 0.8639,
+      "step": 11142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.940091725587617e-06,
+      "loss": 0.666,
+      "step": 11143
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939996178100516e-06,
+      "loss": 1.5116,
+      "step": 11144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9399006306134155e-06,
+      "loss": 0.9769,
+      "step": 11145
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939805083126314e-06,
+      "loss": 1.3092,
+      "step": 11146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939709535639213e-06,
+      "loss": 1.1246,
+      "step": 11147
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939613988152112e-06,
+      "loss": 1.2017,
+      "step": 11148
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939518440665011e-06,
+      "loss": 0.457,
+      "step": 11149
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.93942289317791e-06,
+      "loss": 0.6856,
+      "step": 11150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939327345690809e-06,
+      "loss": 1.2021,
+      "step": 11151
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939231798203708e-06,
+      "loss": 1.1244,
+      "step": 11152
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939136250716606e-06,
+      "loss": 0.8461,
+      "step": 11153
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.939040703229505e-06,
+      "loss": 1.2047,
+      "step": 11154
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938945155742404e-06,
+      "loss": 0.8397,
+      "step": 11155
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938849608255303e-06,
+      "loss": 0.8263,
+      "step": 11156
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938754060768202e-06,
+      "loss": 1.9661,
+      "step": 11157
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938658513281101e-06,
+      "loss": 0.9588,
+      "step": 11158
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938562965794e-06,
+      "loss": 0.8344,
+      "step": 11159
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9384674183068985e-06,
+      "loss": 0.5791,
+      "step": 11160
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9383718708197975e-06,
+      "loss": 1.0296,
+      "step": 11161
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9382763233326966e-06,
+      "loss": 0.6292,
+      "step": 11162
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938180775845596e-06,
+      "loss": 0.8883,
+      "step": 11163
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.938085228358495e-06,
+      "loss": 0.7322,
+      "step": 11164
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937989680871394e-06,
+      "loss": 0.9736,
+      "step": 11165
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937894133384293e-06,
+      "loss": 1.2364,
+      "step": 11166
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937798585897191e-06,
+      "loss": 0.9885,
+      "step": 11167
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.93770303841009e-06,
+      "loss": 0.9768,
+      "step": 11168
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937607490922989e-06,
+      "loss": 1.0906,
+      "step": 11169
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937511943435888e-06,
+      "loss": 0.9036,
+      "step": 11170
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937416395948787e-06,
+      "loss": 1.2506,
+      "step": 11171
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937320848461686e-06,
+      "loss": 1.1225,
+      "step": 11172
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937225300974585e-06,
+      "loss": 1.0992,
+      "step": 11173
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937129753487483e-06,
+      "loss": 1.0672,
+      "step": 11174
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937034206000382e-06,
+      "loss": 0.7819,
+      "step": 11175
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936938658513281e-06,
+      "loss": 0.7322,
+      "step": 11176
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9368431110261805e-06,
+      "loss": 0.5952,
+      "step": 11177
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9367475635390795e-06,
+      "loss": 1.356,
+      "step": 11178
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9366520160519785e-06,
+      "loss": 1.1194,
+      "step": 11179
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9365564685648776e-06,
+      "loss": 0.947,
+      "step": 11180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936460921077776e-06,
+      "loss": 1.5213,
+      "step": 11181
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936365373590675e-06,
+      "loss": 1.2585,
+      "step": 11182
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936269826103574e-06,
+      "loss": 1.2355,
+      "step": 11183
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.936174278616473e-06,
+      "loss": 0.9915,
+      "step": 11184
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.936078731129372e-06,
+      "loss": 1.3368,
+      "step": 11185
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935983183642271e-06,
+      "loss": 1.4264,
+      "step": 11186
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935887636155169e-06,
+      "loss": 1.4411,
+      "step": 11187
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935792088668068e-06,
+      "loss": 0.7003,
+      "step": 11188
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935696541180967e-06,
+      "loss": 0.9816,
+      "step": 11189
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935600993693866e-06,
+      "loss": 1.3835,
+      "step": 11190
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935505446206765e-06,
+      "loss": 0.9108,
+      "step": 11191
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9354098987196635e-06,
+      "loss": 1.0569,
+      "step": 11192
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935314351232563e-06,
+      "loss": 1.5374,
+      "step": 11193
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9352188037454616e-06,
+      "loss": 0.9098,
+      "step": 11194
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.935123256258361e-06,
+      "loss": 1.0954,
+      "step": 11195
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.93502770877126e-06,
+      "loss": 1.1685,
+      "step": 11196
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934932161284159e-06,
+      "loss": 1.4117,
+      "step": 11197
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934836613797057e-06,
+      "loss": 1.1453,
+      "step": 11198
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934741066309956e-06,
+      "loss": 0.6638,
+      "step": 11199
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934645518822856e-06,
+      "loss": 1.0284,
+      "step": 11200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934549971335754e-06,
+      "loss": 1.1961,
+      "step": 11201
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934454423848653e-06,
+      "loss": 1.2546,
+      "step": 11202
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934358876361552e-06,
+      "loss": 0.9621,
+      "step": 11203
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934263328874451e-06,
+      "loss": 1.1676,
+      "step": 11204
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934167781387349e-06,
+      "loss": 0.7829,
+      "step": 11205
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.934072233900248e-06,
+      "loss": 1.2283,
+      "step": 11206
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933976686413148e-06,
+      "loss": 1.0886,
+      "step": 11207
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933881138926046e-06,
+      "loss": 0.8268,
+      "step": 11208
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9337855914389455e-06,
+      "loss": 1.099,
+      "step": 11209
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9336900439518445e-06,
+      "loss": 0.8772,
+      "step": 11210
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9335944964647435e-06,
+      "loss": 1.1239,
+      "step": 11211
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933498948977642e-06,
+      "loss": 1.2338,
+      "step": 11212
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933403401490541e-06,
+      "loss": 0.8305,
+      "step": 11213
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933307854003441e-06,
+      "loss": 0.6475,
+      "step": 11214
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933212306516339e-06,
+      "loss": 1.2868,
+      "step": 11215
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933116759029238e-06,
+      "loss": 0.9866,
+      "step": 11216
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.933021211542137e-06,
+      "loss": 0.846,
+      "step": 11217
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932925664055036e-06,
+      "loss": 0.9531,
+      "step": 11218
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932830116567934e-06,
+      "loss": 0.6476,
+      "step": 11219
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932734569080833e-06,
+      "loss": 1.3528,
+      "step": 11220
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932639021593733e-06,
+      "loss": 1.0123,
+      "step": 11221
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932543474106631e-06,
+      "loss": 1.4578,
+      "step": 11222
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.93244792661953e-06,
+      "loss": 0.8357,
+      "step": 11223
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932352379132429e-06,
+      "loss": 0.8053,
+      "step": 11224
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932256831645328e-06,
+      "loss": 0.8074,
+      "step": 11225
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9321612841582266e-06,
+      "loss": 0.9971,
+      "step": 11226
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.932065736671126e-06,
+      "loss": 1.2484,
+      "step": 11227
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931970189184025e-06,
+      "loss": 0.9803,
+      "step": 11228
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931874641696924e-06,
+      "loss": 1.0535,
+      "step": 11229
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931779094209823e-06,
+      "loss": 0.9759,
+      "step": 11230
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931683546722722e-06,
+      "loss": 1.1782,
+      "step": 11231
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931587999235621e-06,
+      "loss": 0.7311,
+      "step": 11232
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931492451748519e-06,
+      "loss": 0.7052,
+      "step": 11233
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931396904261418e-06,
+      "loss": 1.2796,
+      "step": 11234
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931301356774317e-06,
+      "loss": 0.8496,
+      "step": 11235
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931205809287216e-06,
+      "loss": 0.7405,
+      "step": 11236
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931110261800115e-06,
+      "loss": 0.9056,
+      "step": 11237
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931014714313014e-06,
+      "loss": 0.9361,
+      "step": 11238
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930919166825912e-06,
+      "loss": 0.7777,
+      "step": 11239
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930823619338811e-06,
+      "loss": 1.2063,
+      "step": 11240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9307280718517105e-06,
+      "loss": 1.0918,
+      "step": 11241
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9306325243646095e-06,
+      "loss": 1.2209,
+      "step": 11242
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9305369768775085e-06,
+      "loss": 0.8407,
+      "step": 11243
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9304414293904076e-06,
+      "loss": 1.219,
+      "step": 11244
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930345881903307e-06,
+      "loss": 0.8752,
+      "step": 11245
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930250334416205e-06,
+      "loss": 1.0744,
+      "step": 11246
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930154786929104e-06,
+      "loss": 1.0138,
+      "step": 11247
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930059239442003e-06,
+      "loss": 1.177,
+      "step": 11248
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929963691954902e-06,
+      "loss": 1.2835,
+      "step": 11249
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929868144467801e-06,
+      "loss": 1.4472,
+      "step": 11250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9297725969807e-06,
+      "loss": 1.4774,
+      "step": 11251
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929677049493599e-06,
+      "loss": 0.9303,
+      "step": 11252
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929581502006497e-06,
+      "loss": 0.6897,
+      "step": 11253
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929485954519396e-06,
+      "loss": 0.4971,
+      "step": 11254
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929390407032295e-06,
+      "loss": 0.8255,
+      "step": 11255
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929294859545194e-06,
+      "loss": 1.0694,
+      "step": 11256
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929199312058093e-06,
+      "loss": 1.2509,
+      "step": 11257
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929103764570992e-06,
+      "loss": 1.309,
+      "step": 11258
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9290082170838915e-06,
+      "loss": 1.5561,
+      "step": 11259
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92891266959679e-06,
+      "loss": 1.1497,
+      "step": 11260
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928817122109689e-06,
+      "loss": 1.0621,
+      "step": 11261
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928721574622588e-06,
+      "loss": 0.7781,
+      "step": 11262
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928626027135487e-06,
+      "loss": 1.186,
+      "step": 11263
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928530479648385e-06,
+      "loss": 1.1391,
+      "step": 11264
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928434932161285e-06,
+      "loss": 1.1586,
+      "step": 11265
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928339384674184e-06,
+      "loss": 0.5729,
+      "step": 11266
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928243837187082e-06,
+      "loss": 1.203,
+      "step": 11267
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928148289699981e-06,
+      "loss": 1.1665,
+      "step": 11268
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92805274221288e-06,
+      "loss": 1.2892,
+      "step": 11269
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927957194725779e-06,
+      "loss": 1.1183,
+      "step": 11270
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927861647238677e-06,
+      "loss": 0.9984,
+      "step": 11271
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927766099751577e-06,
+      "loss": 1.1235,
+      "step": 11272
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927670552264476e-06,
+      "loss": 1.0002,
+      "step": 11273
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9275750047773745e-06,
+      "loss": 1.2225,
+      "step": 11274
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9274794572902735e-06,
+      "loss": 0.8233,
+      "step": 11275
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9273839098031726e-06,
+      "loss": 1.1019,
+      "step": 11276
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927288362316072e-06,
+      "loss": 0.6914,
+      "step": 11277
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92719281482897e-06,
+      "loss": 1.2095,
+      "step": 11278
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92709726734187e-06,
+      "loss": 1.0219,
+      "step": 11279
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927001719854768e-06,
+      "loss": 1.1131,
+      "step": 11280
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926906172367667e-06,
+      "loss": 0.8763,
+      "step": 11281
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926810624880566e-06,
+      "loss": 1.0075,
+      "step": 11282
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926715077393465e-06,
+      "loss": 0.9475,
+      "step": 11283
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926619529906364e-06,
+      "loss": 1.2303,
+      "step": 11284
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926523982419262e-06,
+      "loss": 0.7136,
+      "step": 11285
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926428434932162e-06,
+      "loss": 0.9136,
+      "step": 11286
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92633288744506e-06,
+      "loss": 0.8579,
+      "step": 11287
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.926237339957959e-06,
+      "loss": 0.4869,
+      "step": 11288
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.926141792470858e-06,
+      "loss": 1.1307,
+      "step": 11289
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.926046244983757e-06,
+      "loss": 1.0238,
+      "step": 11290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925950697496656e-06,
+      "loss": 1.3029,
+      "step": 11291
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925855150009555e-06,
+      "loss": 0.981,
+      "step": 11292
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9257596025224545e-06,
+      "loss": 1.2578,
+      "step": 11293
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925664055035353e-06,
+      "loss": 1.0472,
+      "step": 11294
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925568507548252e-06,
+      "loss": 0.7859,
+      "step": 11295
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925472960061151e-06,
+      "loss": 1.1845,
+      "step": 11296
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92537741257405e-06,
+      "loss": 0.6108,
+      "step": 11297
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925281865086948e-06,
+      "loss": 1.0022,
+      "step": 11298
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925186317599847e-06,
+      "loss": 1.1427,
+      "step": 11299
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.925090770112746e-06,
+      "loss": 1.2698,
+      "step": 11300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924995222625645e-06,
+      "loss": 0.9854,
+      "step": 11301
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924899675138544e-06,
+      "loss": 1.1152,
+      "step": 11302
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924804127651443e-06,
+      "loss": 0.8735,
+      "step": 11303
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924708580164342e-06,
+      "loss": 0.5802,
+      "step": 11304
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9246130326772405e-06,
+      "loss": 1.2657,
+      "step": 11305
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9245174851901395e-06,
+      "loss": 0.7805,
+      "step": 11306
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9244219377030385e-06,
+      "loss": 1.0796,
+      "step": 11307
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9243263902159376e-06,
+      "loss": 0.9637,
+      "step": 11308
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924230842728837e-06,
+      "loss": 1.2212,
+      "step": 11309
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924135295241736e-06,
+      "loss": 1.1962,
+      "step": 11310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.924039747754635e-06,
+      "loss": 0.8758,
+      "step": 11311
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923944200267533e-06,
+      "loss": 0.876,
+      "step": 11312
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923848652780432e-06,
+      "loss": 0.6981,
+      "step": 11313
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923753105293331e-06,
+      "loss": 1.3141,
+      "step": 11314
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92365755780623e-06,
+      "loss": 1.2015,
+      "step": 11315
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923562010319129e-06,
+      "loss": 1.4198,
+      "step": 11316
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923466462832028e-06,
+      "loss": 1.7241,
+      "step": 11317
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923370915344927e-06,
+      "loss": 1.1424,
+      "step": 11318
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923275367857825e-06,
+      "loss": 0.7874,
+      "step": 11319
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923179820370724e-06,
+      "loss": 1.1438,
+      "step": 11320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.923084272883623e-06,
+      "loss": 1.1403,
+      "step": 11321
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922988725396522e-06,
+      "loss": 0.9838,
+      "step": 11322
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9228931779094215e-06,
+      "loss": 0.869,
+      "step": 11323
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9227976304223205e-06,
+      "loss": 1.201,
+      "step": 11324
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9227020829352195e-06,
+      "loss": 1.0862,
+      "step": 11325
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922606535448118e-06,
+      "loss": 1.1543,
+      "step": 11326
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922510987961017e-06,
+      "loss": 0.8184,
+      "step": 11327
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922415440473916e-06,
+      "loss": 0.9745,
+      "step": 11328
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922319892986815e-06,
+      "loss": 0.925,
+      "step": 11329
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922224345499714e-06,
+      "loss": 0.6564,
+      "step": 11330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922128798012613e-06,
+      "loss": 1.1843,
+      "step": 11331
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922033250525512e-06,
+      "loss": 1.072,
+      "step": 11332
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.92193770303841e-06,
+      "loss": 0.9121,
+      "step": 11333
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921842155551309e-06,
+      "loss": 0.7988,
+      "step": 11334
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921746608064208e-06,
+      "loss": 0.8635,
+      "step": 11335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921651060577107e-06,
+      "loss": 0.7833,
+      "step": 11336
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921555513090006e-06,
+      "loss": 1.0855,
+      "step": 11337
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921459965602905e-06,
+      "loss": 0.4633,
+      "step": 11338
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9213644181158035e-06,
+      "loss": 0.6536,
+      "step": 11339
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9212688706287026e-06,
+      "loss": 0.7463,
+      "step": 11340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.921173323141602e-06,
+      "loss": 1.2872,
+      "step": 11341
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.921077775654501e-06,
+      "loss": 1.0419,
+      "step": 11342
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920982228167399e-06,
+      "loss": 0.8289,
+      "step": 11343
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920886680680299e-06,
+      "loss": 0.915,
+      "step": 11344
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920791133193198e-06,
+      "loss": 1.0872,
+      "step": 11345
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920695585706096e-06,
+      "loss": 1.0908,
+      "step": 11346
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920600038218995e-06,
+      "loss": 1.3459,
+      "step": 11347
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920504490731894e-06,
+      "loss": 0.9756,
+      "step": 11348
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920408943244793e-06,
+      "loss": 1.0114,
+      "step": 11349
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920313395757691e-06,
+      "loss": 1.1189,
+      "step": 11350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920217848270591e-06,
+      "loss": 0.6658,
+      "step": 11351
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.92012230078349e-06,
+      "loss": 0.6035,
+      "step": 11352
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.920026753296388e-06,
+      "loss": 1.2471,
+      "step": 11353
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919931205809287e-06,
+      "loss": 1.0247,
+      "step": 11354
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9198356583221865e-06,
+      "loss": 0.9813,
+      "step": 11355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9197401108350855e-06,
+      "loss": 0.7732,
+      "step": 11356
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919644563347984e-06,
+      "loss": 0.8541,
+      "step": 11357
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9195490158608836e-06,
+      "loss": 1.0629,
+      "step": 11358
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919453468373783e-06,
+      "loss": 1.031,
+      "step": 11359
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919357920886681e-06,
+      "loss": 0.8913,
+      "step": 11360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.91926237339958e-06,
+      "loss": 1.2786,
+      "step": 11361
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919166825912479e-06,
+      "loss": 1.102,
+      "step": 11362
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.919071278425378e-06,
+      "loss": 1.3657,
+      "step": 11363
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918975730938276e-06,
+      "loss": 0.7536,
+      "step": 11364
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918880183451176e-06,
+      "loss": 0.8396,
+      "step": 11365
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918784635964075e-06,
+      "loss": 0.9343,
+      "step": 11366
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918689088476973e-06,
+      "loss": 1.1879,
+      "step": 11367
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918593540989872e-06,
+      "loss": 0.7251,
+      "step": 11368
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918497993502771e-06,
+      "loss": 1.2404,
+      "step": 11369
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.91840244601567e-06,
+      "loss": 1.2225,
+      "step": 11370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9183068985285685e-06,
+      "loss": 1.323,
+      "step": 11371
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9182113510414676e-06,
+      "loss": 1.0727,
+      "step": 11372
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9181158035543675e-06,
+      "loss": 1.1965,
+      "step": 11373
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.918020256067266e-06,
+      "loss": 0.9551,
+      "step": 11374
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917924708580165e-06,
+      "loss": 1.0945,
+      "step": 11375
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917829161093064e-06,
+      "loss": 0.7193,
+      "step": 11376
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917733613605963e-06,
+      "loss": 1.2993,
+      "step": 11377
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917638066118861e-06,
+      "loss": 0.66,
+      "step": 11378
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.91754251863176e-06,
+      "loss": 0.769,
+      "step": 11379
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917446971144659e-06,
+      "loss": 1.6225,
+      "step": 11380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917351423657558e-06,
+      "loss": 1.2433,
+      "step": 11381
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917255876170457e-06,
+      "loss": 1.1852,
+      "step": 11382
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917160328683356e-06,
+      "loss": 1.4166,
+      "step": 11383
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.917064781196255e-06,
+      "loss": 0.9015,
+      "step": 11384
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916969233709153e-06,
+      "loss": 1.1861,
+      "step": 11385
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916873686222052e-06,
+      "loss": 1.0098,
+      "step": 11386
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9167781387349515e-06,
+      "loss": 1.115,
+      "step": 11387
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9166825912478505e-06,
+      "loss": 0.8489,
+      "step": 11388
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9165870437607495e-06,
+      "loss": 0.9322,
+      "step": 11389
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9164914962736486e-06,
+      "loss": 1.0515,
+      "step": 11390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916395948786547e-06,
+      "loss": 0.7842,
+      "step": 11391
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.916300401299446e-06,
+      "loss": 1.0437,
+      "step": 11392
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916204853812345e-06,
+      "loss": 0.97,
+      "step": 11393
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916109306325244e-06,
+      "loss": 0.7493,
+      "step": 11394
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.916013758838143e-06,
+      "loss": 0.9093,
+      "step": 11395
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915918211351042e-06,
+      "loss": 0.8183,
+      "step": 11396
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915822663863941e-06,
+      "loss": 1.1315,
+      "step": 11397
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915727116376839e-06,
+      "loss": 1.3126,
+      "step": 11398
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915631568889738e-06,
+      "loss": 0.9578,
+      "step": 11399
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915536021402637e-06,
+      "loss": 0.8625,
+      "step": 11400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915440473915536e-06,
+      "loss": 1.1733,
+      "step": 11401
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915344926428435e-06,
+      "loss": 0.6162,
+      "step": 11402
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915249378941334e-06,
+      "loss": 1.0029,
+      "step": 11403
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915153831454233e-06,
+      "loss": 0.9907,
+      "step": 11404
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.915058283967132e-06,
+      "loss": 0.7142,
+      "step": 11405
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914962736480031e-06,
+      "loss": 0.8593,
+      "step": 11406
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91486718899293e-06,
+      "loss": 0.8016,
+      "step": 11407
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914771641505829e-06,
+      "loss": 1.2753,
+      "step": 11408
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914676094018728e-06,
+      "loss": 0.8616,
+      "step": 11409
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914580546531627e-06,
+      "loss": 1.0555,
+      "step": 11410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914484999044526e-06,
+      "loss": 0.9413,
+      "step": 11411
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914389451557424e-06,
+      "loss": 1.3247,
+      "step": 11412
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914293904070323e-06,
+      "loss": 0.953,
+      "step": 11413
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914198356583222e-06,
+      "loss": 0.8347,
+      "step": 11414
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.914102809096121e-06,
+      "loss": 0.7507,
+      "step": 11415
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91400726160902e-06,
+      "loss": 0.9093,
+      "step": 11416
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913911714121919e-06,
+      "loss": 1.5691,
+      "step": 11417
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913816166634818e-06,
+      "loss": 1.1786,
+      "step": 11418
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9137206191477165e-06,
+      "loss": 1.0378,
+      "step": 11419
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9136250716606155e-06,
+      "loss": 0.6377,
+      "step": 11420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9135295241735145e-06,
+      "loss": 0.9119,
+      "step": 11421
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9134339766864136e-06,
+      "loss": 1.1463,
+      "step": 11422
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913338429199313e-06,
+      "loss": 0.9965,
+      "step": 11423
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913242881712212e-06,
+      "loss": 0.8765,
+      "step": 11424
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913147334225111e-06,
+      "loss": 0.846,
+      "step": 11425
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.913051786738009e-06,
+      "loss": 0.8402,
+      "step": 11426
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912956239250908e-06,
+      "loss": 1.2948,
+      "step": 11427
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912860691763807e-06,
+      "loss": 0.8168,
+      "step": 11428
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912765144276706e-06,
+      "loss": 0.8565,
+      "step": 11429
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912669596789605e-06,
+      "loss": 0.7373,
+      "step": 11430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912574049302504e-06,
+      "loss": 0.9396,
+      "step": 11431
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912478501815402e-06,
+      "loss": 0.9897,
+      "step": 11432
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912382954328301e-06,
+      "loss": 1.1819,
+      "step": 11433
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9122874068412e-06,
+      "loss": 1.2295,
+      "step": 11434
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912191859354099e-06,
+      "loss": 0.7054,
+      "step": 11435
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912096311866998e-06,
+      "loss": 1.4908,
+      "step": 11436
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9120007643798975e-06,
+      "loss": 0.8538,
+      "step": 11437
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9119052168927965e-06,
+      "loss": 1.0424,
+      "step": 11438
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911809669405695e-06,
+      "loss": 1.1869,
+      "step": 11439
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911714121918594e-06,
+      "loss": 1.1326,
+      "step": 11440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911618574431493e-06,
+      "loss": 0.8036,
+      "step": 11441
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911523026944392e-06,
+      "loss": 0.838,
+      "step": 11442
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91142747945729e-06,
+      "loss": 1.0789,
+      "step": 11443
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911331931970189e-06,
+      "loss": 1.1397,
+      "step": 11444
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911236384483089e-06,
+      "loss": 1.4353,
+      "step": 11445
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.911140836995987e-06,
+      "loss": 0.8618,
+      "step": 11446
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.911045289508886e-06,
+      "loss": 1.0478,
+      "step": 11447
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910949742021785e-06,
+      "loss": 1.1703,
+      "step": 11448
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910854194534684e-06,
+      "loss": 1.0894,
+      "step": 11449
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910758647047582e-06,
+      "loss": 0.935,
+      "step": 11450
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9106630995604815e-06,
+      "loss": 1.4434,
+      "step": 11451
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910567552073381e-06,
+      "loss": 0.9003,
+      "step": 11452
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9104720045862795e-06,
+      "loss": 1.0533,
+      "step": 11453
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9103764570991786e-06,
+      "loss": 1.0786,
+      "step": 11454
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910280909612078e-06,
+      "loss": 0.9493,
+      "step": 11455
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910185362124977e-06,
+      "loss": 1.2794,
+      "step": 11456
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.910089814637875e-06,
+      "loss": 1.1763,
+      "step": 11457
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909994267150774e-06,
+      "loss": 0.7145,
+      "step": 11458
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909898719663674e-06,
+      "loss": 1.0489,
+      "step": 11459
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909803172176572e-06,
+      "loss": 1.4375,
+      "step": 11460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909707624689471e-06,
+      "loss": 1.1605,
+      "step": 11461
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90961207720237e-06,
+      "loss": 0.6911,
+      "step": 11462
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909516529715269e-06,
+      "loss": 0.7083,
+      "step": 11463
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909420982228167e-06,
+      "loss": 0.9425,
+      "step": 11464
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909325434741066e-06,
+      "loss": 1.0763,
+      "step": 11465
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909229887253966e-06,
+      "loss": 1.3298,
+      "step": 11466
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909134339766864e-06,
+      "loss": 1.1101,
+      "step": 11467
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.909038792279763e-06,
+      "loss": 0.9375,
+      "step": 11468
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9089432447926625e-06,
+      "loss": 0.6157,
+      "step": 11469
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9088476973055615e-06,
+      "loss": 0.9587,
+      "step": 11470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90875214981846e-06,
+      "loss": 0.8147,
+      "step": 11471
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908656602331359e-06,
+      "loss": 1.1837,
+      "step": 11472
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908561054844258e-06,
+      "loss": 1.0073,
+      "step": 11473
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908465507357157e-06,
+      "loss": 1.1124,
+      "step": 11474
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908369959870056e-06,
+      "loss": 1.2651,
+      "step": 11475
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908274412382955e-06,
+      "loss": 0.8537,
+      "step": 11476
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908178864895854e-06,
+      "loss": 1.215,
+      "step": 11477
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.908083317408752e-06,
+      "loss": 1.2081,
+      "step": 11478
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907987769921651e-06,
+      "loss": 0.9997,
+      "step": 11479
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90789222243455e-06,
+      "loss": 1.1085,
+      "step": 11480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907796674947449e-06,
+      "loss": 1.3824,
+      "step": 11481
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907701127460348e-06,
+      "loss": 0.7694,
+      "step": 11482
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907605579973247e-06,
+      "loss": 0.6787,
+      "step": 11483
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9075100324861455e-06,
+      "loss": 1.1311,
+      "step": 11484
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9074144849990445e-06,
+      "loss": 0.9509,
+      "step": 11485
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9073189375119436e-06,
+      "loss": 1.11,
+      "step": 11486
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907223390024843e-06,
+      "loss": 0.8276,
+      "step": 11487
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907127842537742e-06,
+      "loss": 1.2936,
+      "step": 11488
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907032295050641e-06,
+      "loss": 1.0163,
+      "step": 11489
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90693674756354e-06,
+      "loss": 0.8507,
+      "step": 11490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906841200076438e-06,
+      "loss": 1.498,
+      "step": 11491
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906745652589337e-06,
+      "loss": 0.5098,
+      "step": 11492
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906650105102236e-06,
+      "loss": 1.2741,
+      "step": 11493
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906554557615135e-06,
+      "loss": 0.9655,
+      "step": 11494
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906459010128034e-06,
+      "loss": 1.0168,
+      "step": 11495
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906363462640933e-06,
+      "loss": 0.4469,
+      "step": 11496
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.906267915153832e-06,
+      "loss": 1.1946,
+      "step": 11497
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90617236766673e-06,
+      "loss": 1.2341,
+      "step": 11498
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.906076820179629e-06,
+      "loss": 1.1127,
+      "step": 11499
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905981272692528e-06,
+      "loss": 1.306,
+      "step": 11500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9058857252054275e-06,
+      "loss": 0.8571,
+      "step": 11501
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9057901777183265e-06,
+      "loss": 0.7214,
+      "step": 11502
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9056946302312255e-06,
+      "loss": 0.8969,
+      "step": 11503
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9055990827441246e-06,
+      "loss": 0.6896,
+      "step": 11504
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905503535257023e-06,
+      "loss": 0.7906,
+      "step": 11505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905407987769922e-06,
+      "loss": 1.2921,
+      "step": 11506
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905312440282821e-06,
+      "loss": 1.3222,
+      "step": 11507
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90521689279572e-06,
+      "loss": 1.1995,
+      "step": 11508
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905121345308619e-06,
+      "loss": 0.8411,
+      "step": 11509
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.905025797821518e-06,
+      "loss": 0.7607,
+      "step": 11510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904930250334417e-06,
+      "loss": 1.1341,
+      "step": 11511
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904834702847315e-06,
+      "loss": 0.7798,
+      "step": 11512
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904739155360214e-06,
+      "loss": 1.2595,
+      "step": 11513
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904643607873113e-06,
+      "loss": 0.7223,
+      "step": 11514
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904548060386012e-06,
+      "loss": 1.2756,
+      "step": 11515
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9044525128989105e-06,
+      "loss": 1.2718,
+      "step": 11516
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90435696541181e-06,
+      "loss": 0.9174,
+      "step": 11517
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904261417924709e-06,
+      "loss": 0.6852,
+      "step": 11518
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904165870437608e-06,
+      "loss": 0.7643,
+      "step": 11519
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.904070322950507e-06,
+      "loss": 0.9925,
+      "step": 11520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903974775463406e-06,
+      "loss": 1.1903,
+      "step": 11521
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903879227976305e-06,
+      "loss": 0.9709,
+      "step": 11522
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903783680489203e-06,
+      "loss": 0.7898,
+      "step": 11523
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903688133002103e-06,
+      "loss": 0.8729,
+      "step": 11524
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903592585515002e-06,
+      "loss": 1.1579,
+      "step": 11525
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9034970380279e-06,
+      "loss": 0.9195,
+      "step": 11526
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903401490540799e-06,
+      "loss": 0.8621,
+      "step": 11527
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903305943053698e-06,
+      "loss": 1.2883,
+      "step": 11528
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903210395566597e-06,
+      "loss": 0.8183,
+      "step": 11529
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903114848079495e-06,
+      "loss": 1.4029,
+      "step": 11530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.903019300592395e-06,
+      "loss": 1.141,
+      "step": 11531
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902923753105293e-06,
+      "loss": 0.7658,
+      "step": 11532
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9028282056181925e-06,
+      "loss": 1.4485,
+      "step": 11533
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9027326581310915e-06,
+      "loss": 0.746,
+      "step": 11534
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9026371106439905e-06,
+      "loss": 1.4243,
+      "step": 11535
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9025415631568896e-06,
+      "loss": 0.7251,
+      "step": 11536
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902446015669788e-06,
+      "loss": 1.2345,
+      "step": 11537
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902350468182688e-06,
+      "loss": 1.0641,
+      "step": 11538
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902254920695586e-06,
+      "loss": 0.9305,
+      "step": 11539
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902159373208485e-06,
+      "loss": 1.3122,
+      "step": 11540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902063825721384e-06,
+      "loss": 1.5961,
+      "step": 11541
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901968278234283e-06,
+      "loss": 1.1749,
+      "step": 11542
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901872730747181e-06,
+      "loss": 1.0033,
+      "step": 11543
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90177718326008e-06,
+      "loss": 0.9337,
+      "step": 11544
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90168163577298e-06,
+      "loss": 1.7045,
+      "step": 11545
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901586088285878e-06,
+      "loss": 1.0448,
+      "step": 11546
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901490540798777e-06,
+      "loss": 1.1505,
+      "step": 11547
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901394993311676e-06,
+      "loss": 0.9127,
+      "step": 11548
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901299445824575e-06,
+      "loss": 1.5264,
+      "step": 11549
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9012038983374736e-06,
+      "loss": 0.6345,
+      "step": 11550
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.901108350850373e-06,
+      "loss": 0.7564,
+      "step": 11551
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.901012803363272e-06,
+      "loss": 0.8986,
+      "step": 11552
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900917255876171e-06,
+      "loss": 1.0644,
+      "step": 11553
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.90082170838907e-06,
+      "loss": 1.2119,
+      "step": 11554
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900726160901969e-06,
+      "loss": 1.1302,
+      "step": 11555
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900630613414868e-06,
+      "loss": 1.0461,
+      "step": 11556
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900535065927766e-06,
+      "loss": 1.1855,
+      "step": 11557
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900439518440665e-06,
+      "loss": 1.3097,
+      "step": 11558
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900343970953564e-06,
+      "loss": 1.2049,
+      "step": 11559
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900248423466463e-06,
+      "loss": 1.1301,
+      "step": 11560
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900152875979362e-06,
+      "loss": 1.1183,
+      "step": 11561
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.900057328492261e-06,
+      "loss": 1.0511,
+      "step": 11562
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89996178100516e-06,
+      "loss": 0.7966,
+      "step": 11563
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.899866233518058e-06,
+      "loss": 1.18,
+      "step": 11564
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8997706860309575e-06,
+      "loss": 1.2386,
+      "step": 11565
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8996751385438565e-06,
+      "loss": 1.3778,
+      "step": 11566
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8995795910567555e-06,
+      "loss": 0.6774,
+      "step": 11567
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8994840435696546e-06,
+      "loss": 1.2866,
+      "step": 11568
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.899388496082554e-06,
+      "loss": 1.2457,
+      "step": 11569
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.899292948595453e-06,
+      "loss": 0.8775,
+      "step": 11570
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.899197401108351e-06,
+      "loss": 1.2484,
+      "step": 11571
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89910185362125e-06,
+      "loss": 0.8895,
+      "step": 11572
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.899006306134149e-06,
+      "loss": 1.0533,
+      "step": 11573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898910758647048e-06,
+      "loss": 0.9318,
+      "step": 11574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898815211159947e-06,
+      "loss": 0.8609,
+      "step": 11575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898719663672846e-06,
+      "loss": 0.9194,
+      "step": 11576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898624116185745e-06,
+      "loss": 0.7583,
+      "step": 11577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898528568698643e-06,
+      "loss": 0.7586,
+      "step": 11578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898433021211542e-06,
+      "loss": 1.5317,
+      "step": 11579
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898337473724441e-06,
+      "loss": 0.6351,
+      "step": 11580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89824192623734e-06,
+      "loss": 1.5393,
+      "step": 11581
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.898146378750239e-06,
+      "loss": 1.0222,
+      "step": 11582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8980508312631385e-06,
+      "loss": 0.8279,
+      "step": 11583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897955283776037e-06,
+      "loss": 0.9917,
+      "step": 11584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897859736288936e-06,
+      "loss": 0.7302,
+      "step": 11585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897764188801835e-06,
+      "loss": 1.0724,
+      "step": 11586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897668641314734e-06,
+      "loss": 1.0174,
+      "step": 11587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897573093827633e-06,
+      "loss": 1.3207,
+      "step": 11588
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897477546340532e-06,
+      "loss": 0.9497,
+      "step": 11589
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897381998853431e-06,
+      "loss": 1.0691,
+      "step": 11590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897286451366329e-06,
+      "loss": 0.7161,
+      "step": 11591
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897190903879228e-06,
+      "loss": 0.6239,
+      "step": 11592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897095356392127e-06,
+      "loss": 0.8177,
+      "step": 11593
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896999808905026e-06,
+      "loss": 1.277,
+      "step": 11594
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896904261417924e-06,
+      "loss": 1.0141,
+      "step": 11595
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896808713930824e-06,
+      "loss": 0.7361,
+      "step": 11596
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896713166443723e-06,
+      "loss": 1.3316,
+      "step": 11597
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8966176189566215e-06,
+      "loss": 1.3615,
+      "step": 11598
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8965220714695205e-06,
+      "loss": 0.7049,
+      "step": 11599
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8964265239824196e-06,
+      "loss": 1.1525,
+      "step": 11600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896330976495319e-06,
+      "loss": 0.9689,
+      "step": 11601
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896235429008217e-06,
+      "loss": 0.7962,
+      "step": 11602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.896139881521117e-06,
+      "loss": 1.0145,
+      "step": 11603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.896044334034016e-06,
+      "loss": 0.87,
+      "step": 11604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895948786546914e-06,
+      "loss": 0.7911,
+      "step": 11605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895853239059813e-06,
+      "loss": 0.8077,
+      "step": 11606
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895757691572712e-06,
+      "loss": 1.4062,
+      "step": 11607
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895662144085611e-06,
+      "loss": 1.3458,
+      "step": 11608
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895566596598509e-06,
+      "loss": 1.7227,
+      "step": 11609
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895471049111409e-06,
+      "loss": 0.6693,
+      "step": 11610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895375501624308e-06,
+      "loss": 1.1276,
+      "step": 11611
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895279954137206e-06,
+      "loss": 0.7808,
+      "step": 11612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895184406650105e-06,
+      "loss": 0.9557,
+      "step": 11613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.895088859163004e-06,
+      "loss": 0.8929,
+      "step": 11614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8949933116759035e-06,
+      "loss": 0.751,
+      "step": 11615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894897764188802e-06,
+      "loss": 0.9153,
+      "step": 11616
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8948022167017015e-06,
+      "loss": 0.7766,
+      "step": 11617
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8947066692146006e-06,
+      "loss": 0.8062,
+      "step": 11618
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894611121727499e-06,
+      "loss": 0.8952,
+      "step": 11619
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894515574240398e-06,
+      "loss": 0.9578,
+      "step": 11620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894420026753297e-06,
+      "loss": 1.4863,
+      "step": 11621
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894324479266196e-06,
+      "loss": 0.9981,
+      "step": 11622
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894228931779094e-06,
+      "loss": 0.7428,
+      "step": 11623
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894133384291993e-06,
+      "loss": 1.292,
+      "step": 11624
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.894037836804892e-06,
+      "loss": 0.9482,
+      "step": 11625
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893942289317791e-06,
+      "loss": 0.7066,
+      "step": 11626
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.89384674183069e-06,
+      "loss": 0.7575,
+      "step": 11627
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893751194343589e-06,
+      "loss": 1.4094,
+      "step": 11628
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893655646856488e-06,
+      "loss": 1.1521,
+      "step": 11629
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8935600993693865e-06,
+      "loss": 0.8642,
+      "step": 11630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8934645518822855e-06,
+      "loss": 1.3836,
+      "step": 11631
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8933690043951846e-06,
+      "loss": 0.6843,
+      "step": 11632
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893273456908084e-06,
+      "loss": 0.8192,
+      "step": 11633
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893177909420983e-06,
+      "loss": 1.3175,
+      "step": 11634
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.893082361933882e-06,
+      "loss": 0.772,
+      "step": 11635
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.89298681444678e-06,
+      "loss": 1.1238,
+      "step": 11636
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892891266959679e-06,
+      "loss": 1.1819,
+      "step": 11637
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892795719472578e-06,
+      "loss": 0.962,
+      "step": 11638
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892700171985477e-06,
+      "loss": 0.5849,
+      "step": 11639
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892604624498376e-06,
+      "loss": 1.2277,
+      "step": 11640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892509077011275e-06,
+      "loss": 1.0171,
+      "step": 11641
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892413529524174e-06,
+      "loss": 0.6499,
+      "step": 11642
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892317982037072e-06,
+      "loss": 1.3976,
+      "step": 11643
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892222434549971e-06,
+      "loss": 0.9183,
+      "step": 11644
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.89212688706287e-06,
+      "loss": 0.939,
+      "step": 11645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.892031339575769e-06,
+      "loss": 1.2164,
+      "step": 11646
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8919357920886685e-06,
+      "loss": 0.7578,
+      "step": 11647
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8918402446015675e-06,
+      "loss": 0.5283,
+      "step": 11648
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8917446971144665e-06,
+      "loss": 1.0421,
+      "step": 11649
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891649149627365e-06,
+      "loss": 1.0193,
+      "step": 11650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891553602140264e-06,
+      "loss": 0.9054,
+      "step": 11651
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891458054653163e-06,
+      "loss": 1.1673,
+      "step": 11652
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891362507166062e-06,
+      "loss": 1.2717,
+      "step": 11653
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891266959678961e-06,
+      "loss": 1.2023,
+      "step": 11654
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.89117141219186e-06,
+      "loss": 0.8911,
+      "step": 11655
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.891075864704759e-06,
+      "loss": 0.7693,
+      "step": 11656
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890980317217657e-06,
+      "loss": 0.832,
+      "step": 11657
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890884769730556e-06,
+      "loss": 0.8594,
+      "step": 11658
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890789222243455e-06,
+      "loss": 1.0671,
+      "step": 11659
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890693674756354e-06,
+      "loss": 0.5578,
+      "step": 11660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890598127269253e-06,
+      "loss": 0.9729,
+      "step": 11661
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890502579782152e-06,
+      "loss": 1.4853,
+      "step": 11662
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890407032295051e-06,
+      "loss": 1.1071,
+      "step": 11663
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8903114848079496e-06,
+      "loss": 1.0321,
+      "step": 11664
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890215937320849e-06,
+      "loss": 0.8751,
+      "step": 11665
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890120389833748e-06,
+      "loss": 1.0227,
+      "step": 11666
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.890024842346647e-06,
+      "loss": 1.0944,
+      "step": 11667
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889929294859546e-06,
+      "loss": 0.8902,
+      "step": 11668
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889833747372445e-06,
+      "loss": 1.1142,
+      "step": 11669
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889738199885344e-06,
+      "loss": 0.8319,
+      "step": 11670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889642652398242e-06,
+      "loss": 1.2545,
+      "step": 11671
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889547104911141e-06,
+      "loss": 0.8266,
+      "step": 11672
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88945155742404e-06,
+      "loss": 1.0764,
+      "step": 11673
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889356009936939e-06,
+      "loss": 0.9824,
+      "step": 11674
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889260462449838e-06,
+      "loss": 1.0376,
+      "step": 11675
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889164914962737e-06,
+      "loss": 0.9404,
+      "step": 11676
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.889069367475635e-06,
+      "loss": 0.8647,
+      "step": 11677
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888973819988534e-06,
+      "loss": 0.8894,
+      "step": 11678
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8888782725014335e-06,
+      "loss": 1.0412,
+      "step": 11679
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8887827250143325e-06,
+      "loss": 1.4701,
+      "step": 11680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8886871775272315e-06,
+      "loss": 1.085,
+      "step": 11681
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8885916300401306e-06,
+      "loss": 1.1376,
+      "step": 11682
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88849608255303e-06,
+      "loss": 0.8817,
+      "step": 11683
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888400535065928e-06,
+      "loss": 0.9763,
+      "step": 11684
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888304987578827e-06,
+      "loss": 1.0746,
+      "step": 11685
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888209440091726e-06,
+      "loss": 0.9049,
+      "step": 11686
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888113892604625e-06,
+      "loss": 0.7512,
+      "step": 11687
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.888018345117523e-06,
+      "loss": 1.4213,
+      "step": 11688
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887922797630423e-06,
+      "loss": 0.7906,
+      "step": 11689
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887827250143322e-06,
+      "loss": 1.6895,
+      "step": 11690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88773170265622e-06,
+      "loss": 1.3979,
+      "step": 11691
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887636155169119e-06,
+      "loss": 1.2562,
+      "step": 11692
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887540607682018e-06,
+      "loss": 0.9486,
+      "step": 11693
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887445060194917e-06,
+      "loss": 0.8155,
+      "step": 11694
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8873495127078155e-06,
+      "loss": 1.1534,
+      "step": 11695
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8872539652207146e-06,
+      "loss": 0.8989,
+      "step": 11696
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8871584177336145e-06,
+      "loss": 1.2197,
+      "step": 11697
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887062870246513e-06,
+      "loss": 1.4581,
+      "step": 11698
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886967322759412e-06,
+      "loss": 1.032,
+      "step": 11699
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886871775272311e-06,
+      "loss": 0.737,
+      "step": 11700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88677622778521e-06,
+      "loss": 1.1254,
+      "step": 11701
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886680680298108e-06,
+      "loss": 0.7923,
+      "step": 11702
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886585132811007e-06,
+      "loss": 0.7661,
+      "step": 11703
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886489585323907e-06,
+      "loss": 1.0728,
+      "step": 11704
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886394037836805e-06,
+      "loss": 1.2594,
+      "step": 11705
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886298490349704e-06,
+      "loss": 0.7014,
+      "step": 11706
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.886202942862603e-06,
+      "loss": 0.8352,
+      "step": 11707
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.886107395375502e-06,
+      "loss": 1.2832,
+      "step": 11708
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8860118478884e-06,
+      "loss": 1.0957,
+      "step": 11709
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885916300401299e-06,
+      "loss": 1.1597,
+      "step": 11710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885820752914199e-06,
+      "loss": 1.1172,
+      "step": 11711
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8857252054270975e-06,
+      "loss": 1.1317,
+      "step": 11712
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8856296579399965e-06,
+      "loss": 1.4647,
+      "step": 11713
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8855341104528956e-06,
+      "loss": 1.0668,
+      "step": 11714
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885438562965795e-06,
+      "loss": 1.0527,
+      "step": 11715
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885343015478693e-06,
+      "loss": 0.7322,
+      "step": 11716
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885247467991592e-06,
+      "loss": 1.5051,
+      "step": 11717
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.885151920504492e-06,
+      "loss": 1.305,
+      "step": 11718
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88505637301739e-06,
+      "loss": 1.1848,
+      "step": 11719
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884960825530289e-06,
+      "loss": 0.791,
+      "step": 11720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884865278043188e-06,
+      "loss": 1.0427,
+      "step": 11721
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884769730556087e-06,
+      "loss": 0.8105,
+      "step": 11722
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884674183068985e-06,
+      "loss": 0.8635,
+      "step": 11723
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884578635581884e-06,
+      "loss": 1.072,
+      "step": 11724
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884483088094783e-06,
+      "loss": 0.9371,
+      "step": 11725
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884387540607682e-06,
+      "loss": 1.2197,
+      "step": 11726
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884291993120581e-06,
+      "loss": 0.9942,
+      "step": 11727
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88419644563348e-06,
+      "loss": 1.1487,
+      "step": 11728
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8841008981463795e-06,
+      "loss": 1.0787,
+      "step": 11729
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.884005350659278e-06,
+      "loss": 1.1843,
+      "step": 11730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883909803172177e-06,
+      "loss": 1.0043,
+      "step": 11731
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883814255685076e-06,
+      "loss": 1.2556,
+      "step": 11732
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883718708197975e-06,
+      "loss": 0.8453,
+      "step": 11733
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883623160710874e-06,
+      "loss": 1.4604,
+      "step": 11734
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883527613223773e-06,
+      "loss": 0.7876,
+      "step": 11735
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883432065736671e-06,
+      "loss": 1.2588,
+      "step": 11736
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88333651824957e-06,
+      "loss": 0.8512,
+      "step": 11737
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883240970762469e-06,
+      "loss": 1.0088,
+      "step": 11738
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883145423275368e-06,
+      "loss": 1.0402,
+      "step": 11739
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.883049875788267e-06,
+      "loss": 0.6254,
+      "step": 11740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882954328301166e-06,
+      "loss": 0.8328,
+      "step": 11741
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882858780814065e-06,
+      "loss": 1.4357,
+      "step": 11742
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8827632333269634e-06,
+      "loss": 1.029,
+      "step": 11743
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8826676858398625e-06,
+      "loss": 0.6197,
+      "step": 11744
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8825721383527615e-06,
+      "loss": 0.8723,
+      "step": 11745
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8824765908656606e-06,
+      "loss": 1.1195,
+      "step": 11746
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88238104337856e-06,
+      "loss": 0.9592,
+      "step": 11747
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882285495891459e-06,
+      "loss": 0.8108,
+      "step": 11748
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882189948404358e-06,
+      "loss": 0.9026,
+      "step": 11749
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882094400917256e-06,
+      "loss": 0.9412,
+      "step": 11750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881998853430155e-06,
+      "loss": 1.0843,
+      "step": 11751
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881903305943054e-06,
+      "loss": 0.9238,
+      "step": 11752
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881807758455953e-06,
+      "loss": 0.8498,
+      "step": 11753
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881712210968852e-06,
+      "loss": 0.9004,
+      "step": 11754
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881616663481751e-06,
+      "loss": 0.9575,
+      "step": 11755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.88152111599465e-06,
+      "loss": 0.8666,
+      "step": 11756
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881425568507548e-06,
+      "loss": 1.1834,
+      "step": 11757
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881330021020447e-06,
+      "loss": 0.7941,
+      "step": 11758
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881234473533346e-06,
+      "loss": 1.0243,
+      "step": 11759
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.881138926046245e-06,
+      "loss": 1.3848,
+      "step": 11760
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8810433785591444e-06,
+      "loss": 1.4271,
+      "step": 11761
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8809478310720435e-06,
+      "loss": 1.0109,
+      "step": 11762
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8808522835849425e-06,
+      "loss": 1.1335,
+      "step": 11763
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880756736097841e-06,
+      "loss": 0.832,
+      "step": 11764
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.88066118861074e-06,
+      "loss": 1.0375,
+      "step": 11765
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880565641123639e-06,
+      "loss": 1.8685,
+      "step": 11766
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880470093636538e-06,
+      "loss": 1.0506,
+      "step": 11767
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880374546149436e-06,
+      "loss": 0.907,
+      "step": 11768
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880278998662336e-06,
+      "loss": 1.0664,
+      "step": 11769
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880183451175235e-06,
+      "loss": 1.389,
+      "step": 11770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.880087903688133e-06,
+      "loss": 1.1264,
+      "step": 11771
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879992356201032e-06,
+      "loss": 0.7215,
+      "step": 11772
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879896808713931e-06,
+      "loss": 1.2601,
+      "step": 11773
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.87980126122683e-06,
+      "loss": 1.0851,
+      "step": 11774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8797057137397284e-06,
+      "loss": 0.7351,
+      "step": 11775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879610166252628e-06,
+      "loss": 0.6346,
+      "step": 11776
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8795146187655265e-06,
+      "loss": 0.735,
+      "step": 11777
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8794190712784256e-06,
+      "loss": 1.1,
+      "step": 11778
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879323523791325e-06,
+      "loss": 0.8153,
+      "step": 11779
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879227976304224e-06,
+      "loss": 0.8886,
+      "step": 11780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879132428817123e-06,
+      "loss": 1.2826,
+      "step": 11781
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.879036881330021e-06,
+      "loss": 1.1756,
+      "step": 11782
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878941333842921e-06,
+      "loss": 0.814,
+      "step": 11783
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878845786355819e-06,
+      "loss": 1.6153,
+      "step": 11784
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878750238868718e-06,
+      "loss": 0.863,
+      "step": 11785
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878654691381617e-06,
+      "loss": 0.665,
+      "step": 11786
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878559143894516e-06,
+      "loss": 1.4304,
+      "step": 11787
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878463596407414e-06,
+      "loss": 1.0967,
+      "step": 11788
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878368048920313e-06,
+      "loss": 0.9737,
+      "step": 11789
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878272501433213e-06,
+      "loss": 1.4564,
+      "step": 11790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.878176953946111e-06,
+      "loss": 1.1181,
+      "step": 11791
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.87808140645901e-06,
+      "loss": 1.0203,
+      "step": 11792
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8779858589719094e-06,
+      "loss": 1.3832,
+      "step": 11793
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8778903114848085e-06,
+      "loss": 0.8756,
+      "step": 11794
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877794763997707e-06,
+      "loss": 1.0317,
+      "step": 11795
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877699216510606e-06,
+      "loss": 0.934,
+      "step": 11796
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877603669023506e-06,
+      "loss": 1.0258,
+      "step": 11797
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877508121536404e-06,
+      "loss": 0.9756,
+      "step": 11798
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877412574049303e-06,
+      "loss": 0.9524,
+      "step": 11799
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877317026562202e-06,
+      "loss": 0.7924,
+      "step": 11800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877221479075101e-06,
+      "loss": 1.0032,
+      "step": 11801
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877125931587999e-06,
+      "loss": 0.7337,
+      "step": 11802
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877030384100898e-06,
+      "loss": 0.7082,
+      "step": 11803
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876934836613797e-06,
+      "loss": 0.8048,
+      "step": 11804
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876839289126696e-06,
+      "loss": 0.6249,
+      "step": 11805
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876743741639595e-06,
+      "loss": 1.4836,
+      "step": 11806
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876648194152494e-06,
+      "loss": 1.1436,
+      "step": 11807
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.876552646665393e-06,
+      "loss": 1.0591,
+      "step": 11808
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8764570991782915e-06,
+      "loss": 0.8739,
+      "step": 11809
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8763615516911906e-06,
+      "loss": 1.3418,
+      "step": 11810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.87626600420409e-06,
+      "loss": 1.1156,
+      "step": 11811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.876170456716989e-06,
+      "loss": 0.5974,
+      "step": 11812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.876074909229888e-06,
+      "loss": 1.3307,
+      "step": 11813
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875979361742787e-06,
+      "loss": 1.1554,
+      "step": 11814
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875883814255686e-06,
+      "loss": 0.7839,
+      "step": 11815
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875788266768584e-06,
+      "loss": 1.3675,
+      "step": 11816
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875692719281483e-06,
+      "loss": 0.9043,
+      "step": 11817
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875597171794382e-06,
+      "loss": 0.9763,
+      "step": 11818
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875501624307281e-06,
+      "loss": 0.6965,
+      "step": 11819
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87540607682018e-06,
+      "loss": 0.9674,
+      "step": 11820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875310529333079e-06,
+      "loss": 1.4272,
+      "step": 11821
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875214981845978e-06,
+      "loss": 0.896,
+      "step": 11822
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875119434358876e-06,
+      "loss": 1.0765,
+      "step": 11823
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.875023886871775e-06,
+      "loss": 1.2291,
+      "step": 11824
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8749283393846744e-06,
+      "loss": 1.035,
+      "step": 11825
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8748327918975735e-06,
+      "loss": 0.9928,
+      "step": 11826
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8747372444104725e-06,
+      "loss": 1.2867,
+      "step": 11827
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8746416969233716e-06,
+      "loss": 1.2332,
+      "step": 11828
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87454614943627e-06,
+      "loss": 1.0747,
+      "step": 11829
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.874450601949169e-06,
+      "loss": 1.3005,
+      "step": 11830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.874355054462068e-06,
+      "loss": 1.5256,
+      "step": 11831
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.874259506974967e-06,
+      "loss": 1.1957,
+      "step": 11832
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.874163959487866e-06,
+      "loss": 1.0891,
+      "step": 11833
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.874068412000765e-06,
+      "loss": 1.2582,
+      "step": 11834
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873972864513664e-06,
+      "loss": 1.0386,
+      "step": 11835
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873877317026562e-06,
+      "loss": 0.6382,
+      "step": 11836
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873781769539461e-06,
+      "loss": 0.7873,
+      "step": 11837
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87368622205236e-06,
+      "loss": 1.3597,
+      "step": 11838
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873590674565259e-06,
+      "loss": 1.0973,
+      "step": 11839
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8734951270781575e-06,
+      "loss": 0.9133,
+      "step": 11840
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873399579591057e-06,
+      "loss": 1.1417,
+      "step": 11841
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873304032103956e-06,
+      "loss": 0.6525,
+      "step": 11842
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873208484616855e-06,
+      "loss": 0.9904,
+      "step": 11843
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873112937129754e-06,
+      "loss": 0.7523,
+      "step": 11844
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.873017389642653e-06,
+      "loss": 1.3012,
+      "step": 11845
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872921842155552e-06,
+      "loss": 0.856,
+      "step": 11846
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87282629466845e-06,
+      "loss": 1.3632,
+      "step": 11847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.87273074718135e-06,
+      "loss": 0.7391,
+      "step": 11848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872635199694249e-06,
+      "loss": 1.1728,
+      "step": 11849
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872539652207147e-06,
+      "loss": 0.5918,
+      "step": 11850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872444104720046e-06,
+      "loss": 1.0224,
+      "step": 11851
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872348557232945e-06,
+      "loss": 0.7082,
+      "step": 11852
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872253009745844e-06,
+      "loss": 1.1665,
+      "step": 11853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872157462258742e-06,
+      "loss": 0.9725,
+      "step": 11854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.872061914771642e-06,
+      "loss": 1.0844,
+      "step": 11855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871966367284541e-06,
+      "loss": 0.9251,
+      "step": 11856
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8718708197974394e-06,
+      "loss": 1.0017,
+      "step": 11857
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8717752723103385e-06,
+      "loss": 1.3419,
+      "step": 11858
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8716797248232375e-06,
+      "loss": 0.7807,
+      "step": 11859
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8715841773361366e-06,
+      "loss": 0.9392,
+      "step": 11860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871488629849035e-06,
+      "loss": 1.1305,
+      "step": 11861
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871393082361935e-06,
+      "loss": 0.9532,
+      "step": 11862
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871297534874834e-06,
+      "loss": 1.1698,
+      "step": 11863
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871201987387732e-06,
+      "loss": 0.918,
+      "step": 11864
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.871106439900631e-06,
+      "loss": 0.7244,
+      "step": 11865
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.87101089241353e-06,
+      "loss": 0.9725,
+      "step": 11866
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870915344926429e-06,
+      "loss": 0.8267,
+      "step": 11867
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870819797439327e-06,
+      "loss": 1.0739,
+      "step": 11868
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870724249952227e-06,
+      "loss": 0.7437,
+      "step": 11869
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870628702465125e-06,
+      "loss": 1.0434,
+      "step": 11870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870533154978024e-06,
+      "loss": 1.3361,
+      "step": 11871
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870437607490923e-06,
+      "loss": 1.0626,
+      "step": 11872
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870342060003822e-06,
+      "loss": 0.9846,
+      "step": 11873
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870246512516721e-06,
+      "loss": 0.8954,
+      "step": 11874
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.87015096502962e-06,
+      "loss": 1.0201,
+      "step": 11875
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.870055417542519e-06,
+      "loss": 1.1054,
+      "step": 11876
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869959870055418e-06,
+      "loss": 1.1711,
+      "step": 11877
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869864322568317e-06,
+      "loss": 0.5745,
+      "step": 11878
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869768775081216e-06,
+      "loss": 1.4135,
+      "step": 11879
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869673227594115e-06,
+      "loss": 0.982,
+      "step": 11880
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869577680107013e-06,
+      "loss": 1.0797,
+      "step": 11881
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869482132619912e-06,
+      "loss": 1.3005,
+      "step": 11882
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869386585132811e-06,
+      "loss": 1.0189,
+      "step": 11883
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.86929103764571e-06,
+      "loss": 1.2531,
+      "step": 11884
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869195490158609e-06,
+      "loss": 0.4619,
+      "step": 11885
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869099942671508e-06,
+      "loss": 1.2571,
+      "step": 11886
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.869004395184407e-06,
+      "loss": 0.9813,
+      "step": 11887
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.868908847697305e-06,
+      "loss": 0.9927,
+      "step": 11888
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8688133002102044e-06,
+      "loss": 1.2626,
+      "step": 11889
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8687177527231035e-06,
+      "loss": 1.195,
+      "step": 11890
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8686222052360025e-06,
+      "loss": 0.9284,
+      "step": 11891
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8685266577489016e-06,
+      "loss": 1.0079,
+      "step": 11892
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.868431110261801e-06,
+      "loss": 0.8869,
+      "step": 11893
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8683355627747e-06,
+      "loss": 1.0096,
+      "step": 11894
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.868240015287598e-06,
+      "loss": 1.2514,
+      "step": 11895
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.868144467800497e-06,
+      "loss": 0.9799,
+      "step": 11896
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.868048920313396e-06,
+      "loss": 0.9124,
+      "step": 11897
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867953372826295e-06,
+      "loss": 1.0111,
+      "step": 11898
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867857825339194e-06,
+      "loss": 0.9976,
+      "step": 11899
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867762277852093e-06,
+      "loss": 0.8786,
+      "step": 11900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867666730364992e-06,
+      "loss": 0.6923,
+      "step": 11901
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.86757118287789e-06,
+      "loss": 1.0252,
+      "step": 11902
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867475635390789e-06,
+      "loss": 1.0159,
+      "step": 11903
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867380087903688e-06,
+      "loss": 1.0364,
+      "step": 11904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867284540416587e-06,
+      "loss": 1.0564,
+      "step": 11905
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.867188992929486e-06,
+      "loss": 1.6758,
+      "step": 11906
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8670934454423854e-06,
+      "loss": 0.7093,
+      "step": 11907
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8669978979552845e-06,
+      "loss": 1.45,
+      "step": 11908
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866902350468183e-06,
+      "loss": 1.2285,
+      "step": 11909
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866806802981082e-06,
+      "loss": 1.0952,
+      "step": 11910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866711255493981e-06,
+      "loss": 1.1301,
+      "step": 11911
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.86661570800688e-06,
+      "loss": 1.2837,
+      "step": 11912
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866520160519779e-06,
+      "loss": 1.1852,
+      "step": 11913
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866424613032678e-06,
+      "loss": 1.2915,
+      "step": 11914
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866329065545577e-06,
+      "loss": 1.1003,
+      "step": 11915
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866233518058475e-06,
+      "loss": 1.1989,
+      "step": 11916
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.866137970571374e-06,
+      "loss": 1.3103,
+      "step": 11917
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.866042423084273e-06,
+      "loss": 0.766,
+      "step": 11918
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865946875597172e-06,
+      "loss": 0.7762,
+      "step": 11919
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865851328110071e-06,
+      "loss": 1.1729,
+      "step": 11920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86575578062297e-06,
+      "loss": 1.0353,
+      "step": 11921
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865660233135869e-06,
+      "loss": 0.8244,
+      "step": 11922
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8655646856487675e-06,
+      "loss": 1.2132,
+      "step": 11923
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8654691381616666e-06,
+      "loss": 1.5396,
+      "step": 11924
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865373590674566e-06,
+      "loss": 0.7933,
+      "step": 11925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865278043187465e-06,
+      "loss": 1.1869,
+      "step": 11926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865182495700364e-06,
+      "loss": 0.7315,
+      "step": 11927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.865086948213263e-06,
+      "loss": 1.1643,
+      "step": 11928
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864991400726161e-06,
+      "loss": 1.2328,
+      "step": 11929
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86489585323906e-06,
+      "loss": 1.5158,
+      "step": 11930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864800305751959e-06,
+      "loss": 0.7939,
+      "step": 11931
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864704758264858e-06,
+      "loss": 0.8992,
+      "step": 11932
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864609210777756e-06,
+      "loss": 0.9354,
+      "step": 11933
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864513663290656e-06,
+      "loss": 0.9096,
+      "step": 11934
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864418115803555e-06,
+      "loss": 1.0437,
+      "step": 11935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864322568316453e-06,
+      "loss": 0.3891,
+      "step": 11936
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864227020829352e-06,
+      "loss": 0.7737,
+      "step": 11937
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.864131473342251e-06,
+      "loss": 0.7433,
+      "step": 11938
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8640359258551504e-06,
+      "loss": 1.0485,
+      "step": 11939
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863940378368049e-06,
+      "loss": 1.1084,
+      "step": 11940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8638448308809485e-06,
+      "loss": 1.1031,
+      "step": 11941
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8637492833938476e-06,
+      "loss": 0.8131,
+      "step": 11942
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863653735906746e-06,
+      "loss": 1.1166,
+      "step": 11943
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863558188419645e-06,
+      "loss": 0.8165,
+      "step": 11944
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863462640932544e-06,
+      "loss": 1.0811,
+      "step": 11945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863367093445443e-06,
+      "loss": 0.8401,
+      "step": 11946
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.863271545958341e-06,
+      "loss": 0.776,
+      "step": 11947
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86317599847124e-06,
+      "loss": 1.0597,
+      "step": 11948
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86308045098414e-06,
+      "loss": 1.3242,
+      "step": 11949
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862984903497038e-06,
+      "loss": 0.6724,
+      "step": 11950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862889356009937e-06,
+      "loss": 1.6628,
+      "step": 11951
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862793808522836e-06,
+      "loss": 1.2357,
+      "step": 11952
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862698261035735e-06,
+      "loss": 0.9067,
+      "step": 11953
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8626027135486335e-06,
+      "loss": 0.632,
+      "step": 11954
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8625071660615325e-06,
+      "loss": 1.0187,
+      "step": 11955
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862411618574432e-06,
+      "loss": 1.0987,
+      "step": 11956
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862316071087331e-06,
+      "loss": 0.8789,
+      "step": 11957
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86222052360023e-06,
+      "loss": 1.3491,
+      "step": 11958
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862124976113129e-06,
+      "loss": 1.3825,
+      "step": 11959
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862029428626028e-06,
+      "loss": 1.0143,
+      "step": 11960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861933881138926e-06,
+      "loss": 0.994,
+      "step": 11961
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861838333651825e-06,
+      "loss": 0.9179,
+      "step": 11962
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861742786164725e-06,
+      "loss": 0.7205,
+      "step": 11963
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861647238677623e-06,
+      "loss": 1.1323,
+      "step": 11964
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861551691190522e-06,
+      "loss": 1.0052,
+      "step": 11965
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861456143703421e-06,
+      "loss": 0.7915,
+      "step": 11966
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.86136059621632e-06,
+      "loss": 0.9426,
+      "step": 11967
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.861265048729218e-06,
+      "loss": 1.1504,
+      "step": 11968
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861169501242117e-06,
+      "loss": 0.966,
+      "step": 11969
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.861073953755016e-06,
+      "loss": 0.9079,
+      "step": 11970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8609784062679154e-06,
+      "loss": 1.4182,
+      "step": 11971
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8608828587808145e-06,
+      "loss": 1.0464,
+      "step": 11972
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8607873112937135e-06,
+      "loss": 0.9095,
+      "step": 11973
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8606917638066126e-06,
+      "loss": 0.8772,
+      "step": 11974
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860596216319511e-06,
+      "loss": 1.074,
+      "step": 11975
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.86050066883241e-06,
+      "loss": 0.9376,
+      "step": 11976
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860405121345309e-06,
+      "loss": 0.7827,
+      "step": 11977
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860309573858208e-06,
+      "loss": 0.9265,
+      "step": 11978
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860214026371107e-06,
+      "loss": 1.0762,
+      "step": 11979
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860118478884006e-06,
+      "loss": 1.4829,
+      "step": 11980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.860022931396904e-06,
+      "loss": 1.2371,
+      "step": 11981
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859927383909803e-06,
+      "loss": 1.1573,
+      "step": 11982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859831836422702e-06,
+      "loss": 0.9549,
+      "step": 11983
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859736288935601e-06,
+      "loss": 0.8579,
+      "step": 11984
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8596407414485e-06,
+      "loss": 0.9101,
+      "step": 11985
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859545193961399e-06,
+      "loss": 1.1207,
+      "step": 11986
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859449646474298e-06,
+      "loss": 1.1054,
+      "step": 11987
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8593540989871966e-06,
+      "loss": 0.7128,
+      "step": 11988
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859258551500096e-06,
+      "loss": 0.6765,
+      "step": 11989
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859163004012995e-06,
+      "loss": 1.0906,
+      "step": 11990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.859067456525894e-06,
+      "loss": 1.1687,
+      "step": 11991
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858971909038793e-06,
+      "loss": 0.7676,
+      "step": 11992
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858876361551692e-06,
+      "loss": 1.1778,
+      "step": 11993
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858780814064591e-06,
+      "loss": 0.4733,
+      "step": 11994
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858685266577489e-06,
+      "loss": 0.9951,
+      "step": 11995
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858589719090388e-06,
+      "loss": 0.7121,
+      "step": 11996
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858494171603287e-06,
+      "loss": 0.5635,
+      "step": 11997
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858398624116186e-06,
+      "loss": 0.8992,
+      "step": 11998
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858303076629085e-06,
+      "loss": 1.2884,
+      "step": 11999
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858207529141984e-06,
+      "loss": 1.0099,
+      "step": 12000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858111981654883e-06,
+      "loss": 1.4084,
+      "step": 12001
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.858016434167781e-06,
+      "loss": 1.0692,
+      "step": 12002
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8579208866806804e-06,
+      "loss": 0.8765,
+      "step": 12003
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8578253391935795e-06,
+      "loss": 1.0008,
+      "step": 12004
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8577297917064785e-06,
+      "loss": 1.0509,
+      "step": 12005
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8576342442193776e-06,
+      "loss": 1.0069,
+      "step": 12006
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857538696732277e-06,
+      "loss": 1.2499,
+      "step": 12007
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857443149245176e-06,
+      "loss": 0.9908,
+      "step": 12008
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857347601758074e-06,
+      "loss": 1.1416,
+      "step": 12009
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857252054270973e-06,
+      "loss": 1.4903,
+      "step": 12010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857156506783872e-06,
+      "loss": 1.2173,
+      "step": 12011
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857060959296771e-06,
+      "loss": 0.9853,
+      "step": 12012
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.85696541180967e-06,
+      "loss": 0.8413,
+      "step": 12013
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856869864322569e-06,
+      "loss": 1.1305,
+      "step": 12014
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856774316835468e-06,
+      "loss": 1.1779,
+      "step": 12015
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856678769348366e-06,
+      "loss": 1.0899,
+      "step": 12016
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856583221861265e-06,
+      "loss": 0.6967,
+      "step": 12017
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856487674374164e-06,
+      "loss": 0.8965,
+      "step": 12018
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.856392126887063e-06,
+      "loss": 0.965,
+      "step": 12019
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8562965793999616e-06,
+      "loss": 0.5993,
+      "step": 12020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8562010319128614e-06,
+      "loss": 0.6876,
+      "step": 12021
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85610548442576e-06,
+      "loss": 1.4265,
+      "step": 12022
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.856009936938659e-06,
+      "loss": 0.9331,
+      "step": 12023
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855914389451558e-06,
+      "loss": 0.8351,
+      "step": 12024
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855818841964457e-06,
+      "loss": 0.7766,
+      "step": 12025
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855723294477356e-06,
+      "loss": 1.0669,
+      "step": 12026
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855627746990254e-06,
+      "loss": 1.5621,
+      "step": 12027
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855532199503154e-06,
+      "loss": 0.6292,
+      "step": 12028
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855436652016052e-06,
+      "loss": 1.2015,
+      "step": 12029
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855341104528951e-06,
+      "loss": 0.6824,
+      "step": 12030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85524555704185e-06,
+      "loss": 0.8519,
+      "step": 12031
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855150009554749e-06,
+      "loss": 0.8338,
+      "step": 12032
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.855054462067647e-06,
+      "loss": 0.876,
+      "step": 12033
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854958914580546e-06,
+      "loss": 0.8892,
+      "step": 12034
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854863367093446e-06,
+      "loss": 1.1236,
+      "step": 12035
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8547678196063445e-06,
+      "loss": 1.4917,
+      "step": 12036
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8546722721192435e-06,
+      "loss": 0.9291,
+      "step": 12037
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8545767246321426e-06,
+      "loss": 1.2297,
+      "step": 12038
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854481177145042e-06,
+      "loss": 1.2454,
+      "step": 12039
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85438562965794e-06,
+      "loss": 0.8288,
+      "step": 12040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854290082170839e-06,
+      "loss": 0.85,
+      "step": 12041
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854194534683739e-06,
+      "loss": 0.9794,
+      "step": 12042
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854098987196637e-06,
+      "loss": 1.2849,
+      "step": 12043
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.854003439709536e-06,
+      "loss": 1.5512,
+      "step": 12044
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853907892222435e-06,
+      "loss": 1.1568,
+      "step": 12045
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853812344735334e-06,
+      "loss": 0.9175,
+      "step": 12046
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853716797248232e-06,
+      "loss": 0.7403,
+      "step": 12047
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853621249761131e-06,
+      "loss": 0.909,
+      "step": 12048
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853525702274031e-06,
+      "loss": 1.0475,
+      "step": 12049
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853430154786929e-06,
+      "loss": 1.0323,
+      "step": 12050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853334607299828e-06,
+      "loss": 0.8837,
+      "step": 12051
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853239059812727e-06,
+      "loss": 1.2204,
+      "step": 12052
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8531435123256264e-06,
+      "loss": 1.289,
+      "step": 12053
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.853047964838525e-06,
+      "loss": 1.0268,
+      "step": 12054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852952417351424e-06,
+      "loss": 1.0125,
+      "step": 12055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852856869864323e-06,
+      "loss": 1.2347,
+      "step": 12056
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852761322377222e-06,
+      "loss": 0.9502,
+      "step": 12057
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852665774890121e-06,
+      "loss": 1.1594,
+      "step": 12058
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85257022740302e-06,
+      "loss": 1.2586,
+      "step": 12059
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852474679915919e-06,
+      "loss": 1.2806,
+      "step": 12060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852379132428817e-06,
+      "loss": 1.0008,
+      "step": 12061
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852283584941716e-06,
+      "loss": 1.0434,
+      "step": 12062
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852188037454615e-06,
+      "loss": 1.0992,
+      "step": 12063
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852092489967514e-06,
+      "loss": 0.8905,
+      "step": 12064
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851996942480413e-06,
+      "loss": 0.942,
+      "step": 12065
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851901394993312e-06,
+      "loss": 0.8237,
+      "step": 12066
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851805847506211e-06,
+      "loss": 1.0217,
+      "step": 12067
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8517103000191095e-06,
+      "loss": 0.7303,
+      "step": 12068
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8516147525320085e-06,
+      "loss": 0.9515,
+      "step": 12069
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8515192050449076e-06,
+      "loss": 1.5749,
+      "step": 12070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851423657557807e-06,
+      "loss": 0.6364,
+      "step": 12071
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851328110070706e-06,
+      "loss": 0.8039,
+      "step": 12072
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.851232562583605e-06,
+      "loss": 1.2294,
+      "step": 12073
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.851137015096503e-06,
+      "loss": 0.833,
+      "step": 12074
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.851041467609402e-06,
+      "loss": 0.6466,
+      "step": 12075
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850945920122301e-06,
+      "loss": 0.9821,
+      "step": 12076
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8508503726352e-06,
+      "loss": 1.0922,
+      "step": 12077
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850754825148099e-06,
+      "loss": 1.0829,
+      "step": 12078
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850659277660998e-06,
+      "loss": 0.896,
+      "step": 12079
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850563730173897e-06,
+      "loss": 0.9985,
+      "step": 12080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850468182686795e-06,
+      "loss": 0.7072,
+      "step": 12081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850372635199694e-06,
+      "loss": 1.3359,
+      "step": 12082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850277087712593e-06,
+      "loss": 1.1941,
+      "step": 12083
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.850181540225492e-06,
+      "loss": 0.7312,
+      "step": 12084
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8500859927383914e-06,
+      "loss": 1.1584,
+      "step": 12085
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8499904452512905e-06,
+      "loss": 1.1226,
+      "step": 12086
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8498948977641895e-06,
+      "loss": 1.1088,
+      "step": 12087
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849799350277088e-06,
+      "loss": 1.0192,
+      "step": 12088
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849703802789987e-06,
+      "loss": 1.0707,
+      "step": 12089
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849608255302886e-06,
+      "loss": 1.1254,
+      "step": 12090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849512707815785e-06,
+      "loss": 0.7862,
+      "step": 12091
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849417160328683e-06,
+      "loss": 0.6904,
+      "step": 12092
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849321612841583e-06,
+      "loss": 0.8974,
+      "step": 12093
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849226065354482e-06,
+      "loss": 0.6334,
+      "step": 12094
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84913051786738e-06,
+      "loss": 1.0378,
+      "step": 12095
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.849034970380279e-06,
+      "loss": 0.8001,
+      "step": 12096
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848939422893178e-06,
+      "loss": 0.6864,
+      "step": 12097
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848843875406077e-06,
+      "loss": 0.9668,
+      "step": 12098
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8487483279189754e-06,
+      "loss": 1.0348,
+      "step": 12099
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848652780431875e-06,
+      "loss": 0.8511,
+      "step": 12100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848557232944774e-06,
+      "loss": 0.99,
+      "step": 12101
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8484616854576726e-06,
+      "loss": 0.7028,
+      "step": 12102
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848366137970572e-06,
+      "loss": 0.8625,
+      "step": 12103
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848270590483471e-06,
+      "loss": 0.8104,
+      "step": 12104
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84817504299637e-06,
+      "loss": 0.8195,
+      "step": 12105
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.848079495509268e-06,
+      "loss": 1.8232,
+      "step": 12106
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847983948022168e-06,
+      "loss": 0.9214,
+      "step": 12107
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847888400535067e-06,
+      "loss": 1.1216,
+      "step": 12108
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847792853047965e-06,
+      "loss": 0.949,
+      "step": 12109
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847697305560864e-06,
+      "loss": 0.7181,
+      "step": 12110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847601758073763e-06,
+      "loss": 1.3117,
+      "step": 12111
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847506210586662e-06,
+      "loss": 1.4175,
+      "step": 12112
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84741066309956e-06,
+      "loss": 1.0539,
+      "step": 12113
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84731511561246e-06,
+      "loss": 1.0946,
+      "step": 12114
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847219568125359e-06,
+      "loss": 1.6241,
+      "step": 12115
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847124020638257e-06,
+      "loss": 1.2177,
+      "step": 12116
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8470284731511564e-06,
+      "loss": 0.8373,
+      "step": 12117
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8469329256640555e-06,
+      "loss": 1.2106,
+      "step": 12118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8468373781769545e-06,
+      "loss": 0.9999,
+      "step": 12119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846741830689853e-06,
+      "loss": 1.0152,
+      "step": 12120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846646283202753e-06,
+      "loss": 0.5311,
+      "step": 12121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846550735715651e-06,
+      "loss": 0.8374,
+      "step": 12122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84645518822855e-06,
+      "loss": 1.3541,
+      "step": 12123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846359640741449e-06,
+      "loss": 1.2725,
+      "step": 12124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846264093254348e-06,
+      "loss": 1.1552,
+      "step": 12125
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.846168545767247e-06,
+      "loss": 1.0951,
+      "step": 12126
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.846072998280145e-06,
+      "loss": 1.211,
+      "step": 12127
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845977450793044e-06,
+      "loss": 1.3385,
+      "step": 12128
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845881903305943e-06,
+      "loss": 1.3872,
+      "step": 12129
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845786355818842e-06,
+      "loss": 0.9309,
+      "step": 12130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845690808331741e-06,
+      "loss": 1.4522,
+      "step": 12131
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84559526084464e-06,
+      "loss": 0.7133,
+      "step": 12132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8454997133575385e-06,
+      "loss": 0.9639,
+      "step": 12133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8454041658704376e-06,
+      "loss": 1.199,
+      "step": 12134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845308618383337e-06,
+      "loss": 1.1472,
+      "step": 12135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845213070896236e-06,
+      "loss": 1.1119,
+      "step": 12136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845117523409135e-06,
+      "loss": 0.9382,
+      "step": 12137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.845021975922034e-06,
+      "loss": 1.1054,
+      "step": 12138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844926428434933e-06,
+      "loss": 0.7685,
+      "step": 12139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844830880947831e-06,
+      "loss": 1.2649,
+      "step": 12140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84473533346073e-06,
+      "loss": 0.6177,
+      "step": 12141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844639785973629e-06,
+      "loss": 0.7262,
+      "step": 12142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844544238486528e-06,
+      "loss": 0.6472,
+      "step": 12143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844448690999427e-06,
+      "loss": 0.9621,
+      "step": 12144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844353143512326e-06,
+      "loss": 0.9497,
+      "step": 12145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844257596025225e-06,
+      "loss": 0.6041,
+      "step": 12146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844162048538123e-06,
+      "loss": 0.8296,
+      "step": 12147
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.844066501051022e-06,
+      "loss": 1.0719,
+      "step": 12148
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8439709535639214e-06,
+      "loss": 1.2229,
+      "step": 12149
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8438754060768205e-06,
+      "loss": 1.0379,
+      "step": 12150
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8437798585897195e-06,
+      "loss": 0.8783,
+      "step": 12151
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8436843111026186e-06,
+      "loss": 1.0539,
+      "step": 12152
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843588763615518e-06,
+      "loss": 0.9418,
+      "step": 12153
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843493216128416e-06,
+      "loss": 0.9805,
+      "step": 12154
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843397668641315e-06,
+      "loss": 1.114,
+      "step": 12155
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843302121154214e-06,
+      "loss": 0.9756,
+      "step": 12156
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843206573667113e-06,
+      "loss": 1.0574,
+      "step": 12157
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843111026180012e-06,
+      "loss": 0.8038,
+      "step": 12158
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.843015478692911e-06,
+      "loss": 0.8461,
+      "step": 12159
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84291993120581e-06,
+      "loss": 0.8554,
+      "step": 12160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842824383718708e-06,
+      "loss": 1.1174,
+      "step": 12161
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842728836231607e-06,
+      "loss": 1.0577,
+      "step": 12162
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842633288744506e-06,
+      "loss": 0.434,
+      "step": 12163
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842537741257405e-06,
+      "loss": 0.8982,
+      "step": 12164
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842442193770304e-06,
+      "loss": 1.1242,
+      "step": 12165
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842346646283203e-06,
+      "loss": 1.1812,
+      "step": 12166
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8422510987961024e-06,
+      "loss": 0.8885,
+      "step": 12167
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842155551309001e-06,
+      "loss": 0.7278,
+      "step": 12168
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8420600038219e-06,
+      "loss": 0.9867,
+      "step": 12169
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841964456334799e-06,
+      "loss": 0.9697,
+      "step": 12170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841868908847698e-06,
+      "loss": 1.3059,
+      "step": 12171
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841773361360597e-06,
+      "loss": 0.7531,
+      "step": 12172
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841677813873496e-06,
+      "loss": 1.2294,
+      "step": 12173
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841582266386394e-06,
+      "loss": 1.186,
+      "step": 12174
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841486718899293e-06,
+      "loss": 1.2475,
+      "step": 12175
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841391171412192e-06,
+      "loss": 1.3375,
+      "step": 12176
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.841295623925091e-06,
+      "loss": 1.4049,
+      "step": 12177
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84120007643799e-06,
+      "loss": 0.6573,
+      "step": 12178
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.841104528950889e-06,
+      "loss": 1.3747,
+      "step": 12179
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.841008981463788e-06,
+      "loss": 0.8651,
+      "step": 12180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8409134339766864e-06,
+      "loss": 0.8594,
+      "step": 12181
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8408178864895855e-06,
+      "loss": 0.883,
+      "step": 12182
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8407223390024845e-06,
+      "loss": 0.9445,
+      "step": 12183
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8406267915153836e-06,
+      "loss": 0.7287,
+      "step": 12184
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840531244028282e-06,
+      "loss": 1.4799,
+      "step": 12185
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840435696541182e-06,
+      "loss": 0.6852,
+      "step": 12186
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840340149054081e-06,
+      "loss": 0.9901,
+      "step": 12187
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840244601566979e-06,
+      "loss": 0.8694,
+      "step": 12188
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840149054079878e-06,
+      "loss": 1.0293,
+      "step": 12189
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.840053506592777e-06,
+      "loss": 1.0719,
+      "step": 12190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839957959105676e-06,
+      "loss": 0.7655,
+      "step": 12191
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839862411618574e-06,
+      "loss": 1.1488,
+      "step": 12192
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839766864131474e-06,
+      "loss": 1.0047,
+      "step": 12193
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839671316644373e-06,
+      "loss": 0.8478,
+      "step": 12194
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839575769157271e-06,
+      "loss": 1.3199,
+      "step": 12195
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.83948022167017e-06,
+      "loss": 1.4413,
+      "step": 12196
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839384674183069e-06,
+      "loss": 0.8828,
+      "step": 12197
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839289126695968e-06,
+      "loss": 1.1732,
+      "step": 12198
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839193579208867e-06,
+      "loss": 1.0318,
+      "step": 12199
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.839098031721766e-06,
+      "loss": 1.1362,
+      "step": 12200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8390024842346655e-06,
+      "loss": 0.755,
+      "step": 12201
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838906936747564e-06,
+      "loss": 0.863,
+      "step": 12202
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838811389260463e-06,
+      "loss": 0.4796,
+      "step": 12203
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838715841773362e-06,
+      "loss": 0.6323,
+      "step": 12204
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838620294286261e-06,
+      "loss": 0.8518,
+      "step": 12205
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838524746799159e-06,
+      "loss": 1.1828,
+      "step": 12206
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838429199312058e-06,
+      "loss": 0.8771,
+      "step": 12207
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838333651824958e-06,
+      "loss": 0.9027,
+      "step": 12208
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838238104337856e-06,
+      "loss": 0.9324,
+      "step": 12209
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838142556850755e-06,
+      "loss": 0.797,
+      "step": 12210
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.838047009363654e-06,
+      "loss": 1.1199,
+      "step": 12211
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837951461876553e-06,
+      "loss": 1.1454,
+      "step": 12212
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8378559143894514e-06,
+      "loss": 1.2187,
+      "step": 12213
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8377603669023505e-06,
+      "loss": 1.1672,
+      "step": 12214
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8376648194152495e-06,
+      "loss": 1.3352,
+      "step": 12215
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8375692719281486e-06,
+      "loss": 1.1234,
+      "step": 12216
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837473724441048e-06,
+      "loss": 1.343,
+      "step": 12217
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837378176953947e-06,
+      "loss": 1.0276,
+      "step": 12218
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837282629466846e-06,
+      "loss": 0.861,
+      "step": 12219
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837187081979744e-06,
+      "loss": 0.8665,
+      "step": 12220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837091534492643e-06,
+      "loss": 1.2255,
+      "step": 12221
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836995987005542e-06,
+      "loss": 1.0561,
+      "step": 12222
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836900439518441e-06,
+      "loss": 0.9179,
+      "step": 12223
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.83680489203134e-06,
+      "loss": 0.8975,
+      "step": 12224
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836709344544239e-06,
+      "loss": 0.9966,
+      "step": 12225
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836613797057137e-06,
+      "loss": 1.2939,
+      "step": 12226
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836518249570036e-06,
+      "loss": 1.0068,
+      "step": 12227
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836422702082935e-06,
+      "loss": 1.0869,
+      "step": 12228
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836327154595834e-06,
+      "loss": 1.2561,
+      "step": 12229
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.836231607108733e-06,
+      "loss": 1.139,
+      "step": 12230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8361360596216324e-06,
+      "loss": 1.4513,
+      "step": 12231
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8360405121345315e-06,
+      "loss": 1.0499,
+      "step": 12232
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83594496464743e-06,
+      "loss": 0.745,
+      "step": 12233
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835849417160329e-06,
+      "loss": 0.7872,
+      "step": 12234
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835753869673228e-06,
+      "loss": 1.1516,
+      "step": 12235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835658322186127e-06,
+      "loss": 0.7799,
+      "step": 12236
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835562774699026e-06,
+      "loss": 1.0223,
+      "step": 12237
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835467227211925e-06,
+      "loss": 0.8447,
+      "step": 12238
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835371679724824e-06,
+      "loss": 1.3349,
+      "step": 12239
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835276132237722e-06,
+      "loss": 0.6185,
+      "step": 12240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.835180584750621e-06,
+      "loss": 0.8383,
+      "step": 12241
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83508503726352e-06,
+      "loss": 1.4778,
+      "step": 12242
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834989489776419e-06,
+      "loss": 0.7628,
+      "step": 12243
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834893942289318e-06,
+      "loss": 0.9982,
+      "step": 12244
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834798394802217e-06,
+      "loss": 1.1102,
+      "step": 12245
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834702847315116e-06,
+      "loss": 0.7473,
+      "step": 12246
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8346072998280145e-06,
+      "loss": 0.8703,
+      "step": 12247
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8345117523409136e-06,
+      "loss": 0.8768,
+      "step": 12248
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834416204853813e-06,
+      "loss": 0.9524,
+      "step": 12249
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834320657366712e-06,
+      "loss": 0.7188,
+      "step": 12250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834225109879611e-06,
+      "loss": 1.0037,
+      "step": 12251
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83412956239251e-06,
+      "loss": 1.0447,
+      "step": 12252
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.834034014905409e-06,
+      "loss": 1.407,
+      "step": 12253
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833938467418307e-06,
+      "loss": 1.0343,
+      "step": 12254
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833842919931206e-06,
+      "loss": 0.905,
+      "step": 12255
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833747372444105e-06,
+      "loss": 0.8778,
+      "step": 12256
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833651824957004e-06,
+      "loss": 0.7755,
+      "step": 12257
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833556277469903e-06,
+      "loss": 0.7761,
+      "step": 12258
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833460729982802e-06,
+      "loss": 1.615,
+      "step": 12259
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833365182495701e-06,
+      "loss": 1.0726,
+      "step": 12260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833269635008599e-06,
+      "loss": 1.3352,
+      "step": 12261
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.833174087521498e-06,
+      "loss": 1.1165,
+      "step": 12262
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8330785400343974e-06,
+      "loss": 0.876,
+      "step": 12263
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8329829925472965e-06,
+      "loss": 0.8859,
+      "step": 12264
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8328874450601955e-06,
+      "loss": 1.0318,
+      "step": 12265
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8327918975730946e-06,
+      "loss": 0.7266,
+      "step": 12266
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832696350085993e-06,
+      "loss": 1.1308,
+      "step": 12267
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832600802598892e-06,
+      "loss": 0.7398,
+      "step": 12268
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832505255111791e-06,
+      "loss": 0.9355,
+      "step": 12269
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.83240970762469e-06,
+      "loss": 1.0062,
+      "step": 12270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832314160137589e-06,
+      "loss": 1.6281,
+      "step": 12271
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832218612650487e-06,
+      "loss": 1.0331,
+      "step": 12272
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832123065163387e-06,
+      "loss": 1.2326,
+      "step": 12273
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832027517676285e-06,
+      "loss": 0.8945,
+      "step": 12274
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831931970189184e-06,
+      "loss": 1.0934,
+      "step": 12275
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831836422702083e-06,
+      "loss": 0.9289,
+      "step": 12276
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831740875214982e-06,
+      "loss": 0.6889,
+      "step": 12277
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8316453277278805e-06,
+      "loss": 1.1605,
+      "step": 12278
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8315497802407795e-06,
+      "loss": 0.5488,
+      "step": 12279
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831454232753679e-06,
+      "loss": 0.7553,
+      "step": 12280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831358685266578e-06,
+      "loss": 1.2261,
+      "step": 12281
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831263137779477e-06,
+      "loss": 1.173,
+      "step": 12282
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831167590292376e-06,
+      "loss": 1.0335,
+      "step": 12283
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.831072042805275e-06,
+      "loss": 1.1456,
+      "step": 12284
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830976495318173e-06,
+      "loss": 1.3774,
+      "step": 12285
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830880947831072e-06,
+      "loss": 1.5799,
+      "step": 12286
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830785400343972e-06,
+      "loss": 1.3,
+      "step": 12287
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.83068985285687e-06,
+      "loss": 0.6701,
+      "step": 12288
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830594305369769e-06,
+      "loss": 0.643,
+      "step": 12289
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830498757882668e-06,
+      "loss": 1.0161,
+      "step": 12290
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830403210395567e-06,
+      "loss": 0.8934,
+      "step": 12291
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830307662908465e-06,
+      "loss": 0.9681,
+      "step": 12292
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830212115421364e-06,
+      "loss": 1.3565,
+      "step": 12293
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.830116567934264e-06,
+      "loss": 1.0361,
+      "step": 12294
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8300210204471624e-06,
+      "loss": 0.976,
+      "step": 12295
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8299254729600615e-06,
+      "loss": 0.6597,
+      "step": 12296
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8298299254729605e-06,
+      "loss": 1.2149,
+      "step": 12297
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8297343779858596e-06,
+      "loss": 0.8615,
+      "step": 12298
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829638830498758e-06,
+      "loss": 0.6839,
+      "step": 12299
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829543283011657e-06,
+      "loss": 0.7382,
+      "step": 12300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829447735524557e-06,
+      "loss": 1.0589,
+      "step": 12301
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829352188037455e-06,
+      "loss": 0.8445,
+      "step": 12302
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829256640550354e-06,
+      "loss": 1.0319,
+      "step": 12303
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829161093063253e-06,
+      "loss": 0.8047,
+      "step": 12304
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.829065545576152e-06,
+      "loss": 1.2411,
+      "step": 12305
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.82896999808905e-06,
+      "loss": 0.8266,
+      "step": 12306
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828874450601949e-06,
+      "loss": 0.8178,
+      "step": 12307
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828778903114848e-06,
+      "loss": 1.1514,
+      "step": 12308
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828683355627747e-06,
+      "loss": 0.6925,
+      "step": 12309
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828587808140646e-06,
+      "loss": 1.0004,
+      "step": 12310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828492260653545e-06,
+      "loss": 0.7485,
+      "step": 12311
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828396713166444e-06,
+      "loss": 1.1862,
+      "step": 12312
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828301165679343e-06,
+      "loss": 1.5081,
+      "step": 12313
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828205618192242e-06,
+      "loss": 1.0923,
+      "step": 12314
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.828110070705141e-06,
+      "loss": 0.8781,
+      "step": 12315
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.82801452321804e-06,
+      "loss": 0.7381,
+      "step": 12316
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827918975730939e-06,
+      "loss": 1.2005,
+      "step": 12317
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827823428243838e-06,
+      "loss": 0.7927,
+      "step": 12318
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827727880756737e-06,
+      "loss": 1.2492,
+      "step": 12319
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827632333269635e-06,
+      "loss": 0.7032,
+      "step": 12320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827536785782534e-06,
+      "loss": 0.7639,
+      "step": 12321
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827441238295433e-06,
+      "loss": 0.6374,
+      "step": 12322
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827345690808332e-06,
+      "loss": 1.2756,
+      "step": 12323
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827250143321231e-06,
+      "loss": 1.0232,
+      "step": 12324
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.82715459583413e-06,
+      "loss": 1.1554,
+      "step": 12325
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827059048347028e-06,
+      "loss": 0.7893,
+      "step": 12326
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8269635008599274e-06,
+      "loss": 0.8149,
+      "step": 12327
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8268679533728265e-06,
+      "loss": 0.9049,
+      "step": 12328
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8267724058857255e-06,
+      "loss": 0.6997,
+      "step": 12329
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8266768583986246e-06,
+      "loss": 1.2158,
+      "step": 12330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826581310911524e-06,
+      "loss": 0.7711,
+      "step": 12331
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826485763424423e-06,
+      "loss": 1.1663,
+      "step": 12332
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826390215937321e-06,
+      "loss": 1.2251,
+      "step": 12333
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.82629466845022e-06,
+      "loss": 1.2844,
+      "step": 12334
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826199120963119e-06,
+      "loss": 1.0507,
+      "step": 12335
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826103573476018e-06,
+      "loss": 0.8245,
+      "step": 12336
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.826008025988917e-06,
+      "loss": 1.1329,
+      "step": 12337
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825912478501816e-06,
+      "loss": 0.9914,
+      "step": 12338
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825816931014715e-06,
+      "loss": 0.7937,
+      "step": 12339
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825721383527613e-06,
+      "loss": 0.8001,
+      "step": 12340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825625836040512e-06,
+      "loss": 0.8081,
+      "step": 12341
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825530288553411e-06,
+      "loss": 1.169,
+      "step": 12342
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.82543474106631e-06,
+      "loss": 1.0504,
+      "step": 12343
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8253391935792086e-06,
+      "loss": 0.8974,
+      "step": 12344
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8252436460921084e-06,
+      "loss": 0.978,
+      "step": 12345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8251480986050075e-06,
+      "loss": 0.7556,
+      "step": 12346
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.825052551117906e-06,
+      "loss": 0.9697,
+      "step": 12347
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824957003630805e-06,
+      "loss": 0.9871,
+      "step": 12348
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824861456143704e-06,
+      "loss": 0.829,
+      "step": 12349
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824765908656603e-06,
+      "loss": 0.6879,
+      "step": 12350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824670361169501e-06,
+      "loss": 1.3378,
+      "step": 12351
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824574813682401e-06,
+      "loss": 0.8454,
+      "step": 12352
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8244792661953e-06,
+      "loss": 1.1306,
+      "step": 12353
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824383718708198e-06,
+      "loss": 0.8299,
+      "step": 12354
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824288171221097e-06,
+      "loss": 0.9225,
+      "step": 12355
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824192623733996e-06,
+      "loss": 1.4188,
+      "step": 12356
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824097076246895e-06,
+      "loss": 1.0702,
+      "step": 12357
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.824001528759793e-06,
+      "loss": 0.8696,
+      "step": 12358
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823905981272693e-06,
+      "loss": 0.8532,
+      "step": 12359
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823810433785592e-06,
+      "loss": 0.6997,
+      "step": 12360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8237148862984905e-06,
+      "loss": 0.6844,
+      "step": 12361
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8236193388113896e-06,
+      "loss": 1.6553,
+      "step": 12362
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823523791324289e-06,
+      "loss": 0.7962,
+      "step": 12363
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823428243837188e-06,
+      "loss": 1.0521,
+      "step": 12364
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823332696350086e-06,
+      "loss": 0.9862,
+      "step": 12365
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823237148862986e-06,
+      "loss": 1.0671,
+      "step": 12366
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823141601375884e-06,
+      "loss": 1.3016,
+      "step": 12367
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.823046053888783e-06,
+      "loss": 1.2355,
+      "step": 12368
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822950506401682e-06,
+      "loss": 0.6739,
+      "step": 12369
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822854958914581e-06,
+      "loss": 1.6446,
+      "step": 12370
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.82275941142748e-06,
+      "loss": 0.8079,
+      "step": 12371
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822663863940378e-06,
+      "loss": 0.6239,
+      "step": 12372
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822568316453278e-06,
+      "loss": 1.0497,
+      "step": 12373
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822472768966176e-06,
+      "loss": 1.09,
+      "step": 12374
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822377221479075e-06,
+      "loss": 1.3053,
+      "step": 12375
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822281673991974e-06,
+      "loss": 0.69,
+      "step": 12376
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8221861265048734e-06,
+      "loss": 0.8362,
+      "step": 12377
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822090579017772e-06,
+      "loss": 0.7682,
+      "step": 12378
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821995031530671e-06,
+      "loss": 0.9422,
+      "step": 12379
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.82189948404357e-06,
+      "loss": 1.4185,
+      "step": 12380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821803936556469e-06,
+      "loss": 1.1188,
+      "step": 12381
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821708389069368e-06,
+      "loss": 0.9677,
+      "step": 12382
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821612841582267e-06,
+      "loss": 0.8029,
+      "step": 12383
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821517294095166e-06,
+      "loss": 0.9751,
+      "step": 12384
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821421746608064e-06,
+      "loss": 0.991,
+      "step": 12385
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821326199120963e-06,
+      "loss": 0.9877,
+      "step": 12386
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821230651633862e-06,
+      "loss": 1.3123,
+      "step": 12387
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.821135104146761e-06,
+      "loss": 0.8386,
+      "step": 12388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.82103955665966e-06,
+      "loss": 0.9193,
+      "step": 12389
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820944009172559e-06,
+      "loss": 0.9714,
+      "step": 12390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820848461685458e-06,
+      "loss": 0.7189,
+      "step": 12391
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8207529141983565e-06,
+      "loss": 0.988,
+      "step": 12392
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8206573667112555e-06,
+      "loss": 0.8068,
+      "step": 12393
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8205618192241546e-06,
+      "loss": 0.6777,
+      "step": 12394
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820466271737054e-06,
+      "loss": 1.2454,
+      "step": 12395
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820370724249953e-06,
+      "loss": 0.6725,
+      "step": 12396
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820275176762852e-06,
+      "loss": 0.8659,
+      "step": 12397
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820179629275751e-06,
+      "loss": 0.8412,
+      "step": 12398
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.820084081788649e-06,
+      "loss": 1.0378,
+      "step": 12399
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819988534301548e-06,
+      "loss": 0.8311,
+      "step": 12400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819892986814447e-06,
+      "loss": 0.717,
+      "step": 12401
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819797439327346e-06,
+      "loss": 1.0079,
+      "step": 12402
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819701891840245e-06,
+      "loss": 0.7683,
+      "step": 12403
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819606344353144e-06,
+      "loss": 0.7968,
+      "step": 12404
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819510796866043e-06,
+      "loss": 0.9647,
+      "step": 12405
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819415249378941e-06,
+      "loss": 1.0433,
+      "step": 12406
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.81931970189184e-06,
+      "loss": 1.174,
+      "step": 12407
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.819224154404739e-06,
+      "loss": 0.7255,
+      "step": 12408
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8191286069176384e-06,
+      "loss": 1.2367,
+      "step": 12409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8190330594305375e-06,
+      "loss": 0.8722,
+      "step": 12410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8189375119434365e-06,
+      "loss": 0.5789,
+      "step": 12411
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8188419644563356e-06,
+      "loss": 0.613,
+      "step": 12412
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818746416969234e-06,
+      "loss": 0.9494,
+      "step": 12413
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818650869482133e-06,
+      "loss": 0.9934,
+      "step": 12414
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818555321995032e-06,
+      "loss": 1.0274,
+      "step": 12415
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818459774507931e-06,
+      "loss": 0.988,
+      "step": 12416
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.81836422702083e-06,
+      "loss": 1.0754,
+      "step": 12417
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818268679533729e-06,
+      "loss": 0.5209,
+      "step": 12418
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818173132046627e-06,
+      "loss": 0.7969,
+      "step": 12419
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.818077584559526e-06,
+      "loss": 0.8573,
+      "step": 12420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817982037072425e-06,
+      "loss": 1.1574,
+      "step": 12421
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817886489585324e-06,
+      "loss": 0.9278,
+      "step": 12422
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817790942098223e-06,
+      "loss": 0.8182,
+      "step": 12423
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817695394611122e-06,
+      "loss": 1.1014,
+      "step": 12424
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817599847124021e-06,
+      "loss": 1.2207,
+      "step": 12425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8175042996369196e-06,
+      "loss": 1.3186,
+      "step": 12426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817408752149819e-06,
+      "loss": 1.0769,
+      "step": 12427
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817313204662718e-06,
+      "loss": 1.0355,
+      "step": 12428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817217657175617e-06,
+      "loss": 0.9357,
+      "step": 12429
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817122109688515e-06,
+      "loss": 1.244,
+      "step": 12430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.817026562201415e-06,
+      "loss": 1.1522,
+      "step": 12431
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816931014714314e-06,
+      "loss": 1.118,
+      "step": 12432
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816835467227212e-06,
+      "loss": 0.8843,
+      "step": 12433
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816739919740111e-06,
+      "loss": 1.2283,
+      "step": 12434
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.81664437225301e-06,
+      "loss": 1.441,
+      "step": 12435
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816548824765909e-06,
+      "loss": 0.9301,
+      "step": 12436
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816453277278807e-06,
+      "loss": 1.1307,
+      "step": 12437
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816357729791707e-06,
+      "loss": 0.8375,
+      "step": 12438
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816262182304606e-06,
+      "loss": 1.0398,
+      "step": 12439
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816166634817504e-06,
+      "loss": 0.961,
+      "step": 12440
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8160710873304034e-06,
+      "loss": 0.943,
+      "step": 12441
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8159755398433025e-06,
+      "loss": 0.9591,
+      "step": 12442
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8158799923562015e-06,
+      "loss": 0.539,
+      "step": 12443
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8157844448691e-06,
+      "loss": 1.0164,
+      "step": 12444
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815688897382e-06,
+      "loss": 0.9609,
+      "step": 12445
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815593349894899e-06,
+      "loss": 1.0424,
+      "step": 12446
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815497802407797e-06,
+      "loss": 0.8944,
+      "step": 12447
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815402254920696e-06,
+      "loss": 1.3465,
+      "step": 12448
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815306707433595e-06,
+      "loss": 0.7458,
+      "step": 12449
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815211159946494e-06,
+      "loss": 1.3753,
+      "step": 12450
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815115612459392e-06,
+      "loss": 0.6398,
+      "step": 12451
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.815020064972291e-06,
+      "loss": 0.9446,
+      "step": 12452
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814924517485191e-06,
+      "loss": 1.1866,
+      "step": 12453
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814828969998089e-06,
+      "loss": 1.1879,
+      "step": 12454
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814733422510988e-06,
+      "loss": 0.8234,
+      "step": 12455
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8146378750238873e-06,
+      "loss": 1.6981,
+      "step": 12456
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814542327536786e-06,
+      "loss": 0.9953,
+      "step": 12457
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814446780049685e-06,
+      "loss": 0.8662,
+      "step": 12458
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8143512325625836e-06,
+      "loss": 0.8476,
+      "step": 12459
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814255685075483e-06,
+      "loss": 0.7104,
+      "step": 12460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.814160137588382e-06,
+      "loss": 0.8246,
+      "step": 12461
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8140645901012807e-06,
+      "loss": 0.7307,
+      "step": 12462
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8139690426141797e-06,
+      "loss": 1.2007,
+      "step": 12463
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8138734951270784e-06,
+      "loss": 0.784,
+      "step": 12464
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8137779476399774e-06,
+      "loss": 0.8439,
+      "step": 12465
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.813682400152876e-06,
+      "loss": 1.1088,
+      "step": 12466
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8135868526657755e-06,
+      "loss": 1.0084,
+      "step": 12467
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.813491305178674e-06,
+      "loss": 0.715,
+      "step": 12468
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.813395757691573e-06,
+      "loss": 1.3424,
+      "step": 12469
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8133002102044717e-06,
+      "loss": 0.9644,
+      "step": 12470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8132046627173708e-06,
+      "loss": 0.8325,
+      "step": 12471
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.81310911523027e-06,
+      "loss": 1.3093,
+      "step": 12472
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8130135677431684e-06,
+      "loss": 0.6257,
+      "step": 12473
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812918020256068e-06,
+      "loss": 0.9715,
+      "step": 12474
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8128224727689665e-06,
+      "loss": 0.7906,
+      "step": 12475
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8127269252818656e-06,
+      "loss": 0.9725,
+      "step": 12476
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812631377794764e-06,
+      "loss": 1.1092,
+      "step": 12477
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812535830307663e-06,
+      "loss": 1.0082,
+      "step": 12478
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812440282820562e-06,
+      "loss": 1.3764,
+      "step": 12479
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812344735333461e-06,
+      "loss": 0.7785,
+      "step": 12480
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8122491878463603e-06,
+      "loss": 1.2111,
+      "step": 12481
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812153640359259e-06,
+      "loss": 1.0064,
+      "step": 12482
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812058092872158e-06,
+      "loss": 1.0905,
+      "step": 12483
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8119625453850566e-06,
+      "loss": 1.036,
+      "step": 12484
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8118669978979556e-06,
+      "loss": 1.587,
+      "step": 12485
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8117714504108542e-06,
+      "loss": 1.3305,
+      "step": 12486
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8116759029237533e-06,
+      "loss": 1.2134,
+      "step": 12487
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.811580355436652e-06,
+      "loss": 0.789,
+      "step": 12488
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8114848079495514e-06,
+      "loss": 0.8506,
+      "step": 12489
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8113892604624504e-06,
+      "loss": 0.9686,
+      "step": 12490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.811293712975349e-06,
+      "loss": 1.0108,
+      "step": 12491
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.811198165488248e-06,
+      "loss": 1.1833,
+      "step": 12492
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8111026180011467e-06,
+      "loss": 0.7555,
+      "step": 12493
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8110070705140457e-06,
+      "loss": 1.0085,
+      "step": 12494
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8109115230269443e-06,
+      "loss": 0.8711,
+      "step": 12495
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8108159755398438e-06,
+      "loss": 0.8443,
+      "step": 12496
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.810720428052743e-06,
+      "loss": 1.1745,
+      "step": 12497
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8106248805656414e-06,
+      "loss": 1.1542,
+      "step": 12498
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8105293330785405e-06,
+      "loss": 0.8768,
+      "step": 12499
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.810433785591439e-06,
+      "loss": 1.2675,
+      "step": 12500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.810338238104338e-06,
+      "loss": 0.813,
+      "step": 12501
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8102426906172367e-06,
+      "loss": 1.0087,
+      "step": 12502
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.810147143130136e-06,
+      "loss": 1.1099,
+      "step": 12503
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8100515956430352e-06,
+      "loss": 1.2985,
+      "step": 12504
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.809956048155934e-06,
+      "loss": 0.8331,
+      "step": 12505
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.809860500668833e-06,
+      "loss": 0.9133,
+      "step": 12506
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8097649531817315e-06,
+      "loss": 1.2995,
+      "step": 12507
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8096694056946305e-06,
+      "loss": 0.9681,
+      "step": 12508
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.809573858207529e-06,
+      "loss": 0.8322,
+      "step": 12509
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8094783107204286e-06,
+      "loss": 1.3765,
+      "step": 12510
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8093827632333272e-06,
+      "loss": 1.2781,
+      "step": 12511
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8092872157462263e-06,
+      "loss": 0.8584,
+      "step": 12512
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8091916682591253e-06,
+      "loss": 0.9182,
+      "step": 12513
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.809096120772024e-06,
+      "loss": 1.0763,
+      "step": 12514
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.809000573284923e-06,
+      "loss": 1.0996,
+      "step": 12515
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8089050257978216e-06,
+      "loss": 0.5715,
+      "step": 12516
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.808809478310721e-06,
+      "loss": 1.0526,
+      "step": 12517
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8087139308236197e-06,
+      "loss": 1.4113,
+      "step": 12518
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8086183833365187e-06,
+      "loss": 0.8439,
+      "step": 12519
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8085228358494173e-06,
+      "loss": 0.9142,
+      "step": 12520
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8084272883623164e-06,
+      "loss": 0.6731,
+      "step": 12521
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.808331740875215e-06,
+      "loss": 1.479,
+      "step": 12522
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.808236193388114e-06,
+      "loss": 0.968,
+      "step": 12523
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.808140645901013e-06,
+      "loss": 0.8256,
+      "step": 12524
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.808045098413912e-06,
+      "loss": 1.1898,
+      "step": 12525
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.807949550926811e-06,
+      "loss": 0.531,
+      "step": 12526
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8078540034397097e-06,
+      "loss": 0.7807,
+      "step": 12527
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8077584559526088e-06,
+      "loss": 1.0835,
+      "step": 12528
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8076629084655074e-06,
+      "loss": 0.9853,
+      "step": 12529
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8075673609784064e-06,
+      "loss": 0.7014,
+      "step": 12530
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.807471813491305e-06,
+      "loss": 1.3093,
+      "step": 12531
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8073762660042045e-06,
+      "loss": 0.9227,
+      "step": 12532
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8072807185171035e-06,
+      "loss": 1.1915,
+      "step": 12533
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.807185171030002e-06,
+      "loss": 0.7488,
+      "step": 12534
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.807089623542901e-06,
+      "loss": 0.6351,
+      "step": 12535
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8069940760558e-06,
+      "loss": 0.6762,
+      "step": 12536
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.806898528568699e-06,
+      "loss": 1.2849,
+      "step": 12537
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8068029810815975e-06,
+      "loss": 1.1295,
+      "step": 12538
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.806707433594497e-06,
+      "loss": 0.8865,
+      "step": 12539
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.806611886107396e-06,
+      "loss": 1.055,
+      "step": 12540
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8065163386202946e-06,
+      "loss": 1.1638,
+      "step": 12541
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8064207911331936e-06,
+      "loss": 0.6462,
+      "step": 12542
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8063252436460922e-06,
+      "loss": 1.1975,
+      "step": 12543
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8062296961589913e-06,
+      "loss": 0.98,
+      "step": 12544
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.80613414867189e-06,
+      "loss": 1.1447,
+      "step": 12545
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8060386011847894e-06,
+      "loss": 1.3409,
+      "step": 12546
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8059430536976884e-06,
+      "loss": 1.1979,
+      "step": 12547
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.805847506210587e-06,
+      "loss": 1.0607,
+      "step": 12548
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.805751958723486e-06,
+      "loss": 0.7466,
+      "step": 12549
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8056564112363847e-06,
+      "loss": 0.6148,
+      "step": 12550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8055608637492837e-06,
+      "loss": 0.8536,
+      "step": 12551
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8054653162621823e-06,
+      "loss": 0.8852,
+      "step": 12552
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8053697687750818e-06,
+      "loss": 1.3131,
+      "step": 12553
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.805274221287981e-06,
+      "loss": 0.7269,
+      "step": 12554
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8051786738008794e-06,
+      "loss": 0.7332,
+      "step": 12555
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8050831263137785e-06,
+      "loss": 1.6297,
+      "step": 12556
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804987578826677e-06,
+      "loss": 1.1793,
+      "step": 12557
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804892031339576e-06,
+      "loss": 0.8664,
+      "step": 12558
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8047964838524747e-06,
+      "loss": 1.0212,
+      "step": 12559
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8047009363653738e-06,
+      "loss": 1.0067,
+      "step": 12560
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804605388878273e-06,
+      "loss": 1.0781,
+      "step": 12561
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804509841391172e-06,
+      "loss": 0.8125,
+      "step": 12562
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804414293904071e-06,
+      "loss": 1.1697,
+      "step": 12563
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8043187464169695e-06,
+      "loss": 1.1107,
+      "step": 12564
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8042231989298685e-06,
+      "loss": 0.9575,
+      "step": 12565
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804127651442767e-06,
+      "loss": 1.0235,
+      "step": 12566
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.804032103955666e-06,
+      "loss": 1.0514,
+      "step": 12567
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8039365564685652e-06,
+      "loss": 1.3313,
+      "step": 12568
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8038410089814643e-06,
+      "loss": 1.0262,
+      "step": 12569
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.803745461494363e-06,
+      "loss": 0.8495,
+      "step": 12570
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.803649914007262e-06,
+      "loss": 0.6933,
+      "step": 12571
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8035543665201605e-06,
+      "loss": 0.6085,
+      "step": 12572
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8034588190330596e-06,
+      "loss": 1.164,
+      "step": 12573
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.803363271545958e-06,
+      "loss": 1.2352,
+      "step": 12574
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8032677240588577e-06,
+      "loss": 0.8781,
+      "step": 12575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8031721765717567e-06,
+      "loss": 0.9752,
+      "step": 12576
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8030766290846553e-06,
+      "loss": 0.8579,
+      "step": 12577
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8029810815975544e-06,
+      "loss": 1.1211,
+      "step": 12578
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.802885534110453e-06,
+      "loss": 1.0085,
+      "step": 12579
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.802789986623352e-06,
+      "loss": 1.2109,
+      "step": 12580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8026944391362506e-06,
+      "loss": 0.6722,
+      "step": 12581
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.80259889164915e-06,
+      "loss": 0.7277,
+      "step": 12582
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.802503344162049e-06,
+      "loss": 0.8229,
+      "step": 12583
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8024077966749477e-06,
+      "loss": 0.666,
+      "step": 12584
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8023122491878468e-06,
+      "loss": 0.6778,
+      "step": 12585
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8022167017007454e-06,
+      "loss": 1.2347,
+      "step": 12586
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8021211542136444e-06,
+      "loss": 0.4484,
+      "step": 12587
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.802025606726543e-06,
+      "loss": 0.8583,
+      "step": 12588
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8019300592394425e-06,
+      "loss": 0.8631,
+      "step": 12589
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8018345117523415e-06,
+      "loss": 0.9157,
+      "step": 12590
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.80173896426524e-06,
+      "loss": 0.547,
+      "step": 12591
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801643416778139e-06,
+      "loss": 1.2119,
+      "step": 12592
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801547869291038e-06,
+      "loss": 1.0401,
+      "step": 12593
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801452321803937e-06,
+      "loss": 1.1435,
+      "step": 12594
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8013567743168355e-06,
+      "loss": 0.7166,
+      "step": 12595
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8012612268297345e-06,
+      "loss": 1.1364,
+      "step": 12596
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801165679342634e-06,
+      "loss": 1.3476,
+      "step": 12597
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8010701318555326e-06,
+      "loss": 1.0569,
+      "step": 12598
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8009745843684316e-06,
+      "loss": 0.9728,
+      "step": 12599
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8008790368813302e-06,
+      "loss": 1.1413,
+      "step": 12600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8007834893942293e-06,
+      "loss": 1.1763,
+      "step": 12601
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.800687941907128e-06,
+      "loss": 1.0096,
+      "step": 12602
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.800592394420027e-06,
+      "loss": 1.2069,
+      "step": 12603
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8004968469329264e-06,
+      "loss": 1.1102,
+      "step": 12604
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.800401299445825e-06,
+      "loss": 0.9204,
+      "step": 12605
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.800305751958724e-06,
+      "loss": 0.9444,
+      "step": 12606
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8002102044716227e-06,
+      "loss": 1.1188,
+      "step": 12607
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8001146569845217e-06,
+      "loss": 0.9843,
+      "step": 12608
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.8000191094974203e-06,
+      "loss": 1.149,
+      "step": 12609
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7999235620103194e-06,
+      "loss": 0.9293,
+      "step": 12610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7998280145232184e-06,
+      "loss": 1.3053,
+      "step": 12611
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7997324670361174e-06,
+      "loss": 1.1145,
+      "step": 12612
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.799636919549016e-06,
+      "loss": 0.8233,
+      "step": 12613
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.799541372061915e-06,
+      "loss": 1.1412,
+      "step": 12614
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.799445824574814e-06,
+      "loss": 1.418,
+      "step": 12615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7993502770877127e-06,
+      "loss": 1.2728,
+      "step": 12616
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7992547296006118e-06,
+      "loss": 1.1056,
+      "step": 12617
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.799159182113511e-06,
+      "loss": 1.0706,
+      "step": 12618
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.79906363462641e-06,
+      "loss": 1.026,
+      "step": 12619
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7989680871393085e-06,
+      "loss": 0.9513,
+      "step": 12620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7988725396522075e-06,
+      "loss": 1.2793,
+      "step": 12621
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.798776992165106e-06,
+      "loss": 1.1473,
+      "step": 12622
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.798681444678005e-06,
+      "loss": 1.0592,
+      "step": 12623
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7985858971909038e-06,
+      "loss": 0.8196,
+      "step": 12624
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7984903497038032e-06,
+      "loss": 1.0834,
+      "step": 12625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7983948022167023e-06,
+      "loss": 1.0363,
+      "step": 12626
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.798299254729601e-06,
+      "loss": 0.5314,
+      "step": 12627
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7982037072425e-06,
+      "loss": 0.8318,
+      "step": 12628
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7981081597553985e-06,
+      "loss": 1.1847,
+      "step": 12629
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7980126122682976e-06,
+      "loss": 0.7441,
+      "step": 12630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.797917064781196e-06,
+      "loss": 0.851,
+      "step": 12631
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7978215172940952e-06,
+      "loss": 0.5674,
+      "step": 12632
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7977259698069947e-06,
+      "loss": 0.7769,
+      "step": 12633
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7976304223198933e-06,
+      "loss": 0.6865,
+      "step": 12634
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7975348748327924e-06,
+      "loss": 0.9075,
+      "step": 12635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.797439327345691e-06,
+      "loss": 0.9892,
+      "step": 12636
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.79734377985859e-06,
+      "loss": 1.1896,
+      "step": 12637
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7972482323714886e-06,
+      "loss": 1.0127,
+      "step": 12638
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7971526848843877e-06,
+      "loss": 1.056,
+      "step": 12639
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.797057137397287e-06,
+      "loss": 1.1452,
+      "step": 12640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7969615899101857e-06,
+      "loss": 1.201,
+      "step": 12641
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7968660424230848e-06,
+      "loss": 1.4663,
+      "step": 12642
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7967704949359834e-06,
+      "loss": 0.8808,
+      "step": 12643
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7966749474488824e-06,
+      "loss": 1.0991,
+      "step": 12644
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796579399961781e-06,
+      "loss": 1.1485,
+      "step": 12645
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.79648385247468e-06,
+      "loss": 0.7713,
+      "step": 12646
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7963883049875795e-06,
+      "loss": 1.2458,
+      "step": 12647
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796292757500478e-06,
+      "loss": 1.0205,
+      "step": 12648
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796197210013377e-06,
+      "loss": 0.7824,
+      "step": 12649
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796101662526276e-06,
+      "loss": 0.839,
+      "step": 12650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.796006115039175e-06,
+      "loss": 1.2958,
+      "step": 12651
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7959105675520735e-06,
+      "loss": 1.5627,
+      "step": 12652
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7958150200649725e-06,
+      "loss": 1.4623,
+      "step": 12653
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.795719472577872e-06,
+      "loss": 0.9989,
+      "step": 12654
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7956239250907706e-06,
+      "loss": 0.9757,
+      "step": 12655
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7955283776036696e-06,
+      "loss": 0.8907,
+      "step": 12656
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7954328301165682e-06,
+      "loss": 0.6755,
+      "step": 12657
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7953372826294673e-06,
+      "loss": 1.0998,
+      "step": 12658
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.795241735142366e-06,
+      "loss": 1.2879,
+      "step": 12659
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.795146187655265e-06,
+      "loss": 0.8067,
+      "step": 12660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.795050640168164e-06,
+      "loss": 0.7272,
+      "step": 12661
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.794955092681063e-06,
+      "loss": 1.1734,
+      "step": 12662
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7948595451939616e-06,
+      "loss": 0.9822,
+      "step": 12663
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7947639977068607e-06,
+      "loss": 1.2967,
+      "step": 12664
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7946684502197597e-06,
+      "loss": 0.9082,
+      "step": 12665
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7945729027326583e-06,
+      "loss": 1.2518,
+      "step": 12666
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7944773552455574e-06,
+      "loss": 0.6701,
+      "step": 12667
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.794381807758456e-06,
+      "loss": 1.2273,
+      "step": 12668
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7942862602713554e-06,
+      "loss": 0.7616,
+      "step": 12669
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.794190712784254e-06,
+      "loss": 0.8064,
+      "step": 12670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.794095165297153e-06,
+      "loss": 1.0242,
+      "step": 12671
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7939996178100517e-06,
+      "loss": 1.3509,
+      "step": 12672
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7939040703229507e-06,
+      "loss": 0.9271,
+      "step": 12673
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7938085228358494e-06,
+      "loss": 1.1256,
+      "step": 12674
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7937129753487484e-06,
+      "loss": 1.0069,
+      "step": 12675
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.793617427861648e-06,
+      "loss": 1.1671,
+      "step": 12676
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7935218803745465e-06,
+      "loss": 0.9032,
+      "step": 12677
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7934263328874455e-06,
+      "loss": 0.7697,
+      "step": 12678
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.793330785400344e-06,
+      "loss": 1.3571,
+      "step": 12679
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.793235237913243e-06,
+      "loss": 0.9143,
+      "step": 12680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7931396904261418e-06,
+      "loss": 1.2645,
+      "step": 12681
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.793044142939041e-06,
+      "loss": 0.7118,
+      "step": 12682
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7929485954519403e-06,
+      "loss": 0.9939,
+      "step": 12683
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792853047964839e-06,
+      "loss": 0.8153,
+      "step": 12684
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792757500477738e-06,
+      "loss": 1.2698,
+      "step": 12685
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7926619529906365e-06,
+      "loss": 1.4273,
+      "step": 12686
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7925664055035356e-06,
+      "loss": 1.7197,
+      "step": 12687
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792470858016434e-06,
+      "loss": 0.8127,
+      "step": 12688
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7923753105293332e-06,
+      "loss": 0.8144,
+      "step": 12689
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7922797630422327e-06,
+      "loss": 1.6658,
+      "step": 12690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7921842155551313e-06,
+      "loss": 1.3201,
+      "step": 12691
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7920886680680304e-06,
+      "loss": 1.3576,
+      "step": 12692
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791993120580929e-06,
+      "loss": 0.4046,
+      "step": 12693
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791897573093828e-06,
+      "loss": 0.9296,
+      "step": 12694
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7918020256067266e-06,
+      "loss": 1.2017,
+      "step": 12695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7917064781196257e-06,
+      "loss": 0.9808,
+      "step": 12696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791610930632525e-06,
+      "loss": 1.241,
+      "step": 12697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7915153831454237e-06,
+      "loss": 1.2131,
+      "step": 12698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7914198356583228e-06,
+      "loss": 0.9249,
+      "step": 12699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7913242881712214e-06,
+      "loss": 1.329,
+      "step": 12700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7912287406841204e-06,
+      "loss": 1.4195,
+      "step": 12701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791133193197019e-06,
+      "loss": 1.1176,
+      "step": 12702
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791037645709918e-06,
+      "loss": 1.4122,
+      "step": 12703
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7909420982228167e-06,
+      "loss": 0.5614,
+      "step": 12704
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790846550735716e-06,
+      "loss": 1.096,
+      "step": 12705
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790751003248615e-06,
+      "loss": 0.5583,
+      "step": 12706
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790655455761514e-06,
+      "loss": 0.7893,
+      "step": 12707
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790559908274413e-06,
+      "loss": 1.4495,
+      "step": 12708
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7904643607873115e-06,
+      "loss": 0.7949,
+      "step": 12709
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7903688133002105e-06,
+      "loss": 0.8657,
+      "step": 12710
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790273265813109e-06,
+      "loss": 0.8438,
+      "step": 12711
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7901777183260086e-06,
+      "loss": 1.041,
+      "step": 12712
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790082170838907e-06,
+      "loss": 1.1196,
+      "step": 12713
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7899866233518062e-06,
+      "loss": 1.1662,
+      "step": 12714
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789891075864705e-06,
+      "loss": 1.0979,
+      "step": 12715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789795528377604e-06,
+      "loss": 1.1049,
+      "step": 12716
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789699980890503e-06,
+      "loss": 1.1428,
+      "step": 12717
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7896044334034015e-06,
+      "loss": 0.7974,
+      "step": 12718
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789508885916301e-06,
+      "loss": 0.8305,
+      "step": 12719
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7894133384291996e-06,
+      "loss": 0.7786,
+      "step": 12720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7893177909420987e-06,
+      "loss": 0.833,
+      "step": 12721
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7892222434549973e-06,
+      "loss": 1.007,
+      "step": 12722
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7891266959678963e-06,
+      "loss": 0.8107,
+      "step": 12723
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789031148480795e-06,
+      "loss": 0.7545,
+      "step": 12724
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.788935600993694e-06,
+      "loss": 1.2193,
+      "step": 12725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7888400535065934e-06,
+      "loss": 1.3175,
+      "step": 12726
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.788744506019492e-06,
+      "loss": 0.9915,
+      "step": 12727
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.788648958532391e-06,
+      "loss": 0.9866,
+      "step": 12728
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7885534110452897e-06,
+      "loss": 0.8157,
+      "step": 12729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7884578635581887e-06,
+      "loss": 0.7972,
+      "step": 12730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7883623160710874e-06,
+      "loss": 0.9477,
+      "step": 12731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7882667685839864e-06,
+      "loss": 1.2899,
+      "step": 12732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.788171221096886e-06,
+      "loss": 0.8596,
+      "step": 12733
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7880756736097845e-06,
+      "loss": 1.2402,
+      "step": 12734
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7879801261226835e-06,
+      "loss": 1.0279,
+      "step": 12735
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787884578635582e-06,
+      "loss": 0.9172,
+      "step": 12736
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787789031148481e-06,
+      "loss": 0.8925,
+      "step": 12737
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7876934836613798e-06,
+      "loss": 0.8757,
+      "step": 12738
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787597936174279e-06,
+      "loss": 0.9977,
+      "step": 12739
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7875023886871774e-06,
+      "loss": 0.8988,
+      "step": 12740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787406841200077e-06,
+      "loss": 1.0853,
+      "step": 12741
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787311293712976e-06,
+      "loss": 1.2363,
+      "step": 12742
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7872157462258745e-06,
+      "loss": 1.0491,
+      "step": 12743
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7871201987387736e-06,
+      "loss": 0.9072,
+      "step": 12744
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787024651251672e-06,
+      "loss": 0.6412,
+      "step": 12745
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7869291037645712e-06,
+      "loss": 0.9617,
+      "step": 12746
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.78683355627747e-06,
+      "loss": 1.3523,
+      "step": 12747
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7867380087903693e-06,
+      "loss": 0.7341,
+      "step": 12748
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7866424613032684e-06,
+      "loss": 1.4117,
+      "step": 12749
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786546913816167e-06,
+      "loss": 1.0728,
+      "step": 12750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786451366329066e-06,
+      "loss": 0.5949,
+      "step": 12751
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7863558188419646e-06,
+      "loss": 0.99,
+      "step": 12752
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7862602713548637e-06,
+      "loss": 0.6751,
+      "step": 12753
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7861647238677623e-06,
+      "loss": 0.8779,
+      "step": 12754
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7860691763806617e-06,
+      "loss": 0.5411,
+      "step": 12755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7859736288935608e-06,
+      "loss": 0.9577,
+      "step": 12756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7858780814064594e-06,
+      "loss": 1.475,
+      "step": 12757
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7857825339193584e-06,
+      "loss": 0.6951,
+      "step": 12758
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.785686986432257e-06,
+      "loss": 1.0187,
+      "step": 12759
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.785591438945156e-06,
+      "loss": 0.9136,
+      "step": 12760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7854958914580547e-06,
+      "loss": 0.8464,
+      "step": 12761
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.785400343970954e-06,
+      "loss": 1.0573,
+      "step": 12762
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7853047964838528e-06,
+      "loss": 0.8259,
+      "step": 12763
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.785209248996752e-06,
+      "loss": 0.9843,
+      "step": 12764
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7851137015096504e-06,
+      "loss": 0.9372,
+      "step": 12765
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7850181540225495e-06,
+      "loss": 1.0168,
+      "step": 12766
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7849226065354485e-06,
+      "loss": 1.0556,
+      "step": 12767
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.784827059048347e-06,
+      "loss": 0.8719,
+      "step": 12768
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7847315115612466e-06,
+      "loss": 0.9418,
+      "step": 12769
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.784635964074145e-06,
+      "loss": 1.3131,
+      "step": 12770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7845404165870442e-06,
+      "loss": 1.0245,
+      "step": 12771
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.784444869099943e-06,
+      "loss": 0.8315,
+      "step": 12772
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.784349321612842e-06,
+      "loss": 0.7862,
+      "step": 12773
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7842537741257405e-06,
+      "loss": 0.732,
+      "step": 12774
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7841582266386395e-06,
+      "loss": 0.9621,
+      "step": 12775
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.784062679151538e-06,
+      "loss": 0.7799,
+      "step": 12776
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7839671316644376e-06,
+      "loss": 1.2429,
+      "step": 12777
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7838715841773367e-06,
+      "loss": 0.8337,
+      "step": 12778
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7837760366902353e-06,
+      "loss": 1.0297,
+      "step": 12779
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7836804892031343e-06,
+      "loss": 1.0941,
+      "step": 12780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.783584941716033e-06,
+      "loss": 0.7634,
+      "step": 12781
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.783489394228932e-06,
+      "loss": 1.0354,
+      "step": 12782
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7833938467418306e-06,
+      "loss": 1.1337,
+      "step": 12783
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78329829925473e-06,
+      "loss": 1.0121,
+      "step": 12784
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.783202751767629e-06,
+      "loss": 1.572,
+      "step": 12785
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7831072042805277e-06,
+      "loss": 1.0892,
+      "step": 12786
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7830116567934267e-06,
+      "loss": 0.7286,
+      "step": 12787
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7829161093063254e-06,
+      "loss": 0.991,
+      "step": 12788
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7828205618192244e-06,
+      "loss": 0.9749,
+      "step": 12789
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.782725014332123e-06,
+      "loss": 0.9792,
+      "step": 12790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7826294668450225e-06,
+      "loss": 1.243,
+      "step": 12791
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7825339193579215e-06,
+      "loss": 0.9991,
+      "step": 12792
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78243837187082e-06,
+      "loss": 1.2373,
+      "step": 12793
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.782342824383719e-06,
+      "loss": 0.9753,
+      "step": 12794
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7822472768966178e-06,
+      "loss": 1.0487,
+      "step": 12795
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.782151729409517e-06,
+      "loss": 0.7458,
+      "step": 12796
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7820561819224154e-06,
+      "loss": 0.7773,
+      "step": 12797
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781960634435315e-06,
+      "loss": 1.0532,
+      "step": 12798
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781865086948214e-06,
+      "loss": 1.5693,
+      "step": 12799
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7817695394611125e-06,
+      "loss": 0.6523,
+      "step": 12800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7816739919740116e-06,
+      "loss": 1.2004,
+      "step": 12801
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.78157844448691e-06,
+      "loss": 0.8385,
+      "step": 12802
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7814828969998092e-06,
+      "loss": 0.7309,
+      "step": 12803
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781387349512708e-06,
+      "loss": 0.8826,
+      "step": 12804
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7812918020256073e-06,
+      "loss": 0.8341,
+      "step": 12805
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781196254538506e-06,
+      "loss": 1.0453,
+      "step": 12806
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781100707051405e-06,
+      "loss": 1.163,
+      "step": 12807
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781005159564304e-06,
+      "loss": 0.8184,
+      "step": 12808
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7809096120772026e-06,
+      "loss": 0.8809,
+      "step": 12809
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7808140645901017e-06,
+      "loss": 0.6182,
+      "step": 12810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7807185171030003e-06,
+      "loss": 0.7548,
+      "step": 12811
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7806229696158993e-06,
+      "loss": 0.9772,
+      "step": 12812
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7805274221287984e-06,
+      "loss": 0.743,
+      "step": 12813
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7804318746416974e-06,
+      "loss": 0.8827,
+      "step": 12814
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.780336327154596e-06,
+      "loss": 0.8995,
+      "step": 12815
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.780240779667495e-06,
+      "loss": 0.9525,
+      "step": 12816
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7801452321803937e-06,
+      "loss": 0.6697,
+      "step": 12817
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7800496846932927e-06,
+      "loss": 1.2896,
+      "step": 12818
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7799541372061917e-06,
+      "loss": 0.8722,
+      "step": 12819
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7798585897190908e-06,
+      "loss": 0.7439,
+      "step": 12820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77976304223199e-06,
+      "loss": 1.3128,
+      "step": 12821
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7796674947448884e-06,
+      "loss": 1.4001,
+      "step": 12822
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7795719472577875e-06,
+      "loss": 0.4695,
+      "step": 12823
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.779476399770686e-06,
+      "loss": 0.874,
+      "step": 12824
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.779380852283585e-06,
+      "loss": 1.4638,
+      "step": 12825
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7792853047964837e-06,
+      "loss": 1.2259,
+      "step": 12826
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.779189757309383e-06,
+      "loss": 1.6599,
+      "step": 12827
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7790942098222822e-06,
+      "loss": 1.2539,
+      "step": 12828
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.778998662335181e-06,
+      "loss": 1.0535,
+      "step": 12829
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77890311484808e-06,
+      "loss": 0.8853,
+      "step": 12830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7788075673609785e-06,
+      "loss": 0.9341,
+      "step": 12831
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7787120198738775e-06,
+      "loss": 1.0207,
+      "step": 12832
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.778616472386776e-06,
+      "loss": 1.0001,
+      "step": 12833
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7785209248996756e-06,
+      "loss": 1.0591,
+      "step": 12834
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7784253774125747e-06,
+      "loss": 1.0604,
+      "step": 12835
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7783298299254733e-06,
+      "loss": 0.7455,
+      "step": 12836
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7782342824383723e-06,
+      "loss": 1.3908,
+      "step": 12837
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.778138734951271e-06,
+      "loss": 0.9356,
+      "step": 12838
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77804318746417e-06,
+      "loss": 0.8034,
+      "step": 12839
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7779476399770686e-06,
+      "loss": 0.8549,
+      "step": 12840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777852092489968e-06,
+      "loss": 1.064,
+      "step": 12841
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777756545002867e-06,
+      "loss": 0.7338,
+      "step": 12842
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7776609975157657e-06,
+      "loss": 0.8878,
+      "step": 12843
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7775654500286647e-06,
+      "loss": 1.5234,
+      "step": 12844
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7774699025415634e-06,
+      "loss": 1.1011,
+      "step": 12845
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7773743550544624e-06,
+      "loss": 0.8668,
+      "step": 12846
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777278807567361e-06,
+      "loss": 0.8837,
+      "step": 12847
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77718326008026e-06,
+      "loss": 1.1736,
+      "step": 12848
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7770877125931595e-06,
+      "loss": 0.9306,
+      "step": 12849
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776992165106058e-06,
+      "loss": 0.921,
+      "step": 12850
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776896617618957e-06,
+      "loss": 1.2516,
+      "step": 12851
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7768010701318558e-06,
+      "loss": 1.2531,
+      "step": 12852
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776705522644755e-06,
+      "loss": 1.3149,
+      "step": 12853
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7766099751576534e-06,
+      "loss": 0.7179,
+      "step": 12854
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7765144276705525e-06,
+      "loss": 0.7247,
+      "step": 12855
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7764188801834515e-06,
+      "loss": 1.0375,
+      "step": 12856
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7763233326963505e-06,
+      "loss": 0.9007,
+      "step": 12857
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7762277852092496e-06,
+      "loss": 0.8375,
+      "step": 12858
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776132237722148e-06,
+      "loss": 0.7671,
+      "step": 12859
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7760366902350472e-06,
+      "loss": 0.881,
+      "step": 12860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.775941142747946e-06,
+      "loss": 1.1549,
+      "step": 12861
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.775845595260845e-06,
+      "loss": 0.8337,
+      "step": 12862
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.775750047773744e-06,
+      "loss": 1.2619,
+      "step": 12863
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.775654500286643e-06,
+      "loss": 1.1052,
+      "step": 12864
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7755589527995416e-06,
+      "loss": 1.3094,
+      "step": 12865
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7754634053124406e-06,
+      "loss": 1.2041,
+      "step": 12866
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7753678578253392e-06,
+      "loss": 0.9861,
+      "step": 12867
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7752723103382383e-06,
+      "loss": 1.1887,
+      "step": 12868
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7751767628511373e-06,
+      "loss": 0.9452,
+      "step": 12869
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7750812153640364e-06,
+      "loss": 0.8564,
+      "step": 12870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7749856678769354e-06,
+      "loss": 0.6521,
+      "step": 12871
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774890120389834e-06,
+      "loss": 0.9482,
+      "step": 12872
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774794572902733e-06,
+      "loss": 1.2488,
+      "step": 12873
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7746990254156317e-06,
+      "loss": 0.8746,
+      "step": 12874
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7746034779285307e-06,
+      "loss": 1.0584,
+      "step": 12875
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7745079304414293e-06,
+      "loss": 1.1825,
+      "step": 12876
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7744123829543288e-06,
+      "loss": 0.9957,
+      "step": 12877
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774316835467228e-06,
+      "loss": 1.0801,
+      "step": 12878
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7742212879801264e-06,
+      "loss": 0.7494,
+      "step": 12879
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7741257404930255e-06,
+      "loss": 0.8236,
+      "step": 12880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.774030193005924e-06,
+      "loss": 1.2411,
+      "step": 12881
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.773934645518823e-06,
+      "loss": 1.3853,
+      "step": 12882
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7738390980317217e-06,
+      "loss": 1.1396,
+      "step": 12883
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7737435505446208e-06,
+      "loss": 1.1435,
+      "step": 12884
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7736480030575202e-06,
+      "loss": 0.5825,
+      "step": 12885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.773552455570419e-06,
+      "loss": 0.9439,
+      "step": 12886
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.773456908083318e-06,
+      "loss": 0.8763,
+      "step": 12887
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7733613605962165e-06,
+      "loss": 0.9831,
+      "step": 12888
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7732658131091155e-06,
+      "loss": 1.2151,
+      "step": 12889
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.773170265622014e-06,
+      "loss": 0.8699,
+      "step": 12890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.773074718134913e-06,
+      "loss": 0.6709,
+      "step": 12891
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7729791706478127e-06,
+      "loss": 0.7845,
+      "step": 12892
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7728836231607113e-06,
+      "loss": 1.4492,
+      "step": 12893
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7727880756736103e-06,
+      "loss": 1.4237,
+      "step": 12894
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772692528186509e-06,
+      "loss": 0.5774,
+      "step": 12895
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772596980699408e-06,
+      "loss": 0.9344,
+      "step": 12896
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7725014332123066e-06,
+      "loss": 0.9535,
+      "step": 12897
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7724058857252056e-06,
+      "loss": 1.4953,
+      "step": 12898
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772310338238105e-06,
+      "loss": 1.2962,
+      "step": 12899
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7722147907510037e-06,
+      "loss": 0.8986,
+      "step": 12900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7721192432639027e-06,
+      "loss": 1.0228,
+      "step": 12901
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7720236957768013e-06,
+      "loss": 0.9955,
+      "step": 12902
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7719281482897004e-06,
+      "loss": 1.078,
+      "step": 12903
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771832600802599e-06,
+      "loss": 1.0343,
+      "step": 12904
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771737053315498e-06,
+      "loss": 1.19,
+      "step": 12905
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771641505828397e-06,
+      "loss": 1.1353,
+      "step": 12906
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771545958341296e-06,
+      "loss": 0.957,
+      "step": 12907
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7714504108541947e-06,
+      "loss": 1.1994,
+      "step": 12908
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7713548633670938e-06,
+      "loss": 1.2156,
+      "step": 12909
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771259315879993e-06,
+      "loss": 0.8388,
+      "step": 12910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7711637683928914e-06,
+      "loss": 0.8972,
+      "step": 12911
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7710682209057905e-06,
+      "loss": 1.5871,
+      "step": 12912
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7709726734186895e-06,
+      "loss": 0.9969,
+      "step": 12913
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7708771259315885e-06,
+      "loss": 0.8698,
+      "step": 12914
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.770781578444487e-06,
+      "loss": 1.1174,
+      "step": 12915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.770686030957386e-06,
+      "loss": 1.1383,
+      "step": 12916
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.770590483470285e-06,
+      "loss": 1.3211,
+      "step": 12917
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.770494935983184e-06,
+      "loss": 0.9102,
+      "step": 12918
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7703993884960825e-06,
+      "loss": 1.1159,
+      "step": 12919
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7703038410089815e-06,
+      "loss": 0.8628,
+      "step": 12920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.770208293521881e-06,
+      "loss": 1.0612,
+      "step": 12921
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7701127460347796e-06,
+      "loss": 1.2631,
+      "step": 12922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7700171985476786e-06,
+      "loss": 1.4488,
+      "step": 12923
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7699216510605772e-06,
+      "loss": 1.3421,
+      "step": 12924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7698261035734763e-06,
+      "loss": 1.0916,
+      "step": 12925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.769730556086375e-06,
+      "loss": 1.2008,
+      "step": 12926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.769635008599274e-06,
+      "loss": 0.8267,
+      "step": 12927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7695394611121734e-06,
+      "loss": 1.5648,
+      "step": 12928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.769443913625072e-06,
+      "loss": 1.0185,
+      "step": 12929
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.769348366137971e-06,
+      "loss": 1.5269,
+      "step": 12930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7692528186508697e-06,
+      "loss": 0.8186,
+      "step": 12931
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7691572711637687e-06,
+      "loss": 1.3132,
+      "step": 12932
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7690617236766673e-06,
+      "loss": 0.8562,
+      "step": 12933
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7689661761895663e-06,
+      "loss": 0.7961,
+      "step": 12934
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.768870628702466e-06,
+      "loss": 1.1262,
+      "step": 12935
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7687750812153644e-06,
+      "loss": 0.9933,
+      "step": 12936
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7686795337282635e-06,
+      "loss": 1.3248,
+      "step": 12937
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.768583986241162e-06,
+      "loss": 0.5819,
+      "step": 12938
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.768488438754061e-06,
+      "loss": 1.394,
+      "step": 12939
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7683928912669597e-06,
+      "loss": 0.7418,
+      "step": 12940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7682973437798588e-06,
+      "loss": 0.9203,
+      "step": 12941
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7682017962927582e-06,
+      "loss": 1.0044,
+      "step": 12942
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.768106248805657e-06,
+      "loss": 1.0368,
+      "step": 12943
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.768010701318556e-06,
+      "loss": 0.9776,
+      "step": 12944
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7679151538314545e-06,
+      "loss": 0.664,
+      "step": 12945
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7678196063443535e-06,
+      "loss": 0.6219,
+      "step": 12946
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767724058857252e-06,
+      "loss": 1.1065,
+      "step": 12947
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767628511370151e-06,
+      "loss": 1.4937,
+      "step": 12948
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7675329638830507e-06,
+      "loss": 1.1174,
+      "step": 12949
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7674374163959493e-06,
+      "loss": 0.9734,
+      "step": 12950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7673418689088483e-06,
+      "loss": 1.2853,
+      "step": 12951
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767246321421747e-06,
+      "loss": 0.6843,
+      "step": 12952
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767150773934646e-06,
+      "loss": 0.661,
+      "step": 12953
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7670552264475446e-06,
+      "loss": 0.4945,
+      "step": 12954
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7669596789604436e-06,
+      "loss": 0.5771,
+      "step": 12955
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7668641314733422e-06,
+      "loss": 0.7093,
+      "step": 12956
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7667685839862417e-06,
+      "loss": 0.8248,
+      "step": 12957
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7666730364991403e-06,
+      "loss": 1.0437,
+      "step": 12958
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7665774890120393e-06,
+      "loss": 1.3654,
+      "step": 12959
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7664819415249384e-06,
+      "loss": 0.9203,
+      "step": 12960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766386394037837e-06,
+      "loss": 0.9432,
+      "step": 12961
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766290846550736e-06,
+      "loss": 1.6153,
+      "step": 12962
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7661952990636347e-06,
+      "loss": 1.1255,
+      "step": 12963
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766099751576534e-06,
+      "loss": 0.8634,
+      "step": 12964
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7660042040894327e-06,
+      "loss": 0.7131,
+      "step": 12965
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7659086566023318e-06,
+      "loss": 0.6145,
+      "step": 12966
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7658131091152304e-06,
+      "loss": 0.8158,
+      "step": 12967
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7657175616281294e-06,
+      "loss": 1.3606,
+      "step": 12968
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765622014141028e-06,
+      "loss": 0.6199,
+      "step": 12969
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765526466653927e-06,
+      "loss": 0.7431,
+      "step": 12970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7654309191668265e-06,
+      "loss": 0.739,
+      "step": 12971
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765335371679725e-06,
+      "loss": 0.8303,
+      "step": 12972
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765239824192624e-06,
+      "loss": 0.9166,
+      "step": 12973
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765144276705523e-06,
+      "loss": 1.0009,
+      "step": 12974
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.765048729218422e-06,
+      "loss": 1.1455,
+      "step": 12975
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7649531817313205e-06,
+      "loss": 0.9463,
+      "step": 12976
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7648576342442195e-06,
+      "loss": 1.241,
+      "step": 12977
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.764762086757119e-06,
+      "loss": 0.8499,
+      "step": 12978
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7646665392700176e-06,
+      "loss": 1.0777,
+      "step": 12979
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7645709917829166e-06,
+      "loss": 0.6951,
+      "step": 12980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7644754442958152e-06,
+      "loss": 0.618,
+      "step": 12981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7643798968087143e-06,
+      "loss": 0.7963,
+      "step": 12982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.764284349321613e-06,
+      "loss": 0.9403,
+      "step": 12983
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.764188801834512e-06,
+      "loss": 1.0994,
+      "step": 12984
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7640932543474114e-06,
+      "loss": 0.9947,
+      "step": 12985
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.76399770686031e-06,
+      "loss": 0.9886,
+      "step": 12986
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.763902159373209e-06,
+      "loss": 0.9973,
+      "step": 12987
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7638066118861077e-06,
+      "loss": 1.0492,
+      "step": 12988
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7637110643990067e-06,
+      "loss": 1.6082,
+      "step": 12989
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7636155169119053e-06,
+      "loss": 1.2534,
+      "step": 12990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7635199694248043e-06,
+      "loss": 0.8441,
+      "step": 12991
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.763424421937703e-06,
+      "loss": 0.8026,
+      "step": 12992
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7633288744506024e-06,
+      "loss": 1.3228,
+      "step": 12993
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7632333269635015e-06,
+      "loss": 1.2445,
+      "step": 12994
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7631377794764e-06,
+      "loss": 0.9696,
+      "step": 12995
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.763042231989299e-06,
+      "loss": 1.1821,
+      "step": 12996
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7629466845021977e-06,
+      "loss": 1.2847,
+      "step": 12997
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7628511370150968e-06,
+      "loss": 0.7741,
+      "step": 12998
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7627555895279954e-06,
+      "loss": 1.1847,
+      "step": 12999
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.762660042040895e-06,
+      "loss": 0.6617,
+      "step": 13000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.762564494553794e-06,
+      "loss": 1.0757,
+      "step": 13001
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7624689470666925e-06,
+      "loss": 1.1147,
+      "step": 13002
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7623733995795915e-06,
+      "loss": 0.8257,
+      "step": 13003
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.76227785209249e-06,
+      "loss": 1.1167,
+      "step": 13004
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.762182304605389e-06,
+      "loss": 1.0524,
+      "step": 13005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.762086757118288e-06,
+      "loss": 0.9141,
+      "step": 13006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7619912096311873e-06,
+      "loss": 0.7845,
+      "step": 13007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761895662144086e-06,
+      "loss": 0.7277,
+      "step": 13008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761800114656985e-06,
+      "loss": 1.0316,
+      "step": 13009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7617045671698835e-06,
+      "loss": 0.9013,
+      "step": 13010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7616090196827826e-06,
+      "loss": 1.0222,
+      "step": 13011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7615134721956816e-06,
+      "loss": 1.3363,
+      "step": 13012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7614179247085802e-06,
+      "loss": 0.8919,
+      "step": 13013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7613223772214797e-06,
+      "loss": 0.7838,
+      "step": 13014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7612268297343783e-06,
+      "loss": 1.2462,
+      "step": 13015
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7611312822472773e-06,
+      "loss": 0.7989,
+      "step": 13016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.761035734760176e-06,
+      "loss": 1.03,
+      "step": 13017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.760940187273075e-06,
+      "loss": 1.1399,
+      "step": 13018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7608446397859736e-06,
+      "loss": 0.8317,
+      "step": 13019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7607490922988727e-06,
+      "loss": 1.1993,
+      "step": 13020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.760653544811772e-06,
+      "loss": 1.0529,
+      "step": 13021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7605579973246707e-06,
+      "loss": 0.9179,
+      "step": 13022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7604624498375698e-06,
+      "loss": 0.8135,
+      "step": 13023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7603669023504684e-06,
+      "loss": 1.0047,
+      "step": 13024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7602713548633674e-06,
+      "loss": 1.1017,
+      "step": 13025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.760175807376266e-06,
+      "loss": 0.9115,
+      "step": 13026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.760080259889165e-06,
+      "loss": 1.0729,
+      "step": 13027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7599847124020637e-06,
+      "loss": 1.1137,
+      "step": 13028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.759889164914963e-06,
+      "loss": 0.6546,
+      "step": 13029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.759793617427862e-06,
+      "loss": 0.7795,
+      "step": 13030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.759698069940761e-06,
+      "loss": 1.0319,
+      "step": 13031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.75960252245366e-06,
+      "loss": 1.246,
+      "step": 13032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7595069749665585e-06,
+      "loss": 1.0194,
+      "step": 13033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7594114274794575e-06,
+      "loss": 0.6673,
+      "step": 13034
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.759315879992356e-06,
+      "loss": 1.3303,
+      "step": 13035
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7592203325052556e-06,
+      "loss": 0.7796,
+      "step": 13036
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7591247850181546e-06,
+      "loss": 1.5007,
+      "step": 13037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7590292375310532e-06,
+      "loss": 1.1993,
+      "step": 13038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7589336900439523e-06,
+      "loss": 0.9874,
+      "step": 13039
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.758838142556851e-06,
+      "loss": 0.9413,
+      "step": 13040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.75874259506975e-06,
+      "loss": 0.8723,
+      "step": 13041
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7586470475826485e-06,
+      "loss": 1.3326,
+      "step": 13042
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.758551500095548e-06,
+      "loss": 0.9396,
+      "step": 13043
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.758455952608447e-06,
+      "loss": 1.097,
+      "step": 13044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7583604051213457e-06,
+      "loss": 0.7106,
+      "step": 13045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7582648576342447e-06,
+      "loss": 1.2679,
+      "step": 13046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7581693101471433e-06,
+      "loss": 1.1679,
+      "step": 13047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7580737626600423e-06,
+      "loss": 0.675,
+      "step": 13048
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757978215172941e-06,
+      "loss": 0.6125,
+      "step": 13049
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7578826676858404e-06,
+      "loss": 0.8543,
+      "step": 13050
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7577871201987395e-06,
+      "loss": 0.9542,
+      "step": 13051
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757691572711638e-06,
+      "loss": 0.9811,
+      "step": 13052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757596025224537e-06,
+      "loss": 1.2797,
+      "step": 13053
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7575004777374357e-06,
+      "loss": 1.3928,
+      "step": 13054
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7574049302503348e-06,
+      "loss": 0.7842,
+      "step": 13055
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7573093827632334e-06,
+      "loss": 0.9019,
+      "step": 13056
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757213835276133e-06,
+      "loss": 1.2327,
+      "step": 13057
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7571182877890315e-06,
+      "loss": 0.9108,
+      "step": 13058
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7570227403019305e-06,
+      "loss": 0.9422,
+      "step": 13059
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756927192814829e-06,
+      "loss": 1.2669,
+      "step": 13060
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756831645327728e-06,
+      "loss": 0.9053,
+      "step": 13061
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756736097840627e-06,
+      "loss": 1.2001,
+      "step": 13062
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756640550353526e-06,
+      "loss": 1.0196,
+      "step": 13063
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756545002866425e-06,
+      "loss": 1.1103,
+      "step": 13064
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756449455379324e-06,
+      "loss": 1.1003,
+      "step": 13065
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756353907892223e-06,
+      "loss": 1.1117,
+      "step": 13066
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7562583604051215e-06,
+      "loss": 1.0708,
+      "step": 13067
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7561628129180206e-06,
+      "loss": 0.7945,
+      "step": 13068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.756067265430919e-06,
+      "loss": 1.2716,
+      "step": 13069
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7559717179438182e-06,
+      "loss": 1.1752,
+      "step": 13070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.755876170456717e-06,
+      "loss": 1.1025,
+      "step": 13071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7557806229696163e-06,
+      "loss": 1.1293,
+      "step": 13072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7556850754825153e-06,
+      "loss": 1.1593,
+      "step": 13073
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.755589527995414e-06,
+      "loss": 0.7397,
+      "step": 13074
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.755493980508313e-06,
+      "loss": 1.0775,
+      "step": 13075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7553984330212116e-06,
+      "loss": 1.0617,
+      "step": 13076
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7553028855341107e-06,
+      "loss": 0.9486,
+      "step": 13077
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7552073380470093e-06,
+      "loss": 1.2448,
+      "step": 13078
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7551117905599087e-06,
+      "loss": 1.1268,
+      "step": 13079
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7550162430728078e-06,
+      "loss": 0.8925,
+      "step": 13080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7549206955857064e-06,
+      "loss": 0.7266,
+      "step": 13081
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7548251480986054e-06,
+      "loss": 1.0119,
+      "step": 13082
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.754729600611504e-06,
+      "loss": 0.9888,
+      "step": 13083
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.754634053124403e-06,
+      "loss": 0.8435,
+      "step": 13084
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7545385056373017e-06,
+      "loss": 1.4458,
+      "step": 13085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.754442958150201e-06,
+      "loss": 0.8615,
+      "step": 13086
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7543474106631e-06,
+      "loss": 0.9771,
+      "step": 13087
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.754251863175999e-06,
+      "loss": 1.1546,
+      "step": 13088
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.754156315688898e-06,
+      "loss": 1.5231,
+      "step": 13089
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7540607682017965e-06,
+      "loss": 0.9526,
+      "step": 13090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7539652207146955e-06,
+      "loss": 0.6355,
+      "step": 13091
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.753869673227594e-06,
+      "loss": 1.1575,
+      "step": 13092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7537741257404936e-06,
+      "loss": 1.0344,
+      "step": 13093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7536785782533926e-06,
+      "loss": 1.1341,
+      "step": 13094
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7535830307662912e-06,
+      "loss": 1.1913,
+      "step": 13095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7534874832791903e-06,
+      "loss": 1.0648,
+      "step": 13096
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.753391935792089e-06,
+      "loss": 0.5796,
+      "step": 13097
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.753296388304988e-06,
+      "loss": 0.7103,
+      "step": 13098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7532008408178865e-06,
+      "loss": 1.3953,
+      "step": 13099
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7531052933307856e-06,
+      "loss": 0.7222,
+      "step": 13100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7530097458436846e-06,
+      "loss": 0.8134,
+      "step": 13101
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7529141983565837e-06,
+      "loss": 1.0515,
+      "step": 13102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7528186508694827e-06,
+      "loss": 0.9865,
+      "step": 13103
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7527231033823813e-06,
+      "loss": 1.4285,
+      "step": 13104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7526275558952803e-06,
+      "loss": 0.7887,
+      "step": 13105
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.752532008408179e-06,
+      "loss": 0.7691,
+      "step": 13106
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.752436460921078e-06,
+      "loss": 1.1643,
+      "step": 13107
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.752340913433977e-06,
+      "loss": 1.0934,
+      "step": 13108
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.752245365946876e-06,
+      "loss": 0.7512,
+      "step": 13109
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7521498184597747e-06,
+      "loss": 1.1684,
+      "step": 13110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7520542709726737e-06,
+      "loss": 0.914,
+      "step": 13111
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7519587234855723e-06,
+      "loss": 0.7256,
+      "step": 13112
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7518631759984714e-06,
+      "loss": 1.0456,
+      "step": 13113
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7517676285113704e-06,
+      "loss": 0.8483,
+      "step": 13114
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7516720810242695e-06,
+      "loss": 0.8808,
+      "step": 13115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7515765335371685e-06,
+      "loss": 0.8646,
+      "step": 13116
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751480986050067e-06,
+      "loss": 1.1685,
+      "step": 13117
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751385438562966e-06,
+      "loss": 1.0718,
+      "step": 13118
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7512898910758648e-06,
+      "loss": 0.994,
+      "step": 13119
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.751194343588764e-06,
+      "loss": 1.3808,
+      "step": 13120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7510987961016624e-06,
+      "loss": 0.8634,
+      "step": 13121
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.751003248614562e-06,
+      "loss": 1.1795,
+      "step": 13122
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.750907701127461e-06,
+      "loss": 0.701,
+      "step": 13123
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7508121536403595e-06,
+      "loss": 0.612,
+      "step": 13124
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7507166061532586e-06,
+      "loss": 1.3205,
+      "step": 13125
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.750621058666157e-06,
+      "loss": 0.9258,
+      "step": 13126
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7505255111790562e-06,
+      "loss": 0.8797,
+      "step": 13127
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.750429963691955e-06,
+      "loss": 0.7838,
+      "step": 13128
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7503344162048543e-06,
+      "loss": 1.0094,
+      "step": 13129
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7502388687177533e-06,
+      "loss": 0.9759,
+      "step": 13130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.750143321230652e-06,
+      "loss": 1.0166,
+      "step": 13131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.750047773743551e-06,
+      "loss": 0.9846,
+      "step": 13132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7499522262564496e-06,
+      "loss": 1.1396,
+      "step": 13133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7498566787693487e-06,
+      "loss": 1.1423,
+      "step": 13134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7497611312822473e-06,
+      "loss": 1.1365,
+      "step": 13135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7496655837951463e-06,
+      "loss": 0.7706,
+      "step": 13136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7495700363080458e-06,
+      "loss": 0.6294,
+      "step": 13137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7494744888209444e-06,
+      "loss": 1.0236,
+      "step": 13138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7493789413338434e-06,
+      "loss": 0.9503,
+      "step": 13139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.749283393846742e-06,
+      "loss": 0.9012,
+      "step": 13140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.749187846359641e-06,
+      "loss": 0.6595,
+      "step": 13141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7490922988725397e-06,
+      "loss": 0.6604,
+      "step": 13142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7489967513854387e-06,
+      "loss": 0.9553,
+      "step": 13143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.748901203898338e-06,
+      "loss": 0.7866,
+      "step": 13144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.748805656411237e-06,
+      "loss": 0.9886,
+      "step": 13145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.748710108924136e-06,
+      "loss": 0.7594,
+      "step": 13146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7486145614370345e-06,
+      "loss": 0.7845,
+      "step": 13147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7485190139499335e-06,
+      "loss": 1.2671,
+      "step": 13148
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.748423466462832e-06,
+      "loss": 0.8564,
+      "step": 13149
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.748327918975731e-06,
+      "loss": 1.2655,
+      "step": 13150
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.74823237148863e-06,
+      "loss": 1.0414,
+      "step": 13151
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7481368240015292e-06,
+      "loss": 1.0488,
+      "step": 13152
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7480412765144283e-06,
+      "loss": 1.0101,
+      "step": 13153
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747945729027327e-06,
+      "loss": 0.8266,
+      "step": 13154
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747850181540226e-06,
+      "loss": 1.0436,
+      "step": 13155
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7477546340531245e-06,
+      "loss": 0.8449,
+      "step": 13156
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7476590865660236e-06,
+      "loss": 0.844,
+      "step": 13157
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7475635390789226e-06,
+      "loss": 0.7858,
+      "step": 13158
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7474679915918217e-06,
+      "loss": 0.9445,
+      "step": 13159
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7473724441047203e-06,
+      "loss": 1.4629,
+      "step": 13160
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7472768966176193e-06,
+      "loss": 0.883,
+      "step": 13161
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747181349130518e-06,
+      "loss": 1.2235,
+      "step": 13162
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747085801643417e-06,
+      "loss": 1.2025,
+      "step": 13163
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746990254156316e-06,
+      "loss": 0.9039,
+      "step": 13164
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746894706669215e-06,
+      "loss": 1.1914,
+      "step": 13165
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746799159182114e-06,
+      "loss": 0.8547,
+      "step": 13166
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7467036116950127e-06,
+      "loss": 1.0733,
+      "step": 13167
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7466080642079117e-06,
+      "loss": 0.9255,
+      "step": 13168
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7465125167208103e-06,
+      "loss": 1.2872,
+      "step": 13169
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7464169692337094e-06,
+      "loss": 0.7516,
+      "step": 13170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746321421746608e-06,
+      "loss": 1.2497,
+      "step": 13171
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.746225874259507e-06,
+      "loss": 1.0453,
+      "step": 13172
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7461303267724065e-06,
+      "loss": 0.6674,
+      "step": 13173
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.746034779285305e-06,
+      "loss": 0.6397,
+      "step": 13174
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.745939231798204e-06,
+      "loss": 0.7207,
+      "step": 13175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7458436843111028e-06,
+      "loss": 0.7887,
+      "step": 13176
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.745748136824002e-06,
+      "loss": 1.3043,
+      "step": 13177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7456525893369004e-06,
+      "loss": 1.7673,
+      "step": 13178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7455570418497995e-06,
+      "loss": 0.7015,
+      "step": 13179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.745461494362699e-06,
+      "loss": 1.2288,
+      "step": 13180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7453659468755975e-06,
+      "loss": 1.2642,
+      "step": 13181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7452703993884966e-06,
+      "loss": 1.2842,
+      "step": 13182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.745174851901395e-06,
+      "loss": 0.9644,
+      "step": 13183
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7450793044142942e-06,
+      "loss": 0.7716,
+      "step": 13184
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.744983756927193e-06,
+      "loss": 1.0667,
+      "step": 13185
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.744888209440092e-06,
+      "loss": 1.3035,
+      "step": 13186
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7447926619529913e-06,
+      "loss": 1.299,
+      "step": 13187
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74469711446589e-06,
+      "loss": 0.8913,
+      "step": 13188
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.744601566978789e-06,
+      "loss": 0.6367,
+      "step": 13189
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7445060194916876e-06,
+      "loss": 1.1972,
+      "step": 13190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7444104720045867e-06,
+      "loss": 0.9484,
+      "step": 13191
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7443149245174853e-06,
+      "loss": 0.893,
+      "step": 13192
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7442193770303843e-06,
+      "loss": 1.3399,
+      "step": 13193
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7441238295432838e-06,
+      "loss": 1.5314,
+      "step": 13194
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7440282820561824e-06,
+      "loss": 1.2196,
+      "step": 13195
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7439327345690814e-06,
+      "loss": 1.0365,
+      "step": 13196
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74383718708198e-06,
+      "loss": 1.099,
+      "step": 13197
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.743741639594879e-06,
+      "loss": 1.3611,
+      "step": 13198
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7436460921077777e-06,
+      "loss": 1.0639,
+      "step": 13199
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7435505446206767e-06,
+      "loss": 0.932,
+      "step": 13200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7434549971335758e-06,
+      "loss": 0.6658,
+      "step": 13201
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.743359449646475e-06,
+      "loss": 1.1197,
+      "step": 13202
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7432639021593734e-06,
+      "loss": 1.1457,
+      "step": 13203
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7431683546722725e-06,
+      "loss": 1.3432,
+      "step": 13204
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7430728071851715e-06,
+      "loss": 0.4988,
+      "step": 13205
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74297725969807e-06,
+      "loss": 0.8812,
+      "step": 13206
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.742881712210969e-06,
+      "loss": 1.0222,
+      "step": 13207
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7427861647238678e-06,
+      "loss": 0.8628,
+      "step": 13208
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7426906172367672e-06,
+      "loss": 1.3241,
+      "step": 13209
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.742595069749666e-06,
+      "loss": 1.4002,
+      "step": 13210
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.742499522262565e-06,
+      "loss": 1.3241,
+      "step": 13211
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7424039747754635e-06,
+      "loss": 1.1221,
+      "step": 13212
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7423084272883625e-06,
+      "loss": 1.0476,
+      "step": 13213
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.742212879801261e-06,
+      "loss": 0.9137,
+      "step": 13214
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.74211733231416e-06,
+      "loss": 0.9552,
+      "step": 13215
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7420217848270597e-06,
+      "loss": 0.9481,
+      "step": 13216
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7419262373399583e-06,
+      "loss": 0.8589,
+      "step": 13217
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7418306898528573e-06,
+      "loss": 1.0536,
+      "step": 13218
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741735142365756e-06,
+      "loss": 1.025,
+      "step": 13219
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741639594878655e-06,
+      "loss": 1.2134,
+      "step": 13220
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7415440473915536e-06,
+      "loss": 0.6723,
+      "step": 13221
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7414484999044526e-06,
+      "loss": 0.8749,
+      "step": 13222
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741352952417352e-06,
+      "loss": 1.2286,
+      "step": 13223
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7412574049302507e-06,
+      "loss": 0.565,
+      "step": 13224
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7411618574431497e-06,
+      "loss": 1.3715,
+      "step": 13225
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7410663099560483e-06,
+      "loss": 1.05,
+      "step": 13226
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7409707624689474e-06,
+      "loss": 1.1124,
+      "step": 13227
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.740875214981846e-06,
+      "loss": 0.7732,
+      "step": 13228
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.740779667494745e-06,
+      "loss": 1.2356,
+      "step": 13229
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7406841200076445e-06,
+      "loss": 0.841,
+      "step": 13230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.740588572520543e-06,
+      "loss": 0.8013,
+      "step": 13231
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.740493025033442e-06,
+      "loss": 1.4,
+      "step": 13232
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7403974775463408e-06,
+      "loss": 1.215,
+      "step": 13233
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.74030193005924e-06,
+      "loss": 1.0073,
+      "step": 13234
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7402063825721384e-06,
+      "loss": 0.8689,
+      "step": 13235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7401108350850375e-06,
+      "loss": 1.1453,
+      "step": 13236
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.740015287597937e-06,
+      "loss": 0.7413,
+      "step": 13237
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7399197401108355e-06,
+      "loss": 0.7272,
+      "step": 13238
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7398241926237346e-06,
+      "loss": 0.826,
+      "step": 13239
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.739728645136633e-06,
+      "loss": 0.9662,
+      "step": 13240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7396330976495322e-06,
+      "loss": 0.8216,
+      "step": 13241
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.739537550162431e-06,
+      "loss": 0.5225,
+      "step": 13242
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.73944200267533e-06,
+      "loss": 1.0657,
+      "step": 13243
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7393464551882285e-06,
+      "loss": 0.5407,
+      "step": 13244
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.739250907701128e-06,
+      "loss": 1.3814,
+      "step": 13245
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.739155360214027e-06,
+      "loss": 1.2098,
+      "step": 13246
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7390598127269256e-06,
+      "loss": 1.1417,
+      "step": 13247
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7389642652398247e-06,
+      "loss": 0.9899,
+      "step": 13248
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7388687177527233e-06,
+      "loss": 0.9884,
+      "step": 13249
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7387731702656223e-06,
+      "loss": 0.8713,
+      "step": 13250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.738677622778521e-06,
+      "loss": 0.6625,
+      "step": 13251
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7385820752914204e-06,
+      "loss": 1.0348,
+      "step": 13252
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.738486527804319e-06,
+      "loss": 1.3184,
+      "step": 13253
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.738390980317218e-06,
+      "loss": 1.469,
+      "step": 13254
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.738295432830117e-06,
+      "loss": 1.2513,
+      "step": 13255
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7381998853430157e-06,
+      "loss": 1.2429,
+      "step": 13256
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7381043378559147e-06,
+      "loss": 1.1699,
+      "step": 13257
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7380087903688133e-06,
+      "loss": 1.0279,
+      "step": 13258
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737913242881713e-06,
+      "loss": 1.4777,
+      "step": 13259
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7378176953946114e-06,
+      "loss": 0.865,
+      "step": 13260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7377221479075105e-06,
+      "loss": 1.0839,
+      "step": 13261
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737626600420409e-06,
+      "loss": 1.4396,
+      "step": 13262
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737531052933308e-06,
+      "loss": 0.9189,
+      "step": 13263
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7374355054462067e-06,
+      "loss": 0.7724,
+      "step": 13264
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7373399579591058e-06,
+      "loss": 0.3832,
+      "step": 13265
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7372444104720052e-06,
+      "loss": 1.1233,
+      "step": 13266
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737148862984904e-06,
+      "loss": 1.1525,
+      "step": 13267
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737053315497803e-06,
+      "loss": 1.09,
+      "step": 13268
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7369577680107015e-06,
+      "loss": 1.1696,
+      "step": 13269
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7368622205236005e-06,
+      "loss": 0.9978,
+      "step": 13270
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.736766673036499e-06,
+      "loss": 1.2158,
+      "step": 13271
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.736671125549398e-06,
+      "loss": 1.3353,
+      "step": 13272
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7365755780622977e-06,
+      "loss": 1.0724,
+      "step": 13273
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7364800305751963e-06,
+      "loss": 1.3928,
+      "step": 13274
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7363844830880953e-06,
+      "loss": 1.1128,
+      "step": 13275
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.736288935600994e-06,
+      "loss": 1.1277,
+      "step": 13276
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.736193388113893e-06,
+      "loss": 1.0886,
+      "step": 13277
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7360978406267916e-06,
+      "loss": 0.8236,
+      "step": 13278
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7360022931396906e-06,
+      "loss": 0.9974,
+      "step": 13279
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7359067456525892e-06,
+      "loss": 0.8061,
+      "step": 13280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7358111981654887e-06,
+      "loss": 0.8115,
+      "step": 13281
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7357156506783877e-06,
+      "loss": 0.8592,
+      "step": 13282
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7356201031912863e-06,
+      "loss": 0.9001,
+      "step": 13283
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7355245557041854e-06,
+      "loss": 1.4854,
+      "step": 13284
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.735429008217084e-06,
+      "loss": 1.0738,
+      "step": 13285
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.735333460729983e-06,
+      "loss": 0.9194,
+      "step": 13286
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7352379132428816e-06,
+      "loss": 0.9345,
+      "step": 13287
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.735142365755781e-06,
+      "loss": 0.9557,
+      "step": 13288
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.73504681826868e-06,
+      "loss": 0.9717,
+      "step": 13289
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7349512707815788e-06,
+      "loss": 0.9702,
+      "step": 13290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.734855723294478e-06,
+      "loss": 1.2477,
+      "step": 13291
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7347601758073764e-06,
+      "loss": 0.6612,
+      "step": 13292
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7346646283202755e-06,
+      "loss": 0.9445,
+      "step": 13293
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.734569080833174e-06,
+      "loss": 1.1094,
+      "step": 13294
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7344735333460735e-06,
+      "loss": 0.89,
+      "step": 13295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7343779858589726e-06,
+      "loss": 1.1964,
+      "step": 13296
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.734282438371871e-06,
+      "loss": 0.9499,
+      "step": 13297
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7341868908847702e-06,
+      "loss": 1.0364,
+      "step": 13298
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.734091343397669e-06,
+      "loss": 1.0494,
+      "step": 13299
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.733995795910568e-06,
+      "loss": 1.1088,
+      "step": 13300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7339002484234665e-06,
+      "loss": 1.0802,
+      "step": 13301
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.733804700936366e-06,
+      "loss": 1.2755,
+      "step": 13302
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7337091534492646e-06,
+      "loss": 0.9631,
+      "step": 13303
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7336136059621636e-06,
+      "loss": 0.8863,
+      "step": 13304
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7335180584750622e-06,
+      "loss": 0.8323,
+      "step": 13305
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7334225109879613e-06,
+      "loss": 1.355,
+      "step": 13306
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7333269635008603e-06,
+      "loss": 0.7402,
+      "step": 13307
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.733231416013759e-06,
+      "loss": 0.7853,
+      "step": 13308
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7331358685266584e-06,
+      "loss": 1.0431,
+      "step": 13309
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.733040321039557e-06,
+      "loss": 1.5747,
+      "step": 13310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.732944773552456e-06,
+      "loss": 0.5421,
+      "step": 13311
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7328492260653546e-06,
+      "loss": 0.6218,
+      "step": 13312
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7327536785782537e-06,
+      "loss": 0.9377,
+      "step": 13313
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7326581310911523e-06,
+      "loss": 1.0122,
+      "step": 13314
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7325625836040513e-06,
+      "loss": 1.5627,
+      "step": 13315
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.73246703611695e-06,
+      "loss": 0.8045,
+      "step": 13316
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7323714886298494e-06,
+      "loss": 1.1807,
+      "step": 13317
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7322759411427485e-06,
+      "loss": 1.0322,
+      "step": 13318
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.732180393655647e-06,
+      "loss": 1.3588,
+      "step": 13319
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.732084846168546e-06,
+      "loss": 1.0826,
+      "step": 13320
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7319892986814447e-06,
+      "loss": 1.0716,
+      "step": 13321
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7318937511943438e-06,
+      "loss": 1.0322,
+      "step": 13322
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7317982037072424e-06,
+      "loss": 1.3807,
+      "step": 13323
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731702656220142e-06,
+      "loss": 0.6899,
+      "step": 13324
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731607108733041e-06,
+      "loss": 1.2271,
+      "step": 13325
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7315115612459395e-06,
+      "loss": 0.6337,
+      "step": 13326
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7314160137588385e-06,
+      "loss": 1.1747,
+      "step": 13327
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731320466271737e-06,
+      "loss": 0.646,
+      "step": 13328
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731224918784636e-06,
+      "loss": 1.1241,
+      "step": 13329
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.731129371297535e-06,
+      "loss": 1.1478,
+      "step": 13330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7310338238104343e-06,
+      "loss": 0.5249,
+      "step": 13331
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7309382763233333e-06,
+      "loss": 0.7607,
+      "step": 13332
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.730842728836232e-06,
+      "loss": 1.3756,
+      "step": 13333
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.730747181349131e-06,
+      "loss": 1.1694,
+      "step": 13334
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7306516338620296e-06,
+      "loss": 0.5976,
+      "step": 13335
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7305560863749286e-06,
+      "loss": 0.8406,
+      "step": 13336
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7304605388878272e-06,
+      "loss": 1.0158,
+      "step": 13337
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7303649914007267e-06,
+      "loss": 0.9912,
+      "step": 13338
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7302694439136257e-06,
+      "loss": 1.0206,
+      "step": 13339
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7301738964265243e-06,
+      "loss": 0.867,
+      "step": 13340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7300783489394234e-06,
+      "loss": 0.9772,
+      "step": 13341
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729982801452322e-06,
+      "loss": 0.9709,
+      "step": 13342
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729887253965221e-06,
+      "loss": 1.0079,
+      "step": 13343
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7297917064781196e-06,
+      "loss": 0.7945,
+      "step": 13344
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729696158991019e-06,
+      "loss": 0.923,
+      "step": 13345
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729600611503918e-06,
+      "loss": 0.4865,
+      "step": 13346
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7295050640168168e-06,
+      "loss": 1.0654,
+      "step": 13347
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729409516529716e-06,
+      "loss": 0.7807,
+      "step": 13348
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7293139690426144e-06,
+      "loss": 1.2984,
+      "step": 13349
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7292184215555135e-06,
+      "loss": 1.0978,
+      "step": 13350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729122874068412e-06,
+      "loss": 0.8674,
+      "step": 13351
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.729027326581311e-06,
+      "loss": 0.8967,
+      "step": 13352
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.72893177909421e-06,
+      "loss": 1.019,
+      "step": 13353
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.728836231607109e-06,
+      "loss": 1.2135,
+      "step": 13354
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.728740684120008e-06,
+      "loss": 1.1786,
+      "step": 13355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.728645136632907e-06,
+      "loss": 0.6896,
+      "step": 13356
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.728549589145806e-06,
+      "loss": 1.0548,
+      "step": 13357
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7284540416587045e-06,
+      "loss": 0.8861,
+      "step": 13358
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7283584941716035e-06,
+      "loss": 0.9357,
+      "step": 13359
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7282629466845026e-06,
+      "loss": 0.8461,
+      "step": 13360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7281673991974016e-06,
+      "loss": 0.9322,
+      "step": 13361
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7280718517103002e-06,
+      "loss": 0.9009,
+      "step": 13362
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7279763042231993e-06,
+      "loss": 0.6888,
+      "step": 13363
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727880756736098e-06,
+      "loss": 0.8189,
+      "step": 13364
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727785209248997e-06,
+      "loss": 0.8089,
+      "step": 13365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7276896617618955e-06,
+      "loss": 1.2033,
+      "step": 13366
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727594114274795e-06,
+      "loss": 0.8287,
+      "step": 13367
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727498566787694e-06,
+      "loss": 1.1977,
+      "step": 13368
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7274030193005926e-06,
+      "loss": 1.0111,
+      "step": 13369
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7273074718134917e-06,
+      "loss": 0.8561,
+      "step": 13370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7272119243263903e-06,
+      "loss": 1.6036,
+      "step": 13371
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7271163768392893e-06,
+      "loss": 0.8644,
+      "step": 13372
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727020829352188e-06,
+      "loss": 1.1123,
+      "step": 13373
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7269252818650874e-06,
+      "loss": 1.1148,
+      "step": 13374
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7268297343779865e-06,
+      "loss": 1.2477,
+      "step": 13375
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726734186890885e-06,
+      "loss": 0.972,
+      "step": 13376
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726638639403784e-06,
+      "loss": 0.6941,
+      "step": 13377
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7265430919166827e-06,
+      "loss": 1.076,
+      "step": 13378
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7264475444295818e-06,
+      "loss": 1.2452,
+      "step": 13379
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7263519969424804e-06,
+      "loss": 1.1211,
+      "step": 13380
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.72625644945538e-06,
+      "loss": 0.7282,
+      "step": 13381
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726160901968279e-06,
+      "loss": 0.9558,
+      "step": 13382
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7260653544811775e-06,
+      "loss": 0.9226,
+      "step": 13383
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7259698069940765e-06,
+      "loss": 0.9397,
+      "step": 13384
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.725874259506975e-06,
+      "loss": 0.7181,
+      "step": 13385
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.725778712019874e-06,
+      "loss": 0.8202,
+      "step": 13386
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.725683164532773e-06,
+      "loss": 0.6783,
+      "step": 13387
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.725587617045672e-06,
+      "loss": 0.6048,
+      "step": 13388
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7254920695585713e-06,
+      "loss": 1.2192,
+      "step": 13389
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72539652207147e-06,
+      "loss": 0.7905,
+      "step": 13390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.725300974584369e-06,
+      "loss": 0.8078,
+      "step": 13391
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7252054270972676e-06,
+      "loss": 0.8332,
+      "step": 13392
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7251098796101666e-06,
+      "loss": 0.8174,
+      "step": 13393
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7250143321230652e-06,
+      "loss": 0.7266,
+      "step": 13394
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7249187846359643e-06,
+      "loss": 0.8417,
+      "step": 13395
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7248232371488633e-06,
+      "loss": 0.9261,
+      "step": 13396
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7247276896617623e-06,
+      "loss": 1.0788,
+      "step": 13397
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7246321421746614e-06,
+      "loss": 1.3588,
+      "step": 13398
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72453659468756e-06,
+      "loss": 1.262,
+      "step": 13399
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.724441047200459e-06,
+      "loss": 0.9286,
+      "step": 13400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7243454997133576e-06,
+      "loss": 0.8366,
+      "step": 13401
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7242499522262567e-06,
+      "loss": 1.2053,
+      "step": 13402
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7241544047391557e-06,
+      "loss": 0.7103,
+      "step": 13403
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7240588572520548e-06,
+      "loss": 1.1081,
+      "step": 13404
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7239633097649534e-06,
+      "loss": 0.9593,
+      "step": 13405
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7238677622778524e-06,
+      "loss": 0.9421,
+      "step": 13406
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723772214790751e-06,
+      "loss": 0.8134,
+      "step": 13407
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.72367666730365e-06,
+      "loss": 1.3932,
+      "step": 13408
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723581119816549e-06,
+      "loss": 1.5126,
+      "step": 13409
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723485572329448e-06,
+      "loss": 0.8524,
+      "step": 13410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723390024842347e-06,
+      "loss": 1.5279,
+      "step": 13411
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723294477355246e-06,
+      "loss": 1.0718,
+      "step": 13412
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.723198929868145e-06,
+      "loss": 1.0408,
+      "step": 13413
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7231033823810435e-06,
+      "loss": 0.8698,
+      "step": 13414
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7230078348939425e-06,
+      "loss": 0.7585,
+      "step": 13415
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722912287406841e-06,
+      "loss": 0.8191,
+      "step": 13416
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7228167399197406e-06,
+      "loss": 1.1578,
+      "step": 13417
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7227211924326396e-06,
+      "loss": 1.1129,
+      "step": 13418
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7226256449455382e-06,
+      "loss": 0.8967,
+      "step": 13419
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7225300974584373e-06,
+      "loss": 1.0403,
+      "step": 13420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722434549971336e-06,
+      "loss": 1.1329,
+      "step": 13421
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722339002484235e-06,
+      "loss": 1.1279,
+      "step": 13422
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7222434549971335e-06,
+      "loss": 1.4973,
+      "step": 13423
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7221479075100326e-06,
+      "loss": 1.0765,
+      "step": 13424
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722052360022932e-06,
+      "loss": 1.2735,
+      "step": 13425
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7219568125358306e-06,
+      "loss": 0.8533,
+      "step": 13426
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7218612650487297e-06,
+      "loss": 0.7649,
+      "step": 13427
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7217657175616283e-06,
+      "loss": 1.2863,
+      "step": 13428
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7216701700745273e-06,
+      "loss": 1.6305,
+      "step": 13429
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721574622587426e-06,
+      "loss": 0.91,
+      "step": 13430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721479075100325e-06,
+      "loss": 1.1601,
+      "step": 13431
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7213835276132245e-06,
+      "loss": 1.2801,
+      "step": 13432
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721287980126123e-06,
+      "loss": 0.9702,
+      "step": 13433
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721192432639022e-06,
+      "loss": 1.1929,
+      "step": 13434
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7210968851519207e-06,
+      "loss": 0.9886,
+      "step": 13435
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7210013376648198e-06,
+      "loss": 1.3028,
+      "step": 13436
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7209057901777184e-06,
+      "loss": 0.9887,
+      "step": 13437
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7208102426906174e-06,
+      "loss": 1.4459,
+      "step": 13438
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.720714695203517e-06,
+      "loss": 0.7733,
+      "step": 13439
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7206191477164155e-06,
+      "loss": 0.8371,
+      "step": 13440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7205236002293145e-06,
+      "loss": 1.2471,
+      "step": 13441
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.720428052742213e-06,
+      "loss": 1.1272,
+      "step": 13442
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.720332505255112e-06,
+      "loss": 1.4596,
+      "step": 13443
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.720236957768011e-06,
+      "loss": 1.3762,
+      "step": 13444
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.72014141028091e-06,
+      "loss": 0.6863,
+      "step": 13445
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.720045862793809e-06,
+      "loss": 0.7889,
+      "step": 13446
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.719950315306708e-06,
+      "loss": 0.7691,
+      "step": 13447
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.719854767819607e-06,
+      "loss": 1.1018,
+      "step": 13448
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7197592203325056e-06,
+      "loss": 0.5515,
+      "step": 13449
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7196636728454046e-06,
+      "loss": 0.9931,
+      "step": 13450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7195681253583032e-06,
+      "loss": 0.8505,
+      "step": 13451
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7194725778712023e-06,
+      "loss": 1.1262,
+      "step": 13452
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7193770303841013e-06,
+      "loss": 0.728,
+      "step": 13453
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7192814828970003e-06,
+      "loss": 1.1573,
+      "step": 13454
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.719185935409899e-06,
+      "loss": 1.32,
+      "step": 13455
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.719090387922798e-06,
+      "loss": 0.7293,
+      "step": 13456
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7189948404356966e-06,
+      "loss": 1.0968,
+      "step": 13457
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7188992929485956e-06,
+      "loss": 0.8403,
+      "step": 13458
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7188037454614947e-06,
+      "loss": 1.0344,
+      "step": 13459
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7187081979743933e-06,
+      "loss": 1.354,
+      "step": 13460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7186126504872928e-06,
+      "loss": 1.0035,
+      "step": 13461
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7185171030001914e-06,
+      "loss": 0.9335,
+      "step": 13462
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7184215555130904e-06,
+      "loss": 0.7315,
+      "step": 13463
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.718326008025989e-06,
+      "loss": 0.8957,
+      "step": 13464
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.718230460538888e-06,
+      "loss": 0.9313,
+      "step": 13465
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7181349130517867e-06,
+      "loss": 1.1887,
+      "step": 13466
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7180393655646857e-06,
+      "loss": 1.1523,
+      "step": 13467
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717943818077585e-06,
+      "loss": 0.9002,
+      "step": 13468
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717848270590484e-06,
+      "loss": 0.8174,
+      "step": 13469
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717752723103383e-06,
+      "loss": 0.3917,
+      "step": 13470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7176571756162815e-06,
+      "loss": 1.1106,
+      "step": 13471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7175616281291805e-06,
+      "loss": 0.8193,
+      "step": 13472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717466080642079e-06,
+      "loss": 0.7587,
+      "step": 13473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717370533154978e-06,
+      "loss": 0.7849,
+      "step": 13474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7172749856678776e-06,
+      "loss": 0.8861,
+      "step": 13475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7171794381807762e-06,
+      "loss": 0.8984,
+      "step": 13476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7170838906936753e-06,
+      "loss": 1.0023,
+      "step": 13477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716988343206574e-06,
+      "loss": 1.0075,
+      "step": 13478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716892795719473e-06,
+      "loss": 0.9096,
+      "step": 13479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7167972482323715e-06,
+      "loss": 1.0304,
+      "step": 13480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7167017007452706e-06,
+      "loss": 1.0404,
+      "step": 13481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.71660615325817e-06,
+      "loss": 0.7905,
+      "step": 13482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7165106057710686e-06,
+      "loss": 0.9642,
+      "step": 13483
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7164150582839677e-06,
+      "loss": 0.7245,
+      "step": 13484
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7163195107968663e-06,
+      "loss": 1.4162,
+      "step": 13485
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7162239633097653e-06,
+      "loss": 0.6989,
+      "step": 13486
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716128415822664e-06,
+      "loss": 0.9865,
+      "step": 13487
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.716032868335563e-06,
+      "loss": 0.9011,
+      "step": 13488
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7159373208484625e-06,
+      "loss": 0.8278,
+      "step": 13489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.715841773361361e-06,
+      "loss": 0.8375,
+      "step": 13490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71574622587426e-06,
+      "loss": 1.2344,
+      "step": 13491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7156506783871587e-06,
+      "loss": 1.6744,
+      "step": 13492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7155551309000578e-06,
+      "loss": 1.3841,
+      "step": 13493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7154595834129564e-06,
+      "loss": 1.1471,
+      "step": 13494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7153640359258554e-06,
+      "loss": 1.1674,
+      "step": 13495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.715268488438754e-06,
+      "loss": 0.6233,
+      "step": 13496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7151729409516535e-06,
+      "loss": 0.8719,
+      "step": 13497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.715077393464552e-06,
+      "loss": 1.4752,
+      "step": 13498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.714981845977451e-06,
+      "loss": 0.8868,
+      "step": 13499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71488629849035e-06,
+      "loss": 1.1581,
+      "step": 13500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.714790751003249e-06,
+      "loss": 1.1983,
+      "step": 13501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.714695203516148e-06,
+      "loss": 1.2577,
+      "step": 13502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7145996560290465e-06,
+      "loss": 1.1236,
+      "step": 13503
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.714504108541946e-06,
+      "loss": 1.2273,
+      "step": 13504
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7144085610548445e-06,
+      "loss": 0.5475,
+      "step": 13505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7143130135677436e-06,
+      "loss": 1.1867,
+      "step": 13506
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.714217466080642e-06,
+      "loss": 0.7865,
+      "step": 13507
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7141219185935412e-06,
+      "loss": 0.6471,
+      "step": 13508
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.71402637110644e-06,
+      "loss": 1.6046,
+      "step": 13509
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.713930823619339e-06,
+      "loss": 1.6652,
+      "step": 13510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7138352761322383e-06,
+      "loss": 0.5524,
+      "step": 13511
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.713739728645137e-06,
+      "loss": 0.9738,
+      "step": 13512
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.713644181158036e-06,
+      "loss": 1.3708,
+      "step": 13513
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7135486336709346e-06,
+      "loss": 1.1491,
+      "step": 13514
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7134530861838336e-06,
+      "loss": 1.0139,
+      "step": 13515
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7133575386967323e-06,
+      "loss": 0.8686,
+      "step": 13516
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7132619912096313e-06,
+      "loss": 0.8491,
+      "step": 13517
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7131664437225308e-06,
+      "loss": 1.69,
+      "step": 13518
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7130708962354294e-06,
+      "loss": 0.9665,
+      "step": 13519
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7129753487483284e-06,
+      "loss": 1.4882,
+      "step": 13520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712879801261227e-06,
+      "loss": 1.0,
+      "step": 13521
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712784253774126e-06,
+      "loss": 1.3397,
+      "step": 13522
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7126887062870247e-06,
+      "loss": 0.8352,
+      "step": 13523
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7125931587999237e-06,
+      "loss": 1.2741,
+      "step": 13524
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712497611312823e-06,
+      "loss": 0.8923,
+      "step": 13525
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712402063825722e-06,
+      "loss": 1.1543,
+      "step": 13526
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712306516338621e-06,
+      "loss": 1.0603,
+      "step": 13527
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7122109688515195e-06,
+      "loss": 0.992,
+      "step": 13528
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7121154213644185e-06,
+      "loss": 1.1758,
+      "step": 13529
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712019873877317e-06,
+      "loss": 0.7961,
+      "step": 13530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.711924326390216e-06,
+      "loss": 1.041,
+      "step": 13531
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7118287789031148e-06,
+      "loss": 1.1851,
+      "step": 13532
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7117332314160142e-06,
+      "loss": 0.8829,
+      "step": 13533
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7116376839289133e-06,
+      "loss": 0.8767,
+      "step": 13534
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.711542136441812e-06,
+      "loss": 1.3483,
+      "step": 13535
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.711446588954711e-06,
+      "loss": 0.795,
+      "step": 13536
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7113510414676095e-06,
+      "loss": 1.0892,
+      "step": 13537
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7112554939805086e-06,
+      "loss": 0.5555,
+      "step": 13538
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.711159946493407e-06,
+      "loss": 1.0372,
+      "step": 13539
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7110643990063066e-06,
+      "loss": 0.6313,
+      "step": 13540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7109688515192057e-06,
+      "loss": 1.034,
+      "step": 13541
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7108733040321043e-06,
+      "loss": 0.8205,
+      "step": 13542
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7107777565450033e-06,
+      "loss": 0.7466,
+      "step": 13543
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.710682209057902e-06,
+      "loss": 1.143,
+      "step": 13544
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.710586661570801e-06,
+      "loss": 0.9798,
+      "step": 13545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7104911140836996e-06,
+      "loss": 0.9641,
+      "step": 13546
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.710395566596599e-06,
+      "loss": 1.3691,
+      "step": 13547
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7103000191094977e-06,
+      "loss": 0.8809,
+      "step": 13548
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7102044716223967e-06,
+      "loss": 1.0165,
+      "step": 13549
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7101089241352958e-06,
+      "loss": 1.3384,
+      "step": 13550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7100133766481944e-06,
+      "loss": 1.0385,
+      "step": 13551
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7099178291610934e-06,
+      "loss": 0.7417,
+      "step": 13552
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.709822281673992e-06,
+      "loss": 0.5794,
+      "step": 13553
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7097267341868915e-06,
+      "loss": 0.6141,
+      "step": 13554
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.70963118669979e-06,
+      "loss": 0.9721,
+      "step": 13555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.709535639212689e-06,
+      "loss": 1.0688,
+      "step": 13556
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7094400917255878e-06,
+      "loss": 0.7195,
+      "step": 13557
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.709344544238487e-06,
+      "loss": 0.5147,
+      "step": 13558
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7092489967513854e-06,
+      "loss": 1.3139,
+      "step": 13559
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7091534492642845e-06,
+      "loss": 1.0111,
+      "step": 13560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.709057901777184e-06,
+      "loss": 0.9359,
+      "step": 13561
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7089623542900825e-06,
+      "loss": 1.274,
+      "step": 13562
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7088668068029816e-06,
+      "loss": 1.3456,
+      "step": 13563
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.70877125931588e-06,
+      "loss": 0.9471,
+      "step": 13564
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7086757118287792e-06,
+      "loss": 0.7946,
+      "step": 13565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.708580164341678e-06,
+      "loss": 1.1015,
+      "step": 13566
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.708484616854577e-06,
+      "loss": 1.0979,
+      "step": 13567
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7083890693674755e-06,
+      "loss": 0.6498,
+      "step": 13568
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.708293521880375e-06,
+      "loss": 1.0815,
+      "step": 13569
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.708197974393274e-06,
+      "loss": 0.9244,
+      "step": 13570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7081024269061726e-06,
+      "loss": 1.1372,
+      "step": 13571
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7080068794190716e-06,
+      "loss": 0.6991,
+      "step": 13572
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7079113319319703e-06,
+      "loss": 1.5288,
+      "step": 13573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7078157844448693e-06,
+      "loss": 0.9997,
+      "step": 13574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.707720236957768e-06,
+      "loss": 1.3083,
+      "step": 13575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7076246894706674e-06,
+      "loss": 0.8973,
+      "step": 13576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7075291419835664e-06,
+      "loss": 0.9817,
+      "step": 13577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.707433594496465e-06,
+      "loss": 0.9581,
+      "step": 13578
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.707338047009364e-06,
+      "loss": 0.3906,
+      "step": 13579
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7072424995222627e-06,
+      "loss": 0.7513,
+      "step": 13580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7071469520351617e-06,
+      "loss": 0.8414,
+      "step": 13581
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7070514045480603e-06,
+      "loss": 0.7161,
+      "step": 13582
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.70695585706096e-06,
+      "loss": 1.0968,
+      "step": 13583
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706860309573859e-06,
+      "loss": 1.1157,
+      "step": 13584
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7067647620867575e-06,
+      "loss": 0.9058,
+      "step": 13585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7066692145996565e-06,
+      "loss": 1.3433,
+      "step": 13586
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706573667112555e-06,
+      "loss": 0.5957,
+      "step": 13587
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706478119625454e-06,
+      "loss": 1.1264,
+      "step": 13588
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7063825721383528e-06,
+      "loss": 0.9322,
+      "step": 13589
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7062870246512522e-06,
+      "loss": 0.9343,
+      "step": 13590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7061914771641513e-06,
+      "loss": 0.8483,
+      "step": 13591
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.70609592967705e-06,
+      "loss": 1.3435,
+      "step": 13592
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.706000382189949e-06,
+      "loss": 1.214,
+      "step": 13593
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7059048347028475e-06,
+      "loss": 0.7473,
+      "step": 13594
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7058092872157466e-06,
+      "loss": 0.7534,
+      "step": 13595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.705713739728645e-06,
+      "loss": 1.567,
+      "step": 13596
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7056181922415446e-06,
+      "loss": 0.8955,
+      "step": 13597
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7055226447544433e-06,
+      "loss": 0.824,
+      "step": 13598
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7054270972673423e-06,
+      "loss": 1.0063,
+      "step": 13599
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.705331549780241e-06,
+      "loss": 1.2001,
+      "step": 13600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70523600229314e-06,
+      "loss": 1.4718,
+      "step": 13601
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.705140454806039e-06,
+      "loss": 1.0119,
+      "step": 13602
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7050449073189376e-06,
+      "loss": 0.8566,
+      "step": 13603
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7049493598318366e-06,
+      "loss": 1.2948,
+      "step": 13604
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7048538123447357e-06,
+      "loss": 1.295,
+      "step": 13605
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7047582648576347e-06,
+      "loss": 0.8337,
+      "step": 13606
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7046627173705333e-06,
+      "loss": 0.9247,
+      "step": 13607
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7045671698834324e-06,
+      "loss": 1.5593,
+      "step": 13608
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.704471622396331e-06,
+      "loss": 0.9837,
+      "step": 13609
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.70437607490923e-06,
+      "loss": 0.6712,
+      "step": 13610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7042805274221286e-06,
+      "loss": 0.8879,
+      "step": 13611
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.704184979935028e-06,
+      "loss": 1.4431,
+      "step": 13612
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.704089432447927e-06,
+      "loss": 0.7869,
+      "step": 13613
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7039938849608258e-06,
+      "loss": 1.1451,
+      "step": 13614
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.703898337473725e-06,
+      "loss": 1.3955,
+      "step": 13615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7038027899866234e-06,
+      "loss": 0.9591,
+      "step": 13616
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7037072424995225e-06,
+      "loss": 0.9836,
+      "step": 13617
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.703611695012421e-06,
+      "loss": 0.8033,
+      "step": 13618
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7035161475253205e-06,
+      "loss": 1.7985,
+      "step": 13619
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7034206000382196e-06,
+      "loss": 0.9396,
+      "step": 13620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.703325052551118e-06,
+      "loss": 1.0772,
+      "step": 13621
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7032295050640172e-06,
+      "loss": 1.0847,
+      "step": 13622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.703133957576916e-06,
+      "loss": 1.0381,
+      "step": 13623
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.703038410089815e-06,
+      "loss": 1.0204,
+      "step": 13624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7029428626027135e-06,
+      "loss": 1.4553,
+      "step": 13625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702847315115613e-06,
+      "loss": 1.0264,
+      "step": 13626
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702751767628512e-06,
+      "loss": 1.5389,
+      "step": 13627
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7026562201414106e-06,
+      "loss": 0.9352,
+      "step": 13628
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7025606726543096e-06,
+      "loss": 1.2012,
+      "step": 13629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7024651251672083e-06,
+      "loss": 0.9235,
+      "step": 13630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7023695776801073e-06,
+      "loss": 1.1074,
+      "step": 13631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702274030193006e-06,
+      "loss": 0.995,
+      "step": 13632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7021784827059054e-06,
+      "loss": 1.2102,
+      "step": 13633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7020829352188044e-06,
+      "loss": 0.8696,
+      "step": 13634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701987387731703e-06,
+      "loss": 0.9157,
+      "step": 13635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701891840244602e-06,
+      "loss": 0.6375,
+      "step": 13636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7017962927575007e-06,
+      "loss": 1.4362,
+      "step": 13637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7017007452703997e-06,
+      "loss": 0.7724,
+      "step": 13638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7016051977832983e-06,
+      "loss": 0.7043,
+      "step": 13639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7015096502961974e-06,
+      "loss": 0.7462,
+      "step": 13640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701414102809097e-06,
+      "loss": 0.9564,
+      "step": 13641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7013185553219955e-06,
+      "loss": 0.8788,
+      "step": 13642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7012230078348945e-06,
+      "loss": 1.2249,
+      "step": 13643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701127460347793e-06,
+      "loss": 0.9079,
+      "step": 13644
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.701031912860692e-06,
+      "loss": 1.1807,
+      "step": 13645
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7009363653735908e-06,
+      "loss": 0.7141,
+      "step": 13646
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.70084081788649e-06,
+      "loss": 1.1469,
+      "step": 13647
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.700745270399389e-06,
+      "loss": 1.0142,
+      "step": 13648
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.700649722912288e-06,
+      "loss": 1.7323,
+      "step": 13649
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7005541754251865e-06,
+      "loss": 0.9579,
+      "step": 13650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7004586279380855e-06,
+      "loss": 1.0503,
+      "step": 13651
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7003630804509846e-06,
+      "loss": 0.9021,
+      "step": 13652
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.700267532963883e-06,
+      "loss": 1.0024,
+      "step": 13653
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7001719854767822e-06,
+      "loss": 1.2527,
+      "step": 13654
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7000764379896813e-06,
+      "loss": 1.0864,
+      "step": 13655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6999808905025803e-06,
+      "loss": 0.6985,
+      "step": 13656
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.699885343015479e-06,
+      "loss": 0.8966,
+      "step": 13657
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.699789795528378e-06,
+      "loss": 0.8571,
+      "step": 13658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6996942480412766e-06,
+      "loss": 0.9045,
+      "step": 13659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6995987005541756e-06,
+      "loss": 0.9566,
+      "step": 13660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6995031530670742e-06,
+      "loss": 1.0516,
+      "step": 13661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6994076055799737e-06,
+      "loss": 1.1135,
+      "step": 13662
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6993120580928727e-06,
+      "loss": 0.917,
+      "step": 13663
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6992165106057713e-06,
+      "loss": 0.8253,
+      "step": 13664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6991209631186704e-06,
+      "loss": 0.749,
+      "step": 13665
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.699025415631569e-06,
+      "loss": 1.1809,
+      "step": 13666
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698929868144468e-06,
+      "loss": 1.1055,
+      "step": 13667
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6988343206573666e-06,
+      "loss": 1.3252,
+      "step": 13668
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698738773170266e-06,
+      "loss": 1.3486,
+      "step": 13669
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698643225683165e-06,
+      "loss": 0.6425,
+      "step": 13670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6985476781960638e-06,
+      "loss": 1.1798,
+      "step": 13671
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698452130708963e-06,
+      "loss": 1.2672,
+      "step": 13672
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6983565832218614e-06,
+      "loss": 0.5571,
+      "step": 13673
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6982610357347604e-06,
+      "loss": 0.9919,
+      "step": 13674
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698165488247659e-06,
+      "loss": 0.7788,
+      "step": 13675
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.698069940760558e-06,
+      "loss": 0.9302,
+      "step": 13676
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6979743932734576e-06,
+      "loss": 1.2359,
+      "step": 13677
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.697878845786356e-06,
+      "loss": 0.9223,
+      "step": 13678
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6977832982992552e-06,
+      "loss": 1.2348,
+      "step": 13679
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.697687750812154e-06,
+      "loss": 1.0555,
+      "step": 13680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.697592203325053e-06,
+      "loss": 1.0398,
+      "step": 13681
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6974966558379515e-06,
+      "loss": 1.0201,
+      "step": 13682
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6974011083508505e-06,
+      "loss": 0.9654,
+      "step": 13683
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.69730556086375e-06,
+      "loss": 0.9647,
+      "step": 13684
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6972100133766486e-06,
+      "loss": 1.2552,
+      "step": 13685
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6971144658895476e-06,
+      "loss": 1.0591,
+      "step": 13686
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6970189184024463e-06,
+      "loss": 0.7325,
+      "step": 13687
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6969233709153453e-06,
+      "loss": 0.9886,
+      "step": 13688
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.696827823428244e-06,
+      "loss": 1.0131,
+      "step": 13689
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.696732275941143e-06,
+      "loss": 1.1524,
+      "step": 13690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6966367284540424e-06,
+      "loss": 1.1673,
+      "step": 13691
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.696541180966941e-06,
+      "loss": 1.3192,
+      "step": 13692
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.69644563347984e-06,
+      "loss": 1.0805,
+      "step": 13693
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6963500859927387e-06,
+      "loss": 1.2675,
+      "step": 13694
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6962545385056377e-06,
+      "loss": 0.7758,
+      "step": 13695
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6961589910185363e-06,
+      "loss": 0.9663,
+      "step": 13696
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6960634435314354e-06,
+      "loss": 1.1103,
+      "step": 13697
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6959678960443344e-06,
+      "loss": 0.9913,
+      "step": 13698
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6958723485572334e-06,
+      "loss": 1.3707,
+      "step": 13699
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.695776801070132e-06,
+      "loss": 0.7945,
+      "step": 13700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.695681253583031e-06,
+      "loss": 1.2155,
+      "step": 13701
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6955857060959297e-06,
+      "loss": 0.9081,
+      "step": 13702
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6954901586088288e-06,
+      "loss": 0.946,
+      "step": 13703
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.695394611121728e-06,
+      "loss": 0.9461,
+      "step": 13704
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.695299063634627e-06,
+      "loss": 1.1012,
+      "step": 13705
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.695203516147526e-06,
+      "loss": 0.9782,
+      "step": 13706
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6951079686604245e-06,
+      "loss": 0.8119,
+      "step": 13707
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6950124211733235e-06,
+      "loss": 1.2762,
+      "step": 13708
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.694916873686222e-06,
+      "loss": 0.9634,
+      "step": 13709
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.694821326199121e-06,
+      "loss": 0.808,
+      "step": 13710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.69472577871202e-06,
+      "loss": 1.0376,
+      "step": 13711
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.694630231224919e-06,
+      "loss": 0.9698,
+      "step": 13712
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6945346837378183e-06,
+      "loss": 0.8146,
+      "step": 13713
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.694439136250717e-06,
+      "loss": 1.0674,
+      "step": 13714
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.694343588763616e-06,
+      "loss": 0.8351,
+      "step": 13715
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6942480412765146e-06,
+      "loss": 1.214,
+      "step": 13716
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6941524937894136e-06,
+      "loss": 1.269,
+      "step": 13717
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6940569463023122e-06,
+      "loss": 0.9753,
+      "step": 13718
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6939613988152113e-06,
+      "loss": 0.5755,
+      "step": 13719
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6938658513281107e-06,
+      "loss": 0.9584,
+      "step": 13720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6937703038410093e-06,
+      "loss": 0.9357,
+      "step": 13721
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6936747563539084e-06,
+      "loss": 0.7765,
+      "step": 13722
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.693579208866807e-06,
+      "loss": 1.0534,
+      "step": 13723
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.693483661379706e-06,
+      "loss": 0.5746,
+      "step": 13724
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6933881138926046e-06,
+      "loss": 1.3755,
+      "step": 13725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6932925664055037e-06,
+      "loss": 1.3096,
+      "step": 13726
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.693197018918403e-06,
+      "loss": 1.0327,
+      "step": 13727
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6931014714313018e-06,
+      "loss": 1.0069,
+      "step": 13728
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.693005923944201e-06,
+      "loss": 1.35,
+      "step": 13729
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6929103764570994e-06,
+      "loss": 0.6915,
+      "step": 13730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6928148289699984e-06,
+      "loss": 0.7641,
+      "step": 13731
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692719281482897e-06,
+      "loss": 1.1761,
+      "step": 13732
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692623733995796e-06,
+      "loss": 0.8668,
+      "step": 13733
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6925281865086956e-06,
+      "loss": 1.3857,
+      "step": 13734
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692432639021594e-06,
+      "loss": 1.0256,
+      "step": 13735
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6923370915344932e-06,
+      "loss": 1.151,
+      "step": 13736
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692241544047392e-06,
+      "loss": 0.7298,
+      "step": 13737
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692145996560291e-06,
+      "loss": 0.7144,
+      "step": 13738
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6920504490731895e-06,
+      "loss": 0.6422,
+      "step": 13739
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6919549015860885e-06,
+      "loss": 0.9292,
+      "step": 13740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6918593540989876e-06,
+      "loss": 1.2714,
+      "step": 13741
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6917638066118866e-06,
+      "loss": 1.0923,
+      "step": 13742
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6916682591247856e-06,
+      "loss": 0.9096,
+      "step": 13743
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6915727116376843e-06,
+      "loss": 0.7865,
+      "step": 13744
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6914771641505833e-06,
+      "loss": 0.7118,
+      "step": 13745
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691381616663482e-06,
+      "loss": 1.5158,
+      "step": 13746
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691286069176381e-06,
+      "loss": 1.0221,
+      "step": 13747
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6911905216892796e-06,
+      "loss": 1.0784,
+      "step": 13748
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691094974202179e-06,
+      "loss": 0.6564,
+      "step": 13749
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6909994267150776e-06,
+      "loss": 0.8722,
+      "step": 13750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6909038792279767e-06,
+      "loss": 0.8645,
+      "step": 13751
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6908083317408753e-06,
+      "loss": 1.0874,
+      "step": 13752
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6907127842537743e-06,
+      "loss": 0.72,
+      "step": 13753
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6906172367666734e-06,
+      "loss": 0.972,
+      "step": 13754
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.690521689279572e-06,
+      "loss": 1.1159,
+      "step": 13755
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6904261417924714e-06,
+      "loss": 0.6079,
+      "step": 13756
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.69033059430537e-06,
+      "loss": 0.979,
+      "step": 13757
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.690235046818269e-06,
+      "loss": 1.0803,
+      "step": 13758
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6901394993311677e-06,
+      "loss": 0.9226,
+      "step": 13759
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6900439518440668e-06,
+      "loss": 0.7331,
+      "step": 13760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6899484043569654e-06,
+      "loss": 1.0544,
+      "step": 13761
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6898528568698644e-06,
+      "loss": 1.1851,
+      "step": 13762
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.689757309382764e-06,
+      "loss": 0.9982,
+      "step": 13763
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6896617618956625e-06,
+      "loss": 0.86,
+      "step": 13764
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6895662144085615e-06,
+      "loss": 0.8082,
+      "step": 13765
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68947066692146e-06,
+      "loss": 0.8084,
+      "step": 13766
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.689375119434359e-06,
+      "loss": 1.0931,
+      "step": 13767
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.689279571947258e-06,
+      "loss": 0.9294,
+      "step": 13768
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.689184024460157e-06,
+      "loss": 0.7123,
+      "step": 13769
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6890884769730563e-06,
+      "loss": 1.3483,
+      "step": 13770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.688992929485955e-06,
+      "loss": 0.6587,
+      "step": 13771
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.688897381998854e-06,
+      "loss": 0.8878,
+      "step": 13772
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6888018345117526e-06,
+      "loss": 0.5191,
+      "step": 13773
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6887062870246516e-06,
+      "loss": 0.5652,
+      "step": 13774
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6886107395375502e-06,
+      "loss": 0.8716,
+      "step": 13775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6885151920504493e-06,
+      "loss": 0.968,
+      "step": 13776
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6884196445633487e-06,
+      "loss": 1.0145,
+      "step": 13777
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6883240970762473e-06,
+      "loss": 0.7679,
+      "step": 13778
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6882285495891464e-06,
+      "loss": 1.2748,
+      "step": 13779
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.688133002102045e-06,
+      "loss": 1.1833,
+      "step": 13780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.688037454614944e-06,
+      "loss": 1.6726,
+      "step": 13781
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6879419071278426e-06,
+      "loss": 0.8649,
+      "step": 13782
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6878463596407417e-06,
+      "loss": 0.7142,
+      "step": 13783
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6877508121536403e-06,
+      "loss": 0.9505,
+      "step": 13784
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6876552646665398e-06,
+      "loss": 0.9013,
+      "step": 13785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.687559717179439e-06,
+      "loss": 1.2467,
+      "step": 13786
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6874641696923374e-06,
+      "loss": 0.8049,
+      "step": 13787
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6873686222052364e-06,
+      "loss": 1.1132,
+      "step": 13788
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.687273074718135e-06,
+      "loss": 1.0846,
+      "step": 13789
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.687177527231034e-06,
+      "loss": 0.9819,
+      "step": 13790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6870819797439327e-06,
+      "loss": 1.4011,
+      "step": 13791
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686986432256832e-06,
+      "loss": 1.3605,
+      "step": 13792
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686890884769731e-06,
+      "loss": 0.812,
+      "step": 13793
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68679533728263e-06,
+      "loss": 1.349,
+      "step": 13794
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686699789795529e-06,
+      "loss": 1.3895,
+      "step": 13795
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6866042423084275e-06,
+      "loss": 1.0361,
+      "step": 13796
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6865086948213265e-06,
+      "loss": 0.9597,
+      "step": 13797
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686413147334225e-06,
+      "loss": 1.122,
+      "step": 13798
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6863175998471246e-06,
+      "loss": 1.0299,
+      "step": 13799
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6862220523600232e-06,
+      "loss": 0.8063,
+      "step": 13800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6861265048729223e-06,
+      "loss": 0.7408,
+      "step": 13801
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.686030957385821e-06,
+      "loss": 1.0169,
+      "step": 13802
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68593540989872e-06,
+      "loss": 1.1621,
+      "step": 13803
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6858398624116185e-06,
+      "loss": 1.0245,
+      "step": 13804
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6857443149245176e-06,
+      "loss": 0.8357,
+      "step": 13805
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.685648767437417e-06,
+      "loss": 1.3819,
+      "step": 13806
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6855532199503156e-06,
+      "loss": 1.2135,
+      "step": 13807
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6854576724632147e-06,
+      "loss": 1.3084,
+      "step": 13808
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6853621249761133e-06,
+      "loss": 0.9676,
+      "step": 13809
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6852665774890123e-06,
+      "loss": 1.0904,
+      "step": 13810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.685171030001911e-06,
+      "loss": 1.2202,
+      "step": 13811
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68507548251481e-06,
+      "loss": 0.6727,
+      "step": 13812
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6849799350277094e-06,
+      "loss": 0.8965,
+      "step": 13813
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.684884387540608e-06,
+      "loss": 0.6995,
+      "step": 13814
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.684788840053507e-06,
+      "loss": 0.8624,
+      "step": 13815
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6846932925664057e-06,
+      "loss": 1.0697,
+      "step": 13816
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6845977450793048e-06,
+      "loss": 1.2146,
+      "step": 13817
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6845021975922034e-06,
+      "loss": 1.1588,
+      "step": 13818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6844066501051024e-06,
+      "loss": 1.1357,
+      "step": 13819
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.684311102618001e-06,
+      "loss": 1.0728,
+      "step": 13820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6842155551309005e-06,
+      "loss": 0.725,
+      "step": 13821
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6841200076437995e-06,
+      "loss": 1.3152,
+      "step": 13822
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.684024460156698e-06,
+      "loss": 0.7905,
+      "step": 13823
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.683928912669597e-06,
+      "loss": 1.3771,
+      "step": 13824
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.683833365182496e-06,
+      "loss": 1.129,
+      "step": 13825
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.683737817695395e-06,
+      "loss": 1.3409,
+      "step": 13826
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6836422702082934e-06,
+      "loss": 0.7639,
+      "step": 13827
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.683546722721193e-06,
+      "loss": 0.9356,
+      "step": 13828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.683451175234092e-06,
+      "loss": 1.0699,
+      "step": 13829
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6833556277469906e-06,
+      "loss": 0.8331,
+      "step": 13830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6832600802598896e-06,
+      "loss": 0.742,
+      "step": 13831
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6831645327727882e-06,
+      "loss": 0.8938,
+      "step": 13832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6830689852856873e-06,
+      "loss": 0.8379,
+      "step": 13833
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682973437798586e-06,
+      "loss": 0.754,
+      "step": 13834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6828778903114853e-06,
+      "loss": 0.5818,
+      "step": 13835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6827823428243844e-06,
+      "loss": 0.7674,
+      "step": 13836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682686795337283e-06,
+      "loss": 0.8868,
+      "step": 13837
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682591247850182e-06,
+      "loss": 1.2997,
+      "step": 13838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6824957003630806e-06,
+      "loss": 1.1649,
+      "step": 13839
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6824001528759797e-06,
+      "loss": 0.8071,
+      "step": 13840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6823046053888783e-06,
+      "loss": 0.6578,
+      "step": 13841
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6822090579017778e-06,
+      "loss": 1.1893,
+      "step": 13842
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6821135104146764e-06,
+      "loss": 1.3653,
+      "step": 13843
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6820179629275754e-06,
+      "loss": 1.11,
+      "step": 13844
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6819224154404744e-06,
+      "loss": 0.9491,
+      "step": 13845
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681826867953373e-06,
+      "loss": 1.1071,
+      "step": 13846
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681731320466272e-06,
+      "loss": 1.281,
+      "step": 13847
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6816357729791707e-06,
+      "loss": 0.7367,
+      "step": 13848
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.68154022549207e-06,
+      "loss": 0.9621,
+      "step": 13849
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681444678004969e-06,
+      "loss": 0.5381,
+      "step": 13850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681349130517868e-06,
+      "loss": 0.9707,
+      "step": 13851
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6812535830307664e-06,
+      "loss": 1.0504,
+      "step": 13852
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6811580355436655e-06,
+      "loss": 1.0751,
+      "step": 13853
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681062488056564e-06,
+      "loss": 1.3635,
+      "step": 13854
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.680966940569463e-06,
+      "loss": 0.9541,
+      "step": 13855
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.680871393082362e-06,
+      "loss": 1.4901,
+      "step": 13856
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6807758455952612e-06,
+      "loss": 0.8871,
+      "step": 13857
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6806802981081603e-06,
+      "loss": 1.1572,
+      "step": 13858
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.680584750621059e-06,
+      "loss": 0.868,
+      "step": 13859
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.680489203133958e-06,
+      "loss": 1.2786,
+      "step": 13860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6803936556468565e-06,
+      "loss": 1.0824,
+      "step": 13861
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6802981081597556e-06,
+      "loss": 0.7998,
+      "step": 13862
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.680202560672654e-06,
+      "loss": 0.9615,
+      "step": 13863
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6801070131855536e-06,
+      "loss": 1.433,
+      "step": 13864
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6800114656984527e-06,
+      "loss": 0.7374,
+      "step": 13865
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6799159182113513e-06,
+      "loss": 1.234,
+      "step": 13866
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6798203707242503e-06,
+      "loss": 1.1346,
+      "step": 13867
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.679724823237149e-06,
+      "loss": 1.3363,
+      "step": 13868
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.679629275750048e-06,
+      "loss": 0.9436,
+      "step": 13869
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6795337282629466e-06,
+      "loss": 1.0607,
+      "step": 13870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.679438180775846e-06,
+      "loss": 0.9812,
+      "step": 13871
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.679342633288745e-06,
+      "loss": 0.8372,
+      "step": 13872
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6792470858016437e-06,
+      "loss": 1.1135,
+      "step": 13873
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6791515383145428e-06,
+      "loss": 1.0463,
+      "step": 13874
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6790559908274414e-06,
+      "loss": 0.5413,
+      "step": 13875
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6789604433403404e-06,
+      "loss": 1.1752,
+      "step": 13876
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678864895853239e-06,
+      "loss": 1.6427,
+      "step": 13877
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6787693483661385e-06,
+      "loss": 0.8971,
+      "step": 13878
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6786738008790375e-06,
+      "loss": 1.1203,
+      "step": 13879
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678578253391936e-06,
+      "loss": 1.2935,
+      "step": 13880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678482705904835e-06,
+      "loss": 0.996,
+      "step": 13881
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678387158417734e-06,
+      "loss": 1.1639,
+      "step": 13882
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678291610930633e-06,
+      "loss": 0.9289,
+      "step": 13883
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6781960634435314e-06,
+      "loss": 1.0157,
+      "step": 13884
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.678100515956431e-06,
+      "loss": 0.849,
+      "step": 13885
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.67800496846933e-06,
+      "loss": 1.2456,
+      "step": 13886
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6779094209822286e-06,
+      "loss": 0.8708,
+      "step": 13887
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6778138734951276e-06,
+      "loss": 1.3484,
+      "step": 13888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6777183260080262e-06,
+      "loss": 1.2624,
+      "step": 13889
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6776227785209253e-06,
+      "loss": 1.2401,
+      "step": 13890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677527231033824e-06,
+      "loss": 1.1983,
+      "step": 13891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677431683546723e-06,
+      "loss": 0.5889,
+      "step": 13892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677336136059622e-06,
+      "loss": 0.6101,
+      "step": 13893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677240588572521e-06,
+      "loss": 1.012,
+      "step": 13894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6771450410854196e-06,
+      "loss": 0.8335,
+      "step": 13895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6770494935983186e-06,
+      "loss": 0.8827,
+      "step": 13896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6769539461112177e-06,
+      "loss": 1.0175,
+      "step": 13897
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6768583986241163e-06,
+      "loss": 1.328,
+      "step": 13898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6767628511370153e-06,
+      "loss": 1.5615,
+      "step": 13899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6766673036499144e-06,
+      "loss": 1.0665,
+      "step": 13900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6765717561628134e-06,
+      "loss": 1.5899,
+      "step": 13901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.676476208675712e-06,
+      "loss": 0.5134,
+      "step": 13902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.676380661188611e-06,
+      "loss": 0.6833,
+      "step": 13903
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6762851137015097e-06,
+      "loss": 0.9919,
+      "step": 13904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6761895662144087e-06,
+      "loss": 0.747,
+      "step": 13905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6760940187273073e-06,
+      "loss": 1.4467,
+      "step": 13906
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675998471240207e-06,
+      "loss": 0.8287,
+      "step": 13907
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675902923753106e-06,
+      "loss": 1.1608,
+      "step": 13908
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6758073762660044e-06,
+      "loss": 1.0576,
+      "step": 13909
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6757118287789035e-06,
+      "loss": 1.4083,
+      "step": 13910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675616281291802e-06,
+      "loss": 1.1673,
+      "step": 13911
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675520733804701e-06,
+      "loss": 0.749,
+      "step": 13912
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6754251863175998e-06,
+      "loss": 1.0317,
+      "step": 13913
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6753296388304992e-06,
+      "loss": 0.9023,
+      "step": 13914
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6752340913433983e-06,
+      "loss": 1.4507,
+      "step": 13915
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675138543856297e-06,
+      "loss": 1.3317,
+      "step": 13916
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.675042996369196e-06,
+      "loss": 1.1722,
+      "step": 13917
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6749474488820945e-06,
+      "loss": 0.926,
+      "step": 13918
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6748519013949936e-06,
+      "loss": 1.0043,
+      "step": 13919
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.674756353907892e-06,
+      "loss": 0.8773,
+      "step": 13920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6746608064207916e-06,
+      "loss": 0.8805,
+      "step": 13921
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6745652589336907e-06,
+      "loss": 0.8878,
+      "step": 13922
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6744697114465893e-06,
+      "loss": 0.9291,
+      "step": 13923
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6743741639594883e-06,
+      "loss": 1.1455,
+      "step": 13924
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.674278616472387e-06,
+      "loss": 1.032,
+      "step": 13925
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.674183068985286e-06,
+      "loss": 0.8992,
+      "step": 13926
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6740875214981846e-06,
+      "loss": 1.128,
+      "step": 13927
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6739919740110836e-06,
+      "loss": 0.7047,
+      "step": 13928
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.673896426523983e-06,
+      "loss": 1.204,
+      "step": 13929
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6738008790368817e-06,
+      "loss": 0.7657,
+      "step": 13930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6737053315497808e-06,
+      "loss": 1.1651,
+      "step": 13931
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6736097840626794e-06,
+      "loss": 0.9862,
+      "step": 13932
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6735142365755784e-06,
+      "loss": 1.3007,
+      "step": 13933
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.673418689088477e-06,
+      "loss": 0.7716,
+      "step": 13934
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.673323141601376e-06,
+      "loss": 0.9065,
+      "step": 13935
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6732275941142755e-06,
+      "loss": 0.6822,
+      "step": 13936
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.673132046627174e-06,
+      "loss": 1.2459,
+      "step": 13937
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.673036499140073e-06,
+      "loss": 1.0934,
+      "step": 13938
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672940951652972e-06,
+      "loss": 0.6254,
+      "step": 13939
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672845404165871e-06,
+      "loss": 0.9143,
+      "step": 13940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6727498566787694e-06,
+      "loss": 0.9744,
+      "step": 13941
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6726543091916685e-06,
+      "loss": 1.1412,
+      "step": 13942
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6725587617045675e-06,
+      "loss": 0.7865,
+      "step": 13943
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6724632142174666e-06,
+      "loss": 0.8078,
+      "step": 13944
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672367666730365e-06,
+      "loss": 0.6828,
+      "step": 13945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6722721192432642e-06,
+      "loss": 1.0746,
+      "step": 13946
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6721765717561633e-06,
+      "loss": 0.639,
+      "step": 13947
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672081024269062e-06,
+      "loss": 0.7289,
+      "step": 13948
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671985476781961e-06,
+      "loss": 0.7067,
+      "step": 13949
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.67188992929486e-06,
+      "loss": 1.2301,
+      "step": 13950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671794381807759e-06,
+      "loss": 1.2757,
+      "step": 13951
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6716988343206576e-06,
+      "loss": 0.8728,
+      "step": 13952
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6716032868335566e-06,
+      "loss": 0.8963,
+      "step": 13953
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6715077393464553e-06,
+      "loss": 1.0321,
+      "step": 13954
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6714121918593543e-06,
+      "loss": 0.5138,
+      "step": 13955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671316644372253e-06,
+      "loss": 1.1979,
+      "step": 13956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6712210968851524e-06,
+      "loss": 0.793,
+      "step": 13957
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6711255493980514e-06,
+      "loss": 1.0023,
+      "step": 13958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.67103000191095e-06,
+      "loss": 1.1017,
+      "step": 13959
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.670934454423849e-06,
+      "loss": 1.224,
+      "step": 13960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6708389069367477e-06,
+      "loss": 0.9396,
+      "step": 13961
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6707433594496467e-06,
+      "loss": 1.2536,
+      "step": 13962
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6706478119625453e-06,
+      "loss": 0.7906,
+      "step": 13963
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6705522644754444e-06,
+      "loss": 1.0663,
+      "step": 13964
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.670456716988344e-06,
+      "loss": 1.1489,
+      "step": 13965
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6703611695012424e-06,
+      "loss": 1.2713,
+      "step": 13966
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6702656220141415e-06,
+      "loss": 1.5291,
+      "step": 13967
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.67017007452704e-06,
+      "loss": 1.1527,
+      "step": 13968
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.670074527039939e-06,
+      "loss": 1.0835,
+      "step": 13969
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6699789795528378e-06,
+      "loss": 1.0899,
+      "step": 13970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.669883432065737e-06,
+      "loss": 0.9893,
+      "step": 13971
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6697878845786363e-06,
+      "loss": 0.783,
+      "step": 13972
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.669692337091535e-06,
+      "loss": 0.9314,
+      "step": 13973
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.669596789604434e-06,
+      "loss": 0.8693,
+      "step": 13974
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6695012421173325e-06,
+      "loss": 0.8728,
+      "step": 13975
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6694056946302316e-06,
+      "loss": 0.7313,
+      "step": 13976
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.66931014714313e-06,
+      "loss": 0.7768,
+      "step": 13977
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6692145996560292e-06,
+      "loss": 1.5221,
+      "step": 13978
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6691190521689287e-06,
+      "loss": 1.0461,
+      "step": 13979
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6690235046818273e-06,
+      "loss": 1.0707,
+      "step": 13980
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6689279571947263e-06,
+      "loss": 1.0154,
+      "step": 13981
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.668832409707625e-06,
+      "loss": 1.0886,
+      "step": 13982
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.668736862220524e-06,
+      "loss": 1.6893,
+      "step": 13983
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6686413147334226e-06,
+      "loss": 1.1679,
+      "step": 13984
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6685457672463216e-06,
+      "loss": 1.0546,
+      "step": 13985
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.668450219759221e-06,
+      "loss": 0.9154,
+      "step": 13986
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6683546722721197e-06,
+      "loss": 1.2357,
+      "step": 13987
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6682591247850188e-06,
+      "loss": 1.2695,
+      "step": 13988
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6681635772979174e-06,
+      "loss": 1.1628,
+      "step": 13989
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6680680298108164e-06,
+      "loss": 0.737,
+      "step": 13990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667972482323715e-06,
+      "loss": 1.0093,
+      "step": 13991
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667876934836614e-06,
+      "loss": 0.7121,
+      "step": 13992
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667781387349513e-06,
+      "loss": 1.2602,
+      "step": 13993
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667685839862412e-06,
+      "loss": 1.1379,
+      "step": 13994
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6675902923753108e-06,
+      "loss": 0.8329,
+      "step": 13995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.66749474488821e-06,
+      "loss": 1.3914,
+      "step": 13996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6673991974011084e-06,
+      "loss": 1.0519,
+      "step": 13997
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6673036499140074e-06,
+      "loss": 0.7286,
+      "step": 13998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6672081024269065e-06,
+      "loss": 0.9329,
+      "step": 13999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667112554939805e-06,
+      "loss": 0.8499,
+      "step": 14000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6670170074527046e-06,
+      "loss": 0.5866,
+      "step": 14001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666921459965603e-06,
+      "loss": 1.2798,
+      "step": 14002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6668259124785022e-06,
+      "loss": 1.0284,
+      "step": 14003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666730364991401e-06,
+      "loss": 0.6528,
+      "step": 14004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6666348175043e-06,
+      "loss": 1.2135,
+      "step": 14005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6665392700171985e-06,
+      "loss": 1.3503,
+      "step": 14006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6664437225300975e-06,
+      "loss": 0.8859,
+      "step": 14007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666348175042997e-06,
+      "loss": 1.0032,
+      "step": 14008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6662526275558956e-06,
+      "loss": 0.7284,
+      "step": 14009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6661570800687946e-06,
+      "loss": 0.885,
+      "step": 14010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6660615325816933e-06,
+      "loss": 0.9547,
+      "step": 14011
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6659659850945923e-06,
+      "loss": 0.8575,
+      "step": 14012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.665870437607491e-06,
+      "loss": 0.6831,
+      "step": 14013
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66577489012039e-06,
+      "loss": 0.7524,
+      "step": 14014
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6656793426332894e-06,
+      "loss": 0.8792,
+      "step": 14015
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.665583795146188e-06,
+      "loss": 0.8139,
+      "step": 14016
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.665488247659087e-06,
+      "loss": 0.8974,
+      "step": 14017
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6653927001719857e-06,
+      "loss": 1.0236,
+      "step": 14018
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6652971526848847e-06,
+      "loss": 1.0829,
+      "step": 14019
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6652016051977833e-06,
+      "loss": 0.9667,
+      "step": 14020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6651060577106824e-06,
+      "loss": 0.858,
+      "step": 14021
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.665010510223582e-06,
+      "loss": 1.2141,
+      "step": 14022
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6649149627364804e-06,
+      "loss": 1.1893,
+      "step": 14023
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6648194152493795e-06,
+      "loss": 1.1542,
+      "step": 14024
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.664723867762278e-06,
+      "loss": 1.0882,
+      "step": 14025
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.664628320275177e-06,
+      "loss": 0.7658,
+      "step": 14026
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6645327727880758e-06,
+      "loss": 1.3152,
+      "step": 14027
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.664437225300975e-06,
+      "loss": 1.0198,
+      "step": 14028
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6643416778138743e-06,
+      "loss": 0.7965,
+      "step": 14029
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.664246130326773e-06,
+      "loss": 1.0921,
+      "step": 14030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.664150582839672e-06,
+      "loss": 0.498,
+      "step": 14031
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6640550353525705e-06,
+      "loss": 0.8309,
+      "step": 14032
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6639594878654696e-06,
+      "loss": 0.6145,
+      "step": 14033
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.663863940378368e-06,
+      "loss": 1.1953,
+      "step": 14034
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6637683928912672e-06,
+      "loss": 0.9842,
+      "step": 14035
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.663672845404166e-06,
+      "loss": 0.9827,
+      "step": 14036
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6635772979170653e-06,
+      "loss": 1.0324,
+      "step": 14037
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6634817504299643e-06,
+      "loss": 1.1514,
+      "step": 14038
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.663386202942863e-06,
+      "loss": 0.5828,
+      "step": 14039
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.663290655455762e-06,
+      "loss": 0.9143,
+      "step": 14040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6631951079686606e-06,
+      "loss": 0.8133,
+      "step": 14041
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6630995604815596e-06,
+      "loss": 0.6061,
+      "step": 14042
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6630040129944583e-06,
+      "loss": 1.1298,
+      "step": 14043
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6629084655073577e-06,
+      "loss": 0.8997,
+      "step": 14044
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6628129180202563e-06,
+      "loss": 0.9188,
+      "step": 14045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6627173705331554e-06,
+      "loss": 0.6185,
+      "step": 14046
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662621823046054e-06,
+      "loss": 1.0413,
+      "step": 14047
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662526275558953e-06,
+      "loss": 0.9135,
+      "step": 14048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662430728071852e-06,
+      "loss": 1.1301,
+      "step": 14049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6623351805847507e-06,
+      "loss": 0.8068,
+      "step": 14050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66223963309765e-06,
+      "loss": 1.0544,
+      "step": 14051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6621440856105488e-06,
+      "loss": 0.7834,
+      "step": 14052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662048538123448e-06,
+      "loss": 0.7397,
+      "step": 14053
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6619529906363464e-06,
+      "loss": 0.6679,
+      "step": 14054
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6618574431492454e-06,
+      "loss": 0.957,
+      "step": 14055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661761895662144e-06,
+      "loss": 1.4673,
+      "step": 14056
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661666348175043e-06,
+      "loss": 0.7869,
+      "step": 14057
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6615708006879426e-06,
+      "loss": 0.6444,
+      "step": 14058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661475253200841e-06,
+      "loss": 1.1046,
+      "step": 14059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6613797057137402e-06,
+      "loss": 1.3549,
+      "step": 14060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661284158226639e-06,
+      "loss": 1.0908,
+      "step": 14061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661188610739538e-06,
+      "loss": 1.1491,
+      "step": 14062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6610930632524365e-06,
+      "loss": 1.0511,
+      "step": 14063
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6609975157653355e-06,
+      "loss": 1.1829,
+      "step": 14064
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660901968278235e-06,
+      "loss": 1.0584,
+      "step": 14065
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6608064207911336e-06,
+      "loss": 1.2134,
+      "step": 14066
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6607108733040326e-06,
+      "loss": 0.8979,
+      "step": 14067
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6606153258169312e-06,
+      "loss": 0.9444,
+      "step": 14068
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6605197783298303e-06,
+      "loss": 0.9542,
+      "step": 14069
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660424230842729e-06,
+      "loss": 1.1662,
+      "step": 14070
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660328683355628e-06,
+      "loss": 1.4557,
+      "step": 14071
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6602331358685266e-06,
+      "loss": 1.0465,
+      "step": 14072
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660137588381426e-06,
+      "loss": 0.935,
+      "step": 14073
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660042040894325e-06,
+      "loss": 0.928,
+      "step": 14074
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6599464934072237e-06,
+      "loss": 1.1064,
+      "step": 14075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6598509459201227e-06,
+      "loss": 1.5505,
+      "step": 14076
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6597553984330213e-06,
+      "loss": 0.752,
+      "step": 14077
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6596598509459204e-06,
+      "loss": 1.0813,
+      "step": 14078
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.659564303458819e-06,
+      "loss": 1.2393,
+      "step": 14079
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6594687559717184e-06,
+      "loss": 0.7155,
+      "step": 14080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6593732084846175e-06,
+      "loss": 1.03,
+      "step": 14081
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.659277660997516e-06,
+      "loss": 1.1148,
+      "step": 14082
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.659182113510415e-06,
+      "loss": 1.046,
+      "step": 14083
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6590865660233137e-06,
+      "loss": 0.9648,
+      "step": 14084
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6589910185362128e-06,
+      "loss": 0.8135,
+      "step": 14085
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6588954710491114e-06,
+      "loss": 0.9895,
+      "step": 14086
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.658799923562011e-06,
+      "loss": 0.5461,
+      "step": 14087
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6587043760749095e-06,
+      "loss": 0.5652,
+      "step": 14088
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6586088285878085e-06,
+      "loss": 0.827,
+      "step": 14089
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6585132811007076e-06,
+      "loss": 1.1028,
+      "step": 14090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.658417733613606e-06,
+      "loss": 1.059,
+      "step": 14091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.658322186126505e-06,
+      "loss": 0.6273,
+      "step": 14092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.658226638639404e-06,
+      "loss": 0.7991,
+      "step": 14093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6581310911523033e-06,
+      "loss": 1.053,
+      "step": 14094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.658035543665202e-06,
+      "loss": 0.9835,
+      "step": 14095
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.657939996178101e-06,
+      "loss": 1.0417,
+      "step": 14096
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6578444486909996e-06,
+      "loss": 1.2091,
+      "step": 14097
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6577489012038986e-06,
+      "loss": 0.8846,
+      "step": 14098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.657653353716797e-06,
+      "loss": 0.8971,
+      "step": 14099
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6575578062296962e-06,
+      "loss": 0.9143,
+      "step": 14100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6574622587425957e-06,
+      "loss": 1.1497,
+      "step": 14101
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6573667112554943e-06,
+      "loss": 0.6924,
+      "step": 14102
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6572711637683934e-06,
+      "loss": 1.0965,
+      "step": 14103
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.657175616281292e-06,
+      "loss": 0.6498,
+      "step": 14104
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.657080068794191e-06,
+      "loss": 0.7307,
+      "step": 14105
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6569845213070896e-06,
+      "loss": 0.9216,
+      "step": 14106
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6568889738199887e-06,
+      "loss": 0.9043,
+      "step": 14107
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6567934263328873e-06,
+      "loss": 1.1401,
+      "step": 14108
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6566978788457867e-06,
+      "loss": 0.9207,
+      "step": 14109
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6566023313586858e-06,
+      "loss": 0.8236,
+      "step": 14110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6565067838715844e-06,
+      "loss": 1.0847,
+      "step": 14111
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6564112363844834e-06,
+      "loss": 0.8888,
+      "step": 14112
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.656315688897382e-06,
+      "loss": 1.7256,
+      "step": 14113
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.656220141410281e-06,
+      "loss": 1.061,
+      "step": 14114
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6561245939231797e-06,
+      "loss": 0.909,
+      "step": 14115
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.656029046436079e-06,
+      "loss": 1.0207,
+      "step": 14116
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.655933498948978e-06,
+      "loss": 0.7884,
+      "step": 14117
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.655837951461877e-06,
+      "loss": 1.4426,
+      "step": 14118
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.655742403974776e-06,
+      "loss": 0.8898,
+      "step": 14119
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6556468564876745e-06,
+      "loss": 0.7215,
+      "step": 14120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6555513090005735e-06,
+      "loss": 0.8369,
+      "step": 14121
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.655455761513472e-06,
+      "loss": 1.3028,
+      "step": 14122
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6553602140263716e-06,
+      "loss": 1.2204,
+      "step": 14123
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6552646665392706e-06,
+      "loss": 0.8246,
+      "step": 14124
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6551691190521692e-06,
+      "loss": 1.0359,
+      "step": 14125
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6550735715650683e-06,
+      "loss": 1.1226,
+      "step": 14126
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.654978024077967e-06,
+      "loss": 1.1607,
+      "step": 14127
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.654882476590866e-06,
+      "loss": 0.4782,
+      "step": 14128
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6547869291037646e-06,
+      "loss": 0.5279,
+      "step": 14129
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.654691381616664e-06,
+      "loss": 0.7945,
+      "step": 14130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.654595834129563e-06,
+      "loss": 1.6624,
+      "step": 14131
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6545002866424617e-06,
+      "loss": 1.6354,
+      "step": 14132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6544047391553607e-06,
+      "loss": 0.801,
+      "step": 14133
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6543091916682593e-06,
+      "loss": 1.3517,
+      "step": 14134
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6542136441811584e-06,
+      "loss": 1.3559,
+      "step": 14135
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.654118096694057e-06,
+      "loss": 1.1297,
+      "step": 14136
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6540225492069564e-06,
+      "loss": 1.439,
+      "step": 14137
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.653927001719855e-06,
+      "loss": 0.6835,
+      "step": 14138
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.653831454232754e-06,
+      "loss": 0.804,
+      "step": 14139
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.653735906745653e-06,
+      "loss": 0.91,
+      "step": 14140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6536403592585517e-06,
+      "loss": 0.8127,
+      "step": 14141
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6535448117714508e-06,
+      "loss": 0.5208,
+      "step": 14142
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6534492642843494e-06,
+      "loss": 1.1424,
+      "step": 14143
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6533537167972484e-06,
+      "loss": 0.7748,
+      "step": 14144
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6532581693101475e-06,
+      "loss": 1.0024,
+      "step": 14145
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6531626218230465e-06,
+      "loss": 0.8829,
+      "step": 14146
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.653067074335945e-06,
+      "loss": 0.7834,
+      "step": 14147
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652971526848844e-06,
+      "loss": 1.1277,
+      "step": 14148
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6528759793617428e-06,
+      "loss": 0.5996,
+      "step": 14149
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652780431874642e-06,
+      "loss": 1.0355,
+      "step": 14150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652684884387541e-06,
+      "loss": 0.8551,
+      "step": 14151
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.65258933690044e-06,
+      "loss": 1.1129,
+      "step": 14152
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652493789413339e-06,
+      "loss": 0.6658,
+      "step": 14153
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6523982419262376e-06,
+      "loss": 0.5262,
+      "step": 14154
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6523026944391366e-06,
+      "loss": 0.9563,
+      "step": 14155
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652207146952035e-06,
+      "loss": 1.3046,
+      "step": 14156
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6521115994649342e-06,
+      "loss": 1.4423,
+      "step": 14157
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.652016051977833e-06,
+      "loss": 1.2262,
+      "step": 14158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6519205044907323e-06,
+      "loss": 1.0533,
+      "step": 14159
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6518249570036314e-06,
+      "loss": 0.6356,
+      "step": 14160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.65172940951653e-06,
+      "loss": 1.0379,
+      "step": 14161
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.651633862029429e-06,
+      "loss": 0.7509,
+      "step": 14162
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6515383145423276e-06,
+      "loss": 1.3025,
+      "step": 14163
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6514427670552267e-06,
+      "loss": 1.2581,
+      "step": 14164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6513472195681253e-06,
+      "loss": 1.035,
+      "step": 14165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6512516720810247e-06,
+      "loss": 0.8668,
+      "step": 14166
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6511561245939238e-06,
+      "loss": 1.1679,
+      "step": 14167
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6510605771068224e-06,
+      "loss": 1.0939,
+      "step": 14168
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6509650296197214e-06,
+      "loss": 1.2812,
+      "step": 14169
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.65086948213262e-06,
+      "loss": 0.7299,
+      "step": 14170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.650773934645519e-06,
+      "loss": 0.6359,
+      "step": 14171
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6506783871584177e-06,
+      "loss": 0.8909,
+      "step": 14172
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.650582839671317e-06,
+      "loss": 1.1273,
+      "step": 14173
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.650487292184216e-06,
+      "loss": 0.6928,
+      "step": 14174
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.650391744697115e-06,
+      "loss": 0.7599,
+      "step": 14175
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.650296197210014e-06,
+      "loss": 0.651,
+      "step": 14176
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6502006497229125e-06,
+      "loss": 1.0028,
+      "step": 14177
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6501051022358115e-06,
+      "loss": 1.1487,
+      "step": 14178
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.65000955474871e-06,
+      "loss": 0.7555,
+      "step": 14179
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.649914007261609e-06,
+      "loss": 0.8979,
+      "step": 14180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6498184597745086e-06,
+      "loss": 1.0895,
+      "step": 14181
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6497229122874072e-06,
+      "loss": 0.8222,
+      "step": 14182
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6496273648003063e-06,
+      "loss": 0.9577,
+      "step": 14183
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.649531817313205e-06,
+      "loss": 0.9664,
+      "step": 14184
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.649436269826104e-06,
+      "loss": 1.3152,
+      "step": 14185
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6493407223390026e-06,
+      "loss": 0.9769,
+      "step": 14186
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6492451748519016e-06,
+      "loss": 0.8068,
+      "step": 14187
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6491496273648006e-06,
+      "loss": 1.2051,
+      "step": 14188
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6490540798776997e-06,
+      "loss": 0.9313,
+      "step": 14189
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6489585323905983e-06,
+      "loss": 0.7347,
+      "step": 14190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6488629849034973e-06,
+      "loss": 1.0752,
+      "step": 14191
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6487674374163964e-06,
+      "loss": 1.5755,
+      "step": 14192
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.648671889929295e-06,
+      "loss": 0.872,
+      "step": 14193
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.648576342442194e-06,
+      "loss": 0.9892,
+      "step": 14194
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.648480794955093e-06,
+      "loss": 1.1341,
+      "step": 14195
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.648385247467992e-06,
+      "loss": 0.8925,
+      "step": 14196
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6482896999808907e-06,
+      "loss": 0.9085,
+      "step": 14197
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6481941524937897e-06,
+      "loss": 0.9184,
+      "step": 14198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6480986050066884e-06,
+      "loss": 1.017,
+      "step": 14199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6480030575195874e-06,
+      "loss": 1.0867,
+      "step": 14200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647907510032486e-06,
+      "loss": 0.9832,
+      "step": 14201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6478119625453855e-06,
+      "loss": 0.764,
+      "step": 14202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6477164150582845e-06,
+      "loss": 0.8324,
+      "step": 14203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647620867571183e-06,
+      "loss": 1.3178,
+      "step": 14204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647525320084082e-06,
+      "loss": 1.3861,
+      "step": 14205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6474297725969808e-06,
+      "loss": 1.0761,
+      "step": 14206
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.64733422510988e-06,
+      "loss": 0.9468,
+      "step": 14207
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6472386776227784e-06,
+      "loss": 0.9627,
+      "step": 14208
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647143130135678e-06,
+      "loss": 1.0069,
+      "step": 14209
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.647047582648577e-06,
+      "loss": 1.1526,
+      "step": 14210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6469520351614756e-06,
+      "loss": 0.6983,
+      "step": 14211
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6468564876743746e-06,
+      "loss": 0.9982,
+      "step": 14212
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646760940187273e-06,
+      "loss": 0.534,
+      "step": 14213
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6466653927001722e-06,
+      "loss": 1.0786,
+      "step": 14214
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646569845213071e-06,
+      "loss": 1.069,
+      "step": 14215
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.64647429772597e-06,
+      "loss": 0.6471,
+      "step": 14216
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6463787502388694e-06,
+      "loss": 0.9216,
+      "step": 14217
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646283202751768e-06,
+      "loss": 1.3355,
+      "step": 14218
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646187655264667e-06,
+      "loss": 0.812,
+      "step": 14219
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6460921077775656e-06,
+      "loss": 0.6021,
+      "step": 14220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6459965602904647e-06,
+      "loss": 0.8253,
+      "step": 14221
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6459010128033633e-06,
+      "loss": 1.3752,
+      "step": 14222
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6458054653162623e-06,
+      "loss": 0.8835,
+      "step": 14223
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6457099178291618e-06,
+      "loss": 0.7918,
+      "step": 14224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6456143703420604e-06,
+      "loss": 1.0894,
+      "step": 14225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6455188228549594e-06,
+      "loss": 1.0812,
+      "step": 14226
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.645423275367858e-06,
+      "loss": 1.065,
+      "step": 14227
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.645327727880757e-06,
+      "loss": 0.6327,
+      "step": 14228
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6452321803936557e-06,
+      "loss": 1.0887,
+      "step": 14229
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6451366329065547e-06,
+      "loss": 1.1606,
+      "step": 14230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.645041085419454e-06,
+      "loss": 1.0566,
+      "step": 14231
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644945537932353e-06,
+      "loss": 0.7989,
+      "step": 14232
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644849990445252e-06,
+      "loss": 0.7826,
+      "step": 14233
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6447544429581505e-06,
+      "loss": 1.2572,
+      "step": 14234
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6446588954710495e-06,
+      "loss": 1.0342,
+      "step": 14235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644563347983948e-06,
+      "loss": 1.4369,
+      "step": 14236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644467800496847e-06,
+      "loss": 1.3949,
+      "step": 14237
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644372253009746e-06,
+      "loss": 0.7929,
+      "step": 14238
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6442767055226452e-06,
+      "loss": 1.0502,
+      "step": 14239
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644181158035544e-06,
+      "loss": 1.4832,
+      "step": 14240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.644085610548443e-06,
+      "loss": 1.0258,
+      "step": 14241
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.643990063061342e-06,
+      "loss": 0.917,
+      "step": 14242
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6438945155742406e-06,
+      "loss": 0.9356,
+      "step": 14243
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6437989680871396e-06,
+      "loss": 0.7974,
+      "step": 14244
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6437034206000386e-06,
+      "loss": 0.9359,
+      "step": 14245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6436078731129377e-06,
+      "loss": 1.0701,
+      "step": 14246
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6435123256258363e-06,
+      "loss": 1.1188,
+      "step": 14247
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6434167781387353e-06,
+      "loss": 1.1612,
+      "step": 14248
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.643321230651634e-06,
+      "loss": 0.903,
+      "step": 14249
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.643225683164533e-06,
+      "loss": 0.8867,
+      "step": 14250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6431301356774316e-06,
+      "loss": 1.1017,
+      "step": 14251
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6430345881903306e-06,
+      "loss": 1.3904,
+      "step": 14252
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.64293904070323e-06,
+      "loss": 1.1427,
+      "step": 14253
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6428434932161287e-06,
+      "loss": 0.8438,
+      "step": 14254
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6427479457290277e-06,
+      "loss": 1.0455,
+      "step": 14255
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6426523982419264e-06,
+      "loss": 0.7709,
+      "step": 14256
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6425568507548254e-06,
+      "loss": 0.7185,
+      "step": 14257
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642461303267724e-06,
+      "loss": 0.8969,
+      "step": 14258
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642365755780623e-06,
+      "loss": 1.2488,
+      "step": 14259
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6422702082935225e-06,
+      "loss": 1.3595,
+      "step": 14260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642174660806421e-06,
+      "loss": 0.7814,
+      "step": 14261
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.64207911331932e-06,
+      "loss": 0.9358,
+      "step": 14262
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6419835658322188e-06,
+      "loss": 1.2436,
+      "step": 14263
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641888018345118e-06,
+      "loss": 0.9631,
+      "step": 14264
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6417924708580164e-06,
+      "loss": 1.2513,
+      "step": 14265
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6416969233709155e-06,
+      "loss": 1.3408,
+      "step": 14266
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641601375883815e-06,
+      "loss": 1.2622,
+      "step": 14267
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6415058283967136e-06,
+      "loss": 1.015,
+      "step": 14268
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6414102809096126e-06,
+      "loss": 0.9276,
+      "step": 14269
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641314733422511e-06,
+      "loss": 1.3674,
+      "step": 14270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6412191859354102e-06,
+      "loss": 1.0438,
+      "step": 14271
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641123638448309e-06,
+      "loss": 1.1465,
+      "step": 14272
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.641028090961208e-06,
+      "loss": 1.2915,
+      "step": 14273
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6409325434741074e-06,
+      "loss": 0.7821,
+      "step": 14274
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.640836995987006e-06,
+      "loss": 1.0506,
+      "step": 14275
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.640741448499905e-06,
+      "loss": 0.9808,
+      "step": 14276
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6406459010128036e-06,
+      "loss": 1.077,
+      "step": 14277
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6405503535257027e-06,
+      "loss": 1.3137,
+      "step": 14278
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6404548060386013e-06,
+      "loss": 1.3182,
+      "step": 14279
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6403592585515003e-06,
+      "loss": 0.8561,
+      "step": 14280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6402637110643998e-06,
+      "loss": 1.0541,
+      "step": 14281
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6401681635772984e-06,
+      "loss": 0.6702,
+      "step": 14282
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6400726160901974e-06,
+      "loss": 0.797,
+      "step": 14283
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639977068603096e-06,
+      "loss": 0.3953,
+      "step": 14284
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639881521115995e-06,
+      "loss": 1.026,
+      "step": 14285
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6397859736288937e-06,
+      "loss": 0.6982,
+      "step": 14286
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6396904261417927e-06,
+      "loss": 0.9459,
+      "step": 14287
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6395948786546914e-06,
+      "loss": 0.9421,
+      "step": 14288
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639499331167591e-06,
+      "loss": 1.0177,
+      "step": 14289
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6394037836804894e-06,
+      "loss": 0.875,
+      "step": 14290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6393082361933885e-06,
+      "loss": 0.7652,
+      "step": 14291
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639212688706287e-06,
+      "loss": 0.4641,
+      "step": 14292
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639117141219186e-06,
+      "loss": 1.1196,
+      "step": 14293
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.639021593732085e-06,
+      "loss": 1.0292,
+      "step": 14294
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6389260462449838e-06,
+      "loss": 0.9408,
+      "step": 14295
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6388304987578832e-06,
+      "loss": 0.8553,
+      "step": 14296
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.638734951270782e-06,
+      "loss": 1.056,
+      "step": 14297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.638639403783681e-06,
+      "loss": 0.8223,
+      "step": 14298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6385438562965795e-06,
+      "loss": 1.0767,
+      "step": 14299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6384483088094786e-06,
+      "loss": 0.5952,
+      "step": 14300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.638352761322377e-06,
+      "loss": 0.9632,
+      "step": 14301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.638257213835276e-06,
+      "loss": 1.2073,
+      "step": 14302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6381616663481757e-06,
+      "loss": 1.085,
+      "step": 14303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6380661188610743e-06,
+      "loss": 1.0249,
+      "step": 14304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6379705713739733e-06,
+      "loss": 1.3351,
+      "step": 14305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.637875023886872e-06,
+      "loss": 1.2449,
+      "step": 14306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.637779476399771e-06,
+      "loss": 1.1106,
+      "step": 14307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6376839289126696e-06,
+      "loss": 1.1335,
+      "step": 14308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6375883814255686e-06,
+      "loss": 0.7061,
+      "step": 14309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.637492833938468e-06,
+      "loss": 1.4054,
+      "step": 14310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6373972864513667e-06,
+      "loss": 0.9606,
+      "step": 14311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6373017389642657e-06,
+      "loss": 0.7848,
+      "step": 14312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6372061914771644e-06,
+      "loss": 1.1854,
+      "step": 14313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6371106439900634e-06,
+      "loss": 0.9581,
+      "step": 14314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.637015096502962e-06,
+      "loss": 1.0336,
+      "step": 14315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636919549015861e-06,
+      "loss": 0.757,
+      "step": 14316
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6368240015287605e-06,
+      "loss": 0.9892,
+      "step": 14317
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636728454041659e-06,
+      "loss": 0.3477,
+      "step": 14318
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636632906554558e-06,
+      "loss": 0.9871,
+      "step": 14319
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6365373590674568e-06,
+      "loss": 1.0004,
+      "step": 14320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636441811580356e-06,
+      "loss": 1.071,
+      "step": 14321
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6363462640932544e-06,
+      "loss": 0.6807,
+      "step": 14322
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6362507166061535e-06,
+      "loss": 0.7442,
+      "step": 14323
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636155169119052e-06,
+      "loss": 1.039,
+      "step": 14324
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6360596216319516e-06,
+      "loss": 1.2678,
+      "step": 14325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6359640741448506e-06,
+      "loss": 0.9835,
+      "step": 14326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.635868526657749e-06,
+      "loss": 1.4552,
+      "step": 14327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6357729791706482e-06,
+      "loss": 1.0427,
+      "step": 14328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.635677431683547e-06,
+      "loss": 1.1195,
+      "step": 14329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.635581884196446e-06,
+      "loss": 0.8908,
+      "step": 14330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6354863367093445e-06,
+      "loss": 1.2967,
+      "step": 14331
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.635390789222244e-06,
+      "loss": 0.739,
+      "step": 14332
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.635295241735143e-06,
+      "loss": 0.8941,
+      "step": 14333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6351996942480416e-06,
+      "loss": 1.1286,
+      "step": 14334
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6351041467609407e-06,
+      "loss": 0.864,
+      "step": 14335
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6350085992738393e-06,
+      "loss": 0.8615,
+      "step": 14336
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6349130517867383e-06,
+      "loss": 1.0675,
+      "step": 14337
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.634817504299637e-06,
+      "loss": 1.1388,
+      "step": 14338
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6347219568125364e-06,
+      "loss": 1.0385,
+      "step": 14339
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.634626409325435e-06,
+      "loss": 1.2222,
+      "step": 14340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.634530861838334e-06,
+      "loss": 0.8727,
+      "step": 14341
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6344353143512327e-06,
+      "loss": 1.2415,
+      "step": 14342
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6343397668641317e-06,
+      "loss": 1.0397,
+      "step": 14343
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6342442193770307e-06,
+      "loss": 1.0722,
+      "step": 14344
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6341486718899294e-06,
+      "loss": 1.1642,
+      "step": 14345
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.634053124402829e-06,
+      "loss": 1.5027,
+      "step": 14346
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6339575769157274e-06,
+      "loss": 0.9545,
+      "step": 14347
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6338620294286265e-06,
+      "loss": 0.9582,
+      "step": 14348
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.633766481941525e-06,
+      "loss": 1.0654,
+      "step": 14349
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.633670934454424e-06,
+      "loss": 0.8202,
+      "step": 14350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6335753869673227e-06,
+      "loss": 1.1615,
+      "step": 14351
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6334798394802218e-06,
+      "loss": 1.7758,
+      "step": 14352
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6333842919931212e-06,
+      "loss": 1.5085,
+      "step": 14353
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.63328874450602e-06,
+      "loss": 0.8374,
+      "step": 14354
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.633193197018919e-06,
+      "loss": 0.6846,
+      "step": 14355
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6330976495318175e-06,
+      "loss": 0.8049,
+      "step": 14356
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6330021020447166e-06,
+      "loss": 0.5552,
+      "step": 14357
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632906554557615e-06,
+      "loss": 0.9101,
+      "step": 14358
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632811007070514e-06,
+      "loss": 0.9704,
+      "step": 14359
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632715459583413e-06,
+      "loss": 1.0667,
+      "step": 14360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6326199120963123e-06,
+      "loss": 0.62,
+      "step": 14361
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6325243646092113e-06,
+      "loss": 0.9934,
+      "step": 14362
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.63242881712211e-06,
+      "loss": 1.1036,
+      "step": 14363
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632333269635009e-06,
+      "loss": 0.7508,
+      "step": 14364
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6322377221479076e-06,
+      "loss": 1.124,
+      "step": 14365
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6321421746608066e-06,
+      "loss": 0.792,
+      "step": 14366
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6320466271737052e-06,
+      "loss": 1.1625,
+      "step": 14367
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6319510796866047e-06,
+      "loss": 1.0034,
+      "step": 14368
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6318555321995037e-06,
+      "loss": 0.7578,
+      "step": 14369
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6317599847124024e-06,
+      "loss": 0.7634,
+      "step": 14370
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6316644372253014e-06,
+      "loss": 0.9783,
+      "step": 14371
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6315688897382e-06,
+      "loss": 1.0938,
+      "step": 14372
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631473342251099e-06,
+      "loss": 0.5445,
+      "step": 14373
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6313777947639977e-06,
+      "loss": 0.9438,
+      "step": 14374
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631282247276897e-06,
+      "loss": 0.7217,
+      "step": 14375
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631186699789796e-06,
+      "loss": 1.0199,
+      "step": 14376
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6310911523026948e-06,
+      "loss": 0.5912,
+      "step": 14377
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.630995604815594e-06,
+      "loss": 1.1291,
+      "step": 14378
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6309000573284924e-06,
+      "loss": 0.8965,
+      "step": 14379
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6308045098413915e-06,
+      "loss": 1.3308,
+      "step": 14380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.63070896235429e-06,
+      "loss": 0.9725,
+      "step": 14381
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6306134148671896e-06,
+      "loss": 0.8281,
+      "step": 14382
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6305178673800886e-06,
+      "loss": 0.8466,
+      "step": 14383
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630422319892987e-06,
+      "loss": 1.0424,
+      "step": 14384
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6303267724058862e-06,
+      "loss": 1.1152,
+      "step": 14385
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630231224918785e-06,
+      "loss": 0.9032,
+      "step": 14386
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630135677431684e-06,
+      "loss": 1.0934,
+      "step": 14387
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6300401299445825e-06,
+      "loss": 0.8852,
+      "step": 14388
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629944582457482e-06,
+      "loss": 0.7772,
+      "step": 14389
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6298490349703806e-06,
+      "loss": 0.7599,
+      "step": 14390
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6297534874832796e-06,
+      "loss": 1.2321,
+      "step": 14391
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6296579399961782e-06,
+      "loss": 0.9641,
+      "step": 14392
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6295623925090773e-06,
+      "loss": 0.4967,
+      "step": 14393
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629466845021976e-06,
+      "loss": 0.4227,
+      "step": 14394
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629371297534875e-06,
+      "loss": 1.3274,
+      "step": 14395
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629275750047774e-06,
+      "loss": 1.0138,
+      "step": 14396
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629180202560673e-06,
+      "loss": 0.6385,
+      "step": 14397
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629084655073572e-06,
+      "loss": 1.3389,
+      "step": 14398
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6289891075864707e-06,
+      "loss": 0.7161,
+      "step": 14399
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6288935600993697e-06,
+      "loss": 0.9293,
+      "step": 14400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6287980126122683e-06,
+      "loss": 0.8598,
+      "step": 14401
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6287024651251674e-06,
+      "loss": 1.2125,
+      "step": 14402
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.628606917638066e-06,
+      "loss": 1.3454,
+      "step": 14403
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6285113701509654e-06,
+      "loss": 1.1249,
+      "step": 14404
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6284158226638645e-06,
+      "loss": 1.0121,
+      "step": 14405
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.628320275176763e-06,
+      "loss": 0.7928,
+      "step": 14406
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.628224727689662e-06,
+      "loss": 1.2628,
+      "step": 14407
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6281291802025607e-06,
+      "loss": 0.9716,
+      "step": 14408
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6280336327154598e-06,
+      "loss": 0.7463,
+      "step": 14409
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6279380852283584e-06,
+      "loss": 0.8764,
+      "step": 14410
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627842537741258e-06,
+      "loss": 1.1974,
+      "step": 14411
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627746990254157e-06,
+      "loss": 0.7415,
+      "step": 14412
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6276514427670555e-06,
+      "loss": 1.0172,
+      "step": 14413
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6275558952799546e-06,
+      "loss": 0.8786,
+      "step": 14414
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627460347792853e-06,
+      "loss": 1.0488,
+      "step": 14415
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627364800305752e-06,
+      "loss": 1.3036,
+      "step": 14416
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627269252818651e-06,
+      "loss": 1.1486,
+      "step": 14417
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6271737053315503e-06,
+      "loss": 1.257,
+      "step": 14418
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6270781578444493e-06,
+      "loss": 0.5712,
+      "step": 14419
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626982610357348e-06,
+      "loss": 0.8203,
+      "step": 14420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626887062870247e-06,
+      "loss": 1.069,
+      "step": 14421
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6267915153831456e-06,
+      "loss": 1.0964,
+      "step": 14422
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6266959678960446e-06,
+      "loss": 1.2956,
+      "step": 14423
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6266004204089432e-06,
+      "loss": 1.0625,
+      "step": 14424
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6265048729218427e-06,
+      "loss": 1.0735,
+      "step": 14425
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6264093254347417e-06,
+      "loss": 1.0693,
+      "step": 14426
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6263137779476404e-06,
+      "loss": 1.3887,
+      "step": 14427
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6262182304605394e-06,
+      "loss": 0.8346,
+      "step": 14428
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626122682973438e-06,
+      "loss": 1.0474,
+      "step": 14429
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626027135486337e-06,
+      "loss": 0.986,
+      "step": 14430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6259315879992357e-06,
+      "loss": 0.7504,
+      "step": 14431
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6258360405121347e-06,
+      "loss": 1.2805,
+      "step": 14432
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6257404930250337e-06,
+      "loss": 0.9185,
+      "step": 14433
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6256449455379328e-06,
+      "loss": 1.0831,
+      "step": 14434
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.625549398050832e-06,
+      "loss": 0.711,
+      "step": 14435
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6254538505637304e-06,
+      "loss": 0.981,
+      "step": 14436
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6253583030766295e-06,
+      "loss": 0.8396,
+      "step": 14437
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.625262755589528e-06,
+      "loss": 0.8496,
+      "step": 14438
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.625167208102427e-06,
+      "loss": 0.9864,
+      "step": 14439
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.625071660615326e-06,
+      "loss": 1.1702,
+      "step": 14440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.624976113128225e-06,
+      "loss": 0.7045,
+      "step": 14441
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.624880565641124e-06,
+      "loss": 0.9636,
+      "step": 14442
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.624785018154023e-06,
+      "loss": 0.752,
+      "step": 14443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6246894706669215e-06,
+      "loss": 0.6697,
+      "step": 14444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6245939231798205e-06,
+      "loss": 1.1804,
+      "step": 14445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6244983756927196e-06,
+      "loss": 1.2698,
+      "step": 14446
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6244028282056186e-06,
+      "loss": 0.4559,
+      "step": 14447
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6243072807185176e-06,
+      "loss": 0.9242,
+      "step": 14448
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6242117332314162e-06,
+      "loss": 0.9753,
+      "step": 14449
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6241161857443153e-06,
+      "loss": 1.1921,
+      "step": 14450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.624020638257214e-06,
+      "loss": 1.0264,
+      "step": 14451
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.623925090770113e-06,
+      "loss": 1.3519,
+      "step": 14452
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6238295432830115e-06,
+      "loss": 0.4367,
+      "step": 14453
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.623733995795911e-06,
+      "loss": 0.8135,
+      "step": 14454
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.62363844830881e-06,
+      "loss": 0.8769,
+      "step": 14455
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6235429008217087e-06,
+      "loss": 0.4947,
+      "step": 14456
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6234473533346077e-06,
+      "loss": 0.7972,
+      "step": 14457
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6233518058475063e-06,
+      "loss": 1.5729,
+      "step": 14458
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6232562583604054e-06,
+      "loss": 0.8552,
+      "step": 14459
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.623160710873304e-06,
+      "loss": 0.4599,
+      "step": 14460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6230651633862034e-06,
+      "loss": 1.2893,
+      "step": 14461
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6229696158991025e-06,
+      "loss": 0.8287,
+      "step": 14462
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622874068412001e-06,
+      "loss": 1.0202,
+      "step": 14463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6227785209249e-06,
+      "loss": 1.1376,
+      "step": 14464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6226829734377987e-06,
+      "loss": 1.1085,
+      "step": 14465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6225874259506978e-06,
+      "loss": 1.1024,
+      "step": 14466
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6224918784635964e-06,
+      "loss": 1.0064,
+      "step": 14467
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6223963309764954e-06,
+      "loss": 0.6147,
+      "step": 14468
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622300783489395e-06,
+      "loss": 0.9093,
+      "step": 14469
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6222052360022935e-06,
+      "loss": 0.4722,
+      "step": 14470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6221096885151926e-06,
+      "loss": 0.6281,
+      "step": 14471
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622014141028091e-06,
+      "loss": 0.7884,
+      "step": 14472
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.62191859354099e-06,
+      "loss": 0.8725,
+      "step": 14473
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621823046053889e-06,
+      "loss": 1.1676,
+      "step": 14474
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621727498566788e-06,
+      "loss": 0.7199,
+      "step": 14475
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6216319510796873e-06,
+      "loss": 1.0653,
+      "step": 14476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621536403592586e-06,
+      "loss": 0.841,
+      "step": 14477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621440856105485e-06,
+      "loss": 0.8096,
+      "step": 14478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6213453086183836e-06,
+      "loss": 1.1848,
+      "step": 14479
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6212497611312826e-06,
+      "loss": 1.0738,
+      "step": 14480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6211542136441812e-06,
+      "loss": 1.1278,
+      "step": 14481
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6210586661570803e-06,
+      "loss": 0.675,
+      "step": 14482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6209631186699793e-06,
+      "loss": 0.811,
+      "step": 14483
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6208675711828784e-06,
+      "loss": 1.4556,
+      "step": 14484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.620772023695777e-06,
+      "loss": 1.6473,
+      "step": 14485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.620676476208676e-06,
+      "loss": 0.9614,
+      "step": 14486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.620580928721575e-06,
+      "loss": 1.2379,
+      "step": 14487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6204853812344737e-06,
+      "loss": 1.1333,
+      "step": 14488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6203898337473727e-06,
+      "loss": 0.6635,
+      "step": 14489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6202942862602717e-06,
+      "loss": 1.1884,
+      "step": 14490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6201987387731708e-06,
+      "loss": 0.9292,
+      "step": 14491
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6201031912860694e-06,
+      "loss": 0.9986,
+      "step": 14492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6200076437989684e-06,
+      "loss": 0.9351,
+      "step": 14493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619912096311867e-06,
+      "loss": 0.9245,
+      "step": 14494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619816548824766e-06,
+      "loss": 0.7658,
+      "step": 14495
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6197210013376647e-06,
+      "loss": 0.9782,
+      "step": 14496
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619625453850564e-06,
+      "loss": 0.9513,
+      "step": 14497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619529906363463e-06,
+      "loss": 1.4544,
+      "step": 14498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619434358876362e-06,
+      "loss": 1.0957,
+      "step": 14499
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619338811389261e-06,
+      "loss": 1.203,
+      "step": 14500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6192432639021595e-06,
+      "loss": 0.6961,
+      "step": 14501
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6191477164150585e-06,
+      "loss": 0.9789,
+      "step": 14502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.619052168927957e-06,
+      "loss": 0.767,
+      "step": 14503
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.618956621440856e-06,
+      "loss": 1.044,
+      "step": 14504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6188610739537556e-06,
+      "loss": 1.0066,
+      "step": 14505
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6187655264666542e-06,
+      "loss": 0.8487,
+      "step": 14506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6186699789795533e-06,
+      "loss": 1.1433,
+      "step": 14507
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.618574431492452e-06,
+      "loss": 0.8889,
+      "step": 14508
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.618478884005351e-06,
+      "loss": 1.016,
+      "step": 14509
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6183833365182495e-06,
+      "loss": 1.2085,
+      "step": 14510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6182877890311486e-06,
+      "loss": 0.6285,
+      "step": 14511
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.618192241544048e-06,
+      "loss": 1.0333,
+      "step": 14512
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6180966940569467e-06,
+      "loss": 1.2126,
+      "step": 14513
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6180011465698457e-06,
+      "loss": 0.9537,
+      "step": 14514
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6179055990827443e-06,
+      "loss": 1.1165,
+      "step": 14515
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6178100515956434e-06,
+      "loss": 0.7472,
+      "step": 14516
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617714504108542e-06,
+      "loss": 1.561,
+      "step": 14517
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617618956621441e-06,
+      "loss": 1.2526,
+      "step": 14518
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6175234091343405e-06,
+      "loss": 1.0771,
+      "step": 14519
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617427861647239e-06,
+      "loss": 0.9044,
+      "step": 14520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617332314160138e-06,
+      "loss": 1.0618,
+      "step": 14521
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6172367666730367e-06,
+      "loss": 1.0614,
+      "step": 14522
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6171412191859358e-06,
+      "loss": 0.8528,
+      "step": 14523
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6170456716988344e-06,
+      "loss": 1.1487,
+      "step": 14524
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6169501242117334e-06,
+      "loss": 1.1518,
+      "step": 14525
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616854576724633e-06,
+      "loss": 0.8439,
+      "step": 14526
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6167590292375315e-06,
+      "loss": 1.2154,
+      "step": 14527
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6166634817504305e-06,
+      "loss": 1.0237,
+      "step": 14528
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616567934263329e-06,
+      "loss": 1.0374,
+      "step": 14529
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616472386776228e-06,
+      "loss": 0.9985,
+      "step": 14530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616376839289127e-06,
+      "loss": 0.4228,
+      "step": 14531
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616281291802026e-06,
+      "loss": 0.9713,
+      "step": 14532
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616185744314925e-06,
+      "loss": 0.7944,
+      "step": 14533
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616090196827824e-06,
+      "loss": 1.0758,
+      "step": 14534
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6159946493407225e-06,
+      "loss": 1.0456,
+      "step": 14535
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6158991018536216e-06,
+      "loss": 1.202,
+      "step": 14536
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6158035543665206e-06,
+      "loss": 1.3721,
+      "step": 14537
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6157080068794192e-06,
+      "loss": 0.7405,
+      "step": 14538
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6156124593923183e-06,
+      "loss": 1.1941,
+      "step": 14539
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.615516911905217e-06,
+      "loss": 0.7606,
+      "step": 14540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6154213644181164e-06,
+      "loss": 1.0691,
+      "step": 14541
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.615325816931015e-06,
+      "loss": 0.9196,
+      "step": 14542
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.615230269443914e-06,
+      "loss": 1.237,
+      "step": 14543
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6151347219568126e-06,
+      "loss": 1.2395,
+      "step": 14544
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6150391744697117e-06,
+      "loss": 1.0809,
+      "step": 14545
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6149436269826103e-06,
+      "loss": 0.9202,
+      "step": 14546
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6148480794955093e-06,
+      "loss": 0.9609,
+      "step": 14547
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6147525320084088e-06,
+      "loss": 0.7009,
+      "step": 14548
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6146569845213074e-06,
+      "loss": 0.8655,
+      "step": 14549
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6145614370342064e-06,
+      "loss": 0.6941,
+      "step": 14550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.614465889547105e-06,
+      "loss": 0.8218,
+      "step": 14551
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.614370342060004e-06,
+      "loss": 0.7257,
+      "step": 14552
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6142747945729027e-06,
+      "loss": 0.8718,
+      "step": 14553
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6141792470858017e-06,
+      "loss": 0.8781,
+      "step": 14554
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.614083699598701e-06,
+      "loss": 1.0055,
+      "step": 14555
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6139881521116e-06,
+      "loss": 1.0163,
+      "step": 14556
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.613892604624499e-06,
+      "loss": 1.024,
+      "step": 14557
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6137970571373975e-06,
+      "loss": 0.8089,
+      "step": 14558
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6137015096502965e-06,
+      "loss": 0.8378,
+      "step": 14559
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.613605962163195e-06,
+      "loss": 0.81,
+      "step": 14560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.613510414676094e-06,
+      "loss": 0.8797,
+      "step": 14561
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6134148671889936e-06,
+      "loss": 1.0485,
+      "step": 14562
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6133193197018922e-06,
+      "loss": 1.0822,
+      "step": 14563
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6132237722147913e-06,
+      "loss": 1.1355,
+      "step": 14564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.61312822472769e-06,
+      "loss": 1.3084,
+      "step": 14565
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.613032677240589e-06,
+      "loss": 0.9562,
+      "step": 14566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6129371297534875e-06,
+      "loss": 0.7745,
+      "step": 14567
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6128415822663866e-06,
+      "loss": 1.1573,
+      "step": 14568
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.612746034779286e-06,
+      "loss": 0.8126,
+      "step": 14569
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6126504872921847e-06,
+      "loss": 0.843,
+      "step": 14570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6125549398050837e-06,
+      "loss": 1.1711,
+      "step": 14571
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6124593923179823e-06,
+      "loss": 0.9393,
+      "step": 14572
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6123638448308814e-06,
+      "loss": 0.6906,
+      "step": 14573
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.61226829734378e-06,
+      "loss": 1.1762,
+      "step": 14574
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.612172749856679e-06,
+      "loss": 0.809,
+      "step": 14575
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6120772023695776e-06,
+      "loss": 1.0079,
+      "step": 14576
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611981654882477e-06,
+      "loss": 1.4603,
+      "step": 14577
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611886107395376e-06,
+      "loss": 0.971,
+      "step": 14578
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6117905599082747e-06,
+      "loss": 1.059,
+      "step": 14579
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6116950124211738e-06,
+      "loss": 0.7035,
+      "step": 14580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6115994649340724e-06,
+      "loss": 1.3341,
+      "step": 14581
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6115039174469714e-06,
+      "loss": 1.127,
+      "step": 14582
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.61140836995987e-06,
+      "loss": 0.6548,
+      "step": 14583
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6113128224727695e-06,
+      "loss": 0.793,
+      "step": 14584
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611217274985668e-06,
+      "loss": 0.629,
+      "step": 14585
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611121727498567e-06,
+      "loss": 0.7078,
+      "step": 14586
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6110261800114658e-06,
+      "loss": 1.4109,
+      "step": 14587
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610930632524365e-06,
+      "loss": 0.9636,
+      "step": 14588
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610835085037264e-06,
+      "loss": 0.8415,
+      "step": 14589
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6107395375501625e-06,
+      "loss": 1.1597,
+      "step": 14590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610643990063062e-06,
+      "loss": 1.3938,
+      "step": 14591
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6105484425759605e-06,
+      "loss": 0.9832,
+      "step": 14592
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6104528950888596e-06,
+      "loss": 1.0164,
+      "step": 14593
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610357347601758e-06,
+      "loss": 0.8989,
+      "step": 14594
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6102618001146572e-06,
+      "loss": 0.5841,
+      "step": 14595
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610166252627556e-06,
+      "loss": 1.1979,
+      "step": 14596
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.610070705140455e-06,
+      "loss": 1.2373,
+      "step": 14597
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6099751576533544e-06,
+      "loss": 1.3941,
+      "step": 14598
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.609879610166253e-06,
+      "loss": 0.9186,
+      "step": 14599
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.609784062679152e-06,
+      "loss": 1.123,
+      "step": 14600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6096885151920506e-06,
+      "loss": 0.6399,
+      "step": 14601
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6095929677049497e-06,
+      "loss": 1.1591,
+      "step": 14602
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6094974202178483e-06,
+      "loss": 1.006,
+      "step": 14603
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6094018727307473e-06,
+      "loss": 1.1384,
+      "step": 14604
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6093063252436468e-06,
+      "loss": 1.2836,
+      "step": 14605
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6092107777565454e-06,
+      "loss": 1.1105,
+      "step": 14606
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6091152302694444e-06,
+      "loss": 0.8155,
+      "step": 14607
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.609019682782343e-06,
+      "loss": 1.1839,
+      "step": 14608
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.608924135295242e-06,
+      "loss": 1.0167,
+      "step": 14609
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6088285878081407e-06,
+      "loss": 0.9224,
+      "step": 14610
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6087330403210397e-06,
+      "loss": 0.8762,
+      "step": 14611
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6086374928339384e-06,
+      "loss": 1.1552,
+      "step": 14612
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.608541945346838e-06,
+      "loss": 1.5324,
+      "step": 14613
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.608446397859737e-06,
+      "loss": 0.8165,
+      "step": 14614
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6083508503726355e-06,
+      "loss": 0.7997,
+      "step": 14615
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6082553028855345e-06,
+      "loss": 1.2227,
+      "step": 14616
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.608159755398433e-06,
+      "loss": 0.8985,
+      "step": 14617
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.608064207911332e-06,
+      "loss": 0.9428,
+      "step": 14618
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6079686604242308e-06,
+      "loss": 1.2973,
+      "step": 14619
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6078731129371302e-06,
+      "loss": 1.4709,
+      "step": 14620
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6077775654500293e-06,
+      "loss": 0.9497,
+      "step": 14621
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.607682017962928e-06,
+      "loss": 0.8222,
+      "step": 14622
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.607586470475827e-06,
+      "loss": 0.7293,
+      "step": 14623
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6074909229887255e-06,
+      "loss": 0.7737,
+      "step": 14624
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6073953755016246e-06,
+      "loss": 1.31,
+      "step": 14625
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.607299828014523e-06,
+      "loss": 1.1722,
+      "step": 14626
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6072042805274227e-06,
+      "loss": 0.9768,
+      "step": 14627
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6071087330403217e-06,
+      "loss": 0.7679,
+      "step": 14628
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6070131855532203e-06,
+      "loss": 1.1231,
+      "step": 14629
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6069176380661194e-06,
+      "loss": 0.991,
+      "step": 14630
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606822090579018e-06,
+      "loss": 0.7929,
+      "step": 14631
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606726543091917e-06,
+      "loss": 0.934,
+      "step": 14632
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6066309956048156e-06,
+      "loss": 1.1937,
+      "step": 14633
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.606535448117715e-06,
+      "loss": 1.2933,
+      "step": 14634
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6064399006306137e-06,
+      "loss": 0.9067,
+      "step": 14635
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6063443531435127e-06,
+      "loss": 0.6751,
+      "step": 14636
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6062488056564114e-06,
+      "loss": 1.1729,
+      "step": 14637
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6061532581693104e-06,
+      "loss": 0.8065,
+      "step": 14638
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6060577106822094e-06,
+      "loss": 1.0103,
+      "step": 14639
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.605962163195108e-06,
+      "loss": 1.0181,
+      "step": 14640
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6058666157080075e-06,
+      "loss": 0.8486,
+      "step": 14641
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.605771068220906e-06,
+      "loss": 0.927,
+      "step": 14642
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.605675520733805e-06,
+      "loss": 1.1019,
+      "step": 14643
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6055799732467038e-06,
+      "loss": 0.772,
+      "step": 14644
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.605484425759603e-06,
+      "loss": 0.9965,
+      "step": 14645
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6053888782725014e-06,
+      "loss": 1.6562,
+      "step": 14646
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6052933307854005e-06,
+      "loss": 0.6656,
+      "step": 14647
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.605197783298299e-06,
+      "loss": 0.8616,
+      "step": 14648
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6051022358111985e-06,
+      "loss": 1.1076,
+      "step": 14649
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6050066883240976e-06,
+      "loss": 0.7725,
+      "step": 14650
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.604911140836996e-06,
+      "loss": 0.7842,
+      "step": 14651
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6048155933498952e-06,
+      "loss": 0.8012,
+      "step": 14652
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.604720045862794e-06,
+      "loss": 0.8265,
+      "step": 14653
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.604624498375693e-06,
+      "loss": 1.0986,
+      "step": 14654
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6045289508885915e-06,
+      "loss": 0.8567,
+      "step": 14655
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.604433403401491e-06,
+      "loss": 1.23,
+      "step": 14656
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60433785591439e-06,
+      "loss": 1.0787,
+      "step": 14657
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6042423084272886e-06,
+      "loss": 1.0796,
+      "step": 14658
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6041467609401877e-06,
+      "loss": 0.5175,
+      "step": 14659
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6040512134530863e-06,
+      "loss": 0.5668,
+      "step": 14660
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6039556659659853e-06,
+      "loss": 0.6556,
+      "step": 14661
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.603860118478884e-06,
+      "loss": 0.7103,
+      "step": 14662
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6037645709917834e-06,
+      "loss": 0.9053,
+      "step": 14663
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6036690235046824e-06,
+      "loss": 1.558,
+      "step": 14664
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.603573476017581e-06,
+      "loss": 0.9929,
+      "step": 14665
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60347792853048e-06,
+      "loss": 0.7615,
+      "step": 14666
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6033823810433787e-06,
+      "loss": 0.9223,
+      "step": 14667
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6032868335562777e-06,
+      "loss": 0.7877,
+      "step": 14668
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6031912860691764e-06,
+      "loss": 0.8027,
+      "step": 14669
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.603095738582076e-06,
+      "loss": 0.8709,
+      "step": 14670
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.603000191094975e-06,
+      "loss": 1.0018,
+      "step": 14671
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6029046436078735e-06,
+      "loss": 1.0526,
+      "step": 14672
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6028090961207725e-06,
+      "loss": 1.1739,
+      "step": 14673
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602713548633671e-06,
+      "loss": 0.9239,
+      "step": 14674
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.60261800114657e-06,
+      "loss": 0.7123,
+      "step": 14675
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6025224536594688e-06,
+      "loss": 1.0784,
+      "step": 14676
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6024269061723682e-06,
+      "loss": 1.1185,
+      "step": 14677
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6023313586852673e-06,
+      "loss": 1.0492,
+      "step": 14678
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602235811198166e-06,
+      "loss": 0.9296,
+      "step": 14679
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602140263711065e-06,
+      "loss": 1.1279,
+      "step": 14680
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6020447162239635e-06,
+      "loss": 1.4385,
+      "step": 14681
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6019491687368626e-06,
+      "loss": 1.1041,
+      "step": 14682
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.601853621249761e-06,
+      "loss": 1.1733,
+      "step": 14683
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6017580737626602e-06,
+      "loss": 0.6308,
+      "step": 14684
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6016625262755593e-06,
+      "loss": 1.16,
+      "step": 14685
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6015669787884583e-06,
+      "loss": 0.8543,
+      "step": 14686
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.601471431301357e-06,
+      "loss": 1.5007,
+      "step": 14687
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.601375883814256e-06,
+      "loss": 0.8442,
+      "step": 14688
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6012803363271546e-06,
+      "loss": 1.21,
+      "step": 14689
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6011847888400536e-06,
+      "loss": 0.9062,
+      "step": 14690
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6010892413529527e-06,
+      "loss": 1.1377,
+      "step": 14691
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6009936938658517e-06,
+      "loss": 0.914,
+      "step": 14692
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6008981463787507e-06,
+      "loss": 1.178,
+      "step": 14693
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6008025988916494e-06,
+      "loss": 0.6215,
+      "step": 14694
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6007070514045484e-06,
+      "loss": 0.8712,
+      "step": 14695
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600611503917447e-06,
+      "loss": 0.7872,
+      "step": 14696
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600515956430346e-06,
+      "loss": 0.6547,
+      "step": 14697
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6004204089432447e-06,
+      "loss": 0.8338,
+      "step": 14698
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600324861456144e-06,
+      "loss": 0.922,
+      "step": 14699
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600229313969043e-06,
+      "loss": 0.7871,
+      "step": 14700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6001337664819418e-06,
+      "loss": 0.7836,
+      "step": 14701
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.600038218994841e-06,
+      "loss": 0.9647,
+      "step": 14702
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5999426715077394e-06,
+      "loss": 0.6529,
+      "step": 14703
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5998471240206385e-06,
+      "loss": 1.18,
+      "step": 14704
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.599751576533537e-06,
+      "loss": 0.593,
+      "step": 14705
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5996560290464365e-06,
+      "loss": 0.9359,
+      "step": 14706
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5995604815593356e-06,
+      "loss": 1.0554,
+      "step": 14707
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.599464934072234e-06,
+      "loss": 1.263,
+      "step": 14708
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5993693865851332e-06,
+      "loss": 1.2764,
+      "step": 14709
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.599273839098032e-06,
+      "loss": 1.2947,
+      "step": 14710
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.599178291610931e-06,
+      "loss": 0.7218,
+      "step": 14711
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5990827441238295e-06,
+      "loss": 0.9008,
+      "step": 14712
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598987196636729e-06,
+      "loss": 0.9064,
+      "step": 14713
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598891649149628e-06,
+      "loss": 0.9998,
+      "step": 14714
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5987961016625266e-06,
+      "loss": 1.1265,
+      "step": 14715
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5987005541754257e-06,
+      "loss": 1.2501,
+      "step": 14716
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5986050066883243e-06,
+      "loss": 0.8469,
+      "step": 14717
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5985094592012233e-06,
+      "loss": 0.8366,
+      "step": 14718
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598413911714122e-06,
+      "loss": 0.9805,
+      "step": 14719
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598318364227021e-06,
+      "loss": 1.1084,
+      "step": 14720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5982228167399204e-06,
+      "loss": 1.1319,
+      "step": 14721
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598127269252819e-06,
+      "loss": 0.7817,
+      "step": 14722
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.598031721765718e-06,
+      "loss": 1.4764,
+      "step": 14723
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5979361742786167e-06,
+      "loss": 1.1101,
+      "step": 14724
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5978406267915157e-06,
+      "loss": 1.4386,
+      "step": 14725
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5977450793044144e-06,
+      "loss": 1.6198,
+      "step": 14726
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5976495318173134e-06,
+      "loss": 0.8204,
+      "step": 14727
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5975539843302124e-06,
+      "loss": 1.0156,
+      "step": 14728
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5974584368431115e-06,
+      "loss": 0.884,
+      "step": 14729
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5973628893560105e-06,
+      "loss": 0.9228,
+      "step": 14730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.597267341868909e-06,
+      "loss": 0.7127,
+      "step": 14731
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.597171794381808e-06,
+      "loss": 0.5573,
+      "step": 14732
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5970762468947068e-06,
+      "loss": 1.0309,
+      "step": 14733
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596980699407606e-06,
+      "loss": 0.7366,
+      "step": 14734
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596885151920505e-06,
+      "loss": 1.3353,
+      "step": 14735
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596789604433404e-06,
+      "loss": 1.1647,
+      "step": 14736
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5966940569463025e-06,
+      "loss": 0.6071,
+      "step": 14737
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5965985094592015e-06,
+      "loss": 0.8309,
+      "step": 14738
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5965029619721e-06,
+      "loss": 0.696,
+      "step": 14739
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596407414484999e-06,
+      "loss": 1.3605,
+      "step": 14740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5963118669978982e-06,
+      "loss": 1.2989,
+      "step": 14741
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5962163195107973e-06,
+      "loss": 0.8511,
+      "step": 14742
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5961207720236963e-06,
+      "loss": 1.0749,
+      "step": 14743
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596025224536595e-06,
+      "loss": 0.6529,
+      "step": 14744
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.595929677049494e-06,
+      "loss": 0.6833,
+      "step": 14745
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5958341295623926e-06,
+      "loss": 1.0143,
+      "step": 14746
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5957385820752916e-06,
+      "loss": 0.8743,
+      "step": 14747
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5956430345881902e-06,
+      "loss": 1.1675,
+      "step": 14748
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5955474871010897e-06,
+      "loss": 0.6449,
+      "step": 14749
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5954519396139887e-06,
+      "loss": 0.8946,
+      "step": 14750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5953563921268874e-06,
+      "loss": 1.1548,
+      "step": 14751
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5952608446397864e-06,
+      "loss": 0.6981,
+      "step": 14752
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.595165297152685e-06,
+      "loss": 0.9484,
+      "step": 14753
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.595069749665584e-06,
+      "loss": 1.3036,
+      "step": 14754
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5949742021784827e-06,
+      "loss": 0.9451,
+      "step": 14755
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5948786546913817e-06,
+      "loss": 0.7984,
+      "step": 14756
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.594783107204281e-06,
+      "loss": 0.9042,
+      "step": 14757
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5946875597171798e-06,
+      "loss": 1.2269,
+      "step": 14758
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.594592012230079e-06,
+      "loss": 1.1111,
+      "step": 14759
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5944964647429774e-06,
+      "loss": 1.1401,
+      "step": 14760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5944009172558765e-06,
+      "loss": 0.8687,
+      "step": 14761
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.594305369768775e-06,
+      "loss": 1.5614,
+      "step": 14762
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.594209822281674e-06,
+      "loss": 1.0587,
+      "step": 14763
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5941142747945736e-06,
+      "loss": 0.5722,
+      "step": 14764
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.594018727307472e-06,
+      "loss": 1.3173,
+      "step": 14765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5939231798203712e-06,
+      "loss": 0.7745,
+      "step": 14766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59382763233327e-06,
+      "loss": 0.9076,
+      "step": 14767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.593732084846169e-06,
+      "loss": 1.1851,
+      "step": 14768
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5936365373590675e-06,
+      "loss": 0.9772,
+      "step": 14769
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5935409898719665e-06,
+      "loss": 1.0395,
+      "step": 14770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.593445442384866e-06,
+      "loss": 0.9011,
+      "step": 14771
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5933498948977646e-06,
+      "loss": 1.2492,
+      "step": 14772
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5932543474106637e-06,
+      "loss": 0.7331,
+      "step": 14773
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5931587999235623e-06,
+      "loss": 1.0039,
+      "step": 14774
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5930632524364613e-06,
+      "loss": 1.2488,
+      "step": 14775
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.59296770494936e-06,
+      "loss": 1.3858,
+      "step": 14776
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.592872157462259e-06,
+      "loss": 1.1479,
+      "step": 14777
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.592776609975158e-06,
+      "loss": 1.2141,
+      "step": 14778
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.592681062488057e-06,
+      "loss": 0.5531,
+      "step": 14779
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5925855150009557e-06,
+      "loss": 1.153,
+      "step": 14780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5924899675138547e-06,
+      "loss": 0.7972,
+      "step": 14781
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5923944200267537e-06,
+      "loss": 0.9803,
+      "step": 14782
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5922988725396524e-06,
+      "loss": 0.806,
+      "step": 14783
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5922033250525514e-06,
+      "loss": 0.9377,
+      "step": 14784
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5921077775654504e-06,
+      "loss": 1.006,
+      "step": 14785
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5920122300783495e-06,
+      "loss": 0.6818,
+      "step": 14786
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591916682591248e-06,
+      "loss": 1.4698,
+      "step": 14787
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591821135104147e-06,
+      "loss": 0.796,
+      "step": 14788
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5917255876170457e-06,
+      "loss": 1.1284,
+      "step": 14789
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5916300401299448e-06,
+      "loss": 1.5679,
+      "step": 14790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5915344926428434e-06,
+      "loss": 0.8094,
+      "step": 14791
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5914389451557424e-06,
+      "loss": 0.7896,
+      "step": 14792
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591343397668642e-06,
+      "loss": 0.9932,
+      "step": 14793
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5912478501815405e-06,
+      "loss": 1.3544,
+      "step": 14794
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5911523026944395e-06,
+      "loss": 1.0339,
+      "step": 14795
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591056755207338e-06,
+      "loss": 0.6698,
+      "step": 14796
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.590961207720237e-06,
+      "loss": 0.5174,
+      "step": 14797
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.590865660233136e-06,
+      "loss": 1.1382,
+      "step": 14798
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.590770112746035e-06,
+      "loss": 0.8429,
+      "step": 14799
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5906745652589343e-06,
+      "loss": 0.5366,
+      "step": 14800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.590579017771833e-06,
+      "loss": 0.8908,
+      "step": 14801
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.590483470284732e-06,
+      "loss": 0.9539,
+      "step": 14802
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5903879227976306e-06,
+      "loss": 0.6777,
+      "step": 14803
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5902923753105296e-06,
+      "loss": 0.673,
+      "step": 14804
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5901968278234282e-06,
+      "loss": 0.6805,
+      "step": 14805
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5901012803363273e-06,
+      "loss": 0.7235,
+      "step": 14806
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5900057328492267e-06,
+      "loss": 1.2896,
+      "step": 14807
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5899101853621254e-06,
+      "loss": 1.0018,
+      "step": 14808
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5898146378750244e-06,
+      "loss": 0.8713,
+      "step": 14809
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.589719090387923e-06,
+      "loss": 0.86,
+      "step": 14810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.589623542900822e-06,
+      "loss": 0.971,
+      "step": 14811
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5895279954137207e-06,
+      "loss": 0.8735,
+      "step": 14812
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5894324479266197e-06,
+      "loss": 0.8471,
+      "step": 14813
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.589336900439519e-06,
+      "loss": 0.9138,
+      "step": 14814
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5892413529524178e-06,
+      "loss": 0.9443,
+      "step": 14815
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.589145805465317e-06,
+      "loss": 0.9411,
+      "step": 14816
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5890502579782154e-06,
+      "loss": 0.9587,
+      "step": 14817
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5889547104911145e-06,
+      "loss": 1.2721,
+      "step": 14818
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.588859163004013e-06,
+      "loss": 0.4426,
+      "step": 14819
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.588763615516912e-06,
+      "loss": 0.9267,
+      "step": 14820
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5886680680298116e-06,
+      "loss": 0.9915,
+      "step": 14821
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.58857252054271e-06,
+      "loss": 0.8404,
+      "step": 14822
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5884769730556092e-06,
+      "loss": 1.3047,
+      "step": 14823
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.588381425568508e-06,
+      "loss": 0.6926,
+      "step": 14824
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.588285878081407e-06,
+      "loss": 0.9353,
+      "step": 14825
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5881903305943055e-06,
+      "loss": 1.1122,
+      "step": 14826
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5880947831072045e-06,
+      "loss": 0.9103,
+      "step": 14827
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587999235620103e-06,
+      "loss": 0.9094,
+      "step": 14828
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5879036881330026e-06,
+      "loss": 0.8602,
+      "step": 14829
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5878081406459012e-06,
+      "loss": 1.3084,
+      "step": 14830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5877125931588003e-06,
+      "loss": 0.7017,
+      "step": 14831
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5876170456716993e-06,
+      "loss": 0.6626,
+      "step": 14832
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587521498184598e-06,
+      "loss": 0.8438,
+      "step": 14833
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587425950697497e-06,
+      "loss": 0.9404,
+      "step": 14834
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5873304032103956e-06,
+      "loss": 1.1539,
+      "step": 14835
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587234855723295e-06,
+      "loss": 1.5068,
+      "step": 14836
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5871393082361937e-06,
+      "loss": 0.8254,
+      "step": 14837
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5870437607490927e-06,
+      "loss": 0.7417,
+      "step": 14838
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5869482132619913e-06,
+      "loss": 1.0778,
+      "step": 14839
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5868526657748904e-06,
+      "loss": 1.0798,
+      "step": 14840
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586757118287789e-06,
+      "loss": 1.1796,
+      "step": 14841
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586661570800688e-06,
+      "loss": 0.9955,
+      "step": 14842
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5865660233135875e-06,
+      "loss": 0.9608,
+      "step": 14843
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586470475826486e-06,
+      "loss": 0.9526,
+      "step": 14844
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586374928339385e-06,
+      "loss": 0.8101,
+      "step": 14845
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5862793808522837e-06,
+      "loss": 1.2452,
+      "step": 14846
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5861838333651828e-06,
+      "loss": 1.0998,
+      "step": 14847
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5860882858780814e-06,
+      "loss": 1.1783,
+      "step": 14848
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5859927383909804e-06,
+      "loss": 1.1433,
+      "step": 14849
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58589719090388e-06,
+      "loss": 1.1403,
+      "step": 14850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5858016434167785e-06,
+      "loss": 0.6672,
+      "step": 14851
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5857060959296775e-06,
+      "loss": 1.2531,
+      "step": 14852
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585610548442576e-06,
+      "loss": 0.8988,
+      "step": 14853
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585515000955475e-06,
+      "loss": 0.8099,
+      "step": 14854
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585419453468374e-06,
+      "loss": 1.2325,
+      "step": 14855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585323905981273e-06,
+      "loss": 1.0358,
+      "step": 14856
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5852283584941723e-06,
+      "loss": 1.0501,
+      "step": 14857
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.585132811007071e-06,
+      "loss": 0.7551,
+      "step": 14858
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58503726351997e-06,
+      "loss": 1.2589,
+      "step": 14859
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5849417160328686e-06,
+      "loss": 0.8533,
+      "step": 14860
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5848461685457676e-06,
+      "loss": 1.1628,
+      "step": 14861
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5847506210586662e-06,
+      "loss": 0.8182,
+      "step": 14862
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5846550735715653e-06,
+      "loss": 0.6284,
+      "step": 14863
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.584559526084464e-06,
+      "loss": 0.8648,
+      "step": 14864
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5844639785973633e-06,
+      "loss": 0.6968,
+      "step": 14865
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5843684311102624e-06,
+      "loss": 0.9905,
+      "step": 14866
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.584272883623161e-06,
+      "loss": 0.6691,
+      "step": 14867
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58417733613606e-06,
+      "loss": 1.0645,
+      "step": 14868
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5840817886489587e-06,
+      "loss": 1.1179,
+      "step": 14869
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5839862411618577e-06,
+      "loss": 0.3595,
+      "step": 14870
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5838906936747563e-06,
+      "loss": 1.4251,
+      "step": 14871
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5837951461876558e-06,
+      "loss": 0.8402,
+      "step": 14872
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.583699598700555e-06,
+      "loss": 1.1134,
+      "step": 14873
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5836040512134534e-06,
+      "loss": 1.0239,
+      "step": 14874
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5835085037263525e-06,
+      "loss": 0.912,
+      "step": 14875
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.583412956239251e-06,
+      "loss": 0.9204,
+      "step": 14876
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58331740875215e-06,
+      "loss": 0.7349,
+      "step": 14877
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5832218612650487e-06,
+      "loss": 0.9659,
+      "step": 14878
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.583126313777948e-06,
+      "loss": 0.9967,
+      "step": 14879
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.583030766290847e-06,
+      "loss": 0.9364,
+      "step": 14880
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582935218803746e-06,
+      "loss": 0.8556,
+      "step": 14881
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5828396713166445e-06,
+      "loss": 0.8404,
+      "step": 14882
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5827441238295435e-06,
+      "loss": 0.7987,
+      "step": 14883
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5826485763424425e-06,
+      "loss": 0.4633,
+      "step": 14884
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582553028855341e-06,
+      "loss": 1.3467,
+      "step": 14885
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5824574813682406e-06,
+      "loss": 1.0319,
+      "step": 14886
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5823619338811392e-06,
+      "loss": 0.6246,
+      "step": 14887
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5822663863940383e-06,
+      "loss": 0.8748,
+      "step": 14888
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582170838906937e-06,
+      "loss": 0.9318,
+      "step": 14889
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582075291419836e-06,
+      "loss": 0.9706,
+      "step": 14890
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5819797439327345e-06,
+      "loss": 0.778,
+      "step": 14891
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5818841964456336e-06,
+      "loss": 1.2926,
+      "step": 14892
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581788648958533e-06,
+      "loss": 0.6614,
+      "step": 14893
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5816931014714317e-06,
+      "loss": 1.0013,
+      "step": 14894
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5815975539843307e-06,
+      "loss": 0.6387,
+      "step": 14895
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5815020064972293e-06,
+      "loss": 0.9892,
+      "step": 14896
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5814064590101283e-06,
+      "loss": 0.8169,
+      "step": 14897
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581310911523027e-06,
+      "loss": 1.1522,
+      "step": 14898
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581215364035926e-06,
+      "loss": 0.9594,
+      "step": 14899
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5811198165488246e-06,
+      "loss": 1.2781,
+      "step": 14900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.581024269061724e-06,
+      "loss": 0.7065,
+      "step": 14901
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.580928721574623e-06,
+      "loss": 0.719,
+      "step": 14902
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5808331740875217e-06,
+      "loss": 0.8767,
+      "step": 14903
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5807376266004208e-06,
+      "loss": 0.992,
+      "step": 14904
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5806420791133194e-06,
+      "loss": 1.1136,
+      "step": 14905
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5805465316262184e-06,
+      "loss": 0.7101,
+      "step": 14906
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.580450984139117e-06,
+      "loss": 1.4113,
+      "step": 14907
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5803554366520165e-06,
+      "loss": 0.8548,
+      "step": 14908
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5802598891649155e-06,
+      "loss": 1.219,
+      "step": 14909
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.580164341677814e-06,
+      "loss": 1.7897,
+      "step": 14910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.580068794190713e-06,
+      "loss": 1.3036,
+      "step": 14911
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579973246703612e-06,
+      "loss": 0.9671,
+      "step": 14912
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579877699216511e-06,
+      "loss": 1.1084,
+      "step": 14913
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5797821517294095e-06,
+      "loss": 0.9774,
+      "step": 14914
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579686604242309e-06,
+      "loss": 0.7161,
+      "step": 14915
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579591056755208e-06,
+      "loss": 1.3198,
+      "step": 14916
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5794955092681066e-06,
+      "loss": 0.9192,
+      "step": 14917
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5793999617810056e-06,
+      "loss": 0.9932,
+      "step": 14918
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5793044142939042e-06,
+      "loss": 0.986,
+      "step": 14919
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5792088668068033e-06,
+      "loss": 1.1373,
+      "step": 14920
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579113319319702e-06,
+      "loss": 1.2863,
+      "step": 14921
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5790177718326013e-06,
+      "loss": 1.0539,
+      "step": 14922
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5789222243455004e-06,
+      "loss": 0.8528,
+      "step": 14923
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.578826676858399e-06,
+      "loss": 0.9948,
+      "step": 14924
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.578731129371298e-06,
+      "loss": 0.9936,
+      "step": 14925
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5786355818841967e-06,
+      "loss": 0.6593,
+      "step": 14926
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5785400343970957e-06,
+      "loss": 0.501,
+      "step": 14927
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5784444869099943e-06,
+      "loss": 1.4537,
+      "step": 14928
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5783489394228938e-06,
+      "loss": 0.7491,
+      "step": 14929
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5782533919357924e-06,
+      "loss": 0.4775,
+      "step": 14930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5781578444486914e-06,
+      "loss": 0.9675,
+      "step": 14931
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.57806229696159e-06,
+      "loss": 0.8561,
+      "step": 14932
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577966749474489e-06,
+      "loss": 0.9928,
+      "step": 14933
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577871201987388e-06,
+      "loss": 0.6518,
+      "step": 14934
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5777756545002867e-06,
+      "loss": 1.387,
+      "step": 14935
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5776801070131858e-06,
+      "loss": 1.3046,
+      "step": 14936
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577584559526085e-06,
+      "loss": 0.8833,
+      "step": 14937
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577489012038984e-06,
+      "loss": 0.9328,
+      "step": 14938
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5773934645518825e-06,
+      "loss": 1.0509,
+      "step": 14939
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5772979170647815e-06,
+      "loss": 1.3515,
+      "step": 14940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.57720236957768e-06,
+      "loss": 0.8517,
+      "step": 14941
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.577106822090579e-06,
+      "loss": 1.0824,
+      "step": 14942
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5770112746034778e-06,
+      "loss": 1.1248,
+      "step": 14943
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5769157271163772e-06,
+      "loss": 1.1373,
+      "step": 14944
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5768201796292763e-06,
+      "loss": 1.036,
+      "step": 14945
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.576724632142175e-06,
+      "loss": 0.9486,
+      "step": 14946
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.576629084655074e-06,
+      "loss": 1.0815,
+      "step": 14947
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5765335371679725e-06,
+      "loss": 0.8666,
+      "step": 14948
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5764379896808716e-06,
+      "loss": 1.6646,
+      "step": 14949
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.57634244219377e-06,
+      "loss": 1.0022,
+      "step": 14950
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5762468947066697e-06,
+      "loss": 0.7337,
+      "step": 14951
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5761513472195687e-06,
+      "loss": 1.0814,
+      "step": 14952
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5760557997324673e-06,
+      "loss": 0.9302,
+      "step": 14953
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5759602522453663e-06,
+      "loss": 0.9645,
+      "step": 14954
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575864704758265e-06,
+      "loss": 1.4023,
+      "step": 14955
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575769157271164e-06,
+      "loss": 0.6739,
+      "step": 14956
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5756736097840626e-06,
+      "loss": 0.8323,
+      "step": 14957
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575578062296962e-06,
+      "loss": 0.8203,
+      "step": 14958
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575482514809861e-06,
+      "loss": 0.9732,
+      "step": 14959
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5753869673227597e-06,
+      "loss": 0.8881,
+      "step": 14960
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5752914198356588e-06,
+      "loss": 0.7948,
+      "step": 14961
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5751958723485574e-06,
+      "loss": 0.8644,
+      "step": 14962
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5751003248614564e-06,
+      "loss": 0.6835,
+      "step": 14963
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.575004777374355e-06,
+      "loss": 1.2186,
+      "step": 14964
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5749092298872545e-06,
+      "loss": 0.5749,
+      "step": 14965
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5748136824001535e-06,
+      "loss": 1.0273,
+      "step": 14966
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.574718134913052e-06,
+      "loss": 1.3983,
+      "step": 14967
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.574622587425951e-06,
+      "loss": 1.0084,
+      "step": 14968
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.57452703993885e-06,
+      "loss": 0.7178,
+      "step": 14969
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.574431492451749e-06,
+      "loss": 0.6706,
+      "step": 14970
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5743359449646475e-06,
+      "loss": 0.789,
+      "step": 14971
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5742403974775465e-06,
+      "loss": 1.4191,
+      "step": 14972
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.574144849990446e-06,
+      "loss": 1.3275,
+      "step": 14973
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5740493025033446e-06,
+      "loss": 1.3237,
+      "step": 14974
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5739537550162436e-06,
+      "loss": 0.699,
+      "step": 14975
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5738582075291422e-06,
+      "loss": 1.0267,
+      "step": 14976
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5737626600420413e-06,
+      "loss": 0.7972,
+      "step": 14977
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.57366711255494e-06,
+      "loss": 0.976,
+      "step": 14978
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.573571565067839e-06,
+      "loss": 0.8367,
+      "step": 14979
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.573476017580738e-06,
+      "loss": 0.6667,
+      "step": 14980
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.573380470093637e-06,
+      "loss": 0.9702,
+      "step": 14981
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5732849226065356e-06,
+      "loss": 1.541,
+      "step": 14982
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5731893751194347e-06,
+      "loss": 1.1852,
+      "step": 14983
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5730938276323333e-06,
+      "loss": 1.5329,
+      "step": 14984
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5729982801452323e-06,
+      "loss": 0.9339,
+      "step": 14985
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5729027326581313e-06,
+      "loss": 0.8268,
+      "step": 14986
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5728071851710304e-06,
+      "loss": 1.0618,
+      "step": 14987
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5727116376839294e-06,
+      "loss": 0.8968,
+      "step": 14988
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.572616090196828e-06,
+      "loss": 0.8016,
+      "step": 14989
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.572520542709727e-06,
+      "loss": 1.0524,
+      "step": 14990
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5724249952226257e-06,
+      "loss": 1.5843,
+      "step": 14991
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5723294477355247e-06,
+      "loss": 1.0888,
+      "step": 14992
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5722339002484233e-06,
+      "loss": 0.8481,
+      "step": 14993
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.572138352761323e-06,
+      "loss": 0.7153,
+      "step": 14994
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.572042805274222e-06,
+      "loss": 0.8229,
+      "step": 14995
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5719472577871205e-06,
+      "loss": 0.8114,
+      "step": 14996
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5718517103000195e-06,
+      "loss": 0.9133,
+      "step": 14997
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571756162812918e-06,
+      "loss": 1.1556,
+      "step": 14998
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571660615325817e-06,
+      "loss": 0.7045,
+      "step": 14999
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5715650678387158e-06,
+      "loss": 1.0204,
+      "step": 15000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5714695203516152e-06,
+      "loss": 0.7621,
+      "step": 15001
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5713739728645143e-06,
+      "loss": 0.7766,
+      "step": 15002
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571278425377413e-06,
+      "loss": 0.917,
+      "step": 15003
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571182877890312e-06,
+      "loss": 0.7152,
+      "step": 15004
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5710873304032105e-06,
+      "loss": 0.8672,
+      "step": 15005
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5709917829161096e-06,
+      "loss": 1.2535,
+      "step": 15006
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.570896235429008e-06,
+      "loss": 1.8489,
+      "step": 15007
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5708006879419072e-06,
+      "loss": 0.8572,
+      "step": 15008
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5707051404548067e-06,
+      "loss": 0.9813,
+      "step": 15009
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5706095929677053e-06,
+      "loss": 0.9424,
+      "step": 15010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5705140454806043e-06,
+      "loss": 0.8013,
+      "step": 15011
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.570418497993503e-06,
+      "loss": 1.1077,
+      "step": 15012
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.570322950506402e-06,
+      "loss": 0.7328,
+      "step": 15013
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5702274030193006e-06,
+      "loss": 0.7961,
+      "step": 15014
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5701318555321997e-06,
+      "loss": 0.7749,
+      "step": 15015
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.570036308045099e-06,
+      "loss": 0.7488,
+      "step": 15016
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5699407605579977e-06,
+      "loss": 0.9427,
+      "step": 15017
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5698452130708968e-06,
+      "loss": 1.003,
+      "step": 15018
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5697496655837954e-06,
+      "loss": 0.7714,
+      "step": 15019
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5696541180966944e-06,
+      "loss": 0.8543,
+      "step": 15020
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.569558570609593e-06,
+      "loss": 1.0649,
+      "step": 15021
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.569463023122492e-06,
+      "loss": 0.6606,
+      "step": 15022
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.569367475635391e-06,
+      "loss": 0.7388,
+      "step": 15023
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.56927192814829e-06,
+      "loss": 1.0503,
+      "step": 15024
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.569176380661189e-06,
+      "loss": 0.885,
+      "step": 15025
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.569080833174088e-06,
+      "loss": 0.5943,
+      "step": 15026
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568985285686987e-06,
+      "loss": 1.1943,
+      "step": 15027
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5688897381998855e-06,
+      "loss": 0.9055,
+      "step": 15028
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5687941907127845e-06,
+      "loss": 0.922,
+      "step": 15029
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5686986432256835e-06,
+      "loss": 1.5498,
+      "step": 15030
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5686030957385826e-06,
+      "loss": 1.0269,
+      "step": 15031
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568507548251481e-06,
+      "loss": 1.1172,
+      "step": 15032
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5684120007643802e-06,
+      "loss": 0.912,
+      "step": 15033
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568316453277279e-06,
+      "loss": 1.1889,
+      "step": 15034
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568220905790178e-06,
+      "loss": 1.0198,
+      "step": 15035
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568125358303077e-06,
+      "loss": 1.3115,
+      "step": 15036
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.568029810815976e-06,
+      "loss": 1.2784,
+      "step": 15037
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567934263328875e-06,
+      "loss": 1.3036,
+      "step": 15038
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5678387158417736e-06,
+      "loss": 1.1377,
+      "step": 15039
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5677431683546727e-06,
+      "loss": 1.262,
+      "step": 15040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5676476208675713e-06,
+      "loss": 0.9564,
+      "step": 15041
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5675520733804703e-06,
+      "loss": 0.9799,
+      "step": 15042
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567456525893369e-06,
+      "loss": 0.8713,
+      "step": 15043
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567360978406268e-06,
+      "loss": 0.6979,
+      "step": 15044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5672654309191674e-06,
+      "loss": 0.5514,
+      "step": 15045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567169883432066e-06,
+      "loss": 1.0206,
+      "step": 15046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567074335944965e-06,
+      "loss": 1.1167,
+      "step": 15047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5669787884578637e-06,
+      "loss": 0.6803,
+      "step": 15048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5668832409707627e-06,
+      "loss": 1.0317,
+      "step": 15049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5667876934836613e-06,
+      "loss": 1.4577,
+      "step": 15050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5666921459965604e-06,
+      "loss": 0.91,
+      "step": 15051
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.56659659850946e-06,
+      "loss": 0.8313,
+      "step": 15052
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5665010510223585e-06,
+      "loss": 0.9698,
+      "step": 15053
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5664055035352575e-06,
+      "loss": 0.9023,
+      "step": 15054
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.566309956048156e-06,
+      "loss": 0.6357,
+      "step": 15055
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.566214408561055e-06,
+      "loss": 0.6161,
+      "step": 15056
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5661188610739538e-06,
+      "loss": 0.8009,
+      "step": 15057
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.566023313586853e-06,
+      "loss": 0.9182,
+      "step": 15058
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5659277660997523e-06,
+      "loss": 0.7339,
+      "step": 15059
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.565832218612651e-06,
+      "loss": 0.7241,
+      "step": 15060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56573667112555e-06,
+      "loss": 0.8868,
+      "step": 15061
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5656411236384485e-06,
+      "loss": 1.0648,
+      "step": 15062
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5655455761513476e-06,
+      "loss": 0.9681,
+      "step": 15063
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.565450028664246e-06,
+      "loss": 0.9864,
+      "step": 15064
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5653544811771452e-06,
+      "loss": 0.888,
+      "step": 15065
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5652589336900447e-06,
+      "loss": 1.1483,
+      "step": 15066
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5651633862029433e-06,
+      "loss": 0.9211,
+      "step": 15067
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5650678387158423e-06,
+      "loss": 1.3661,
+      "step": 15068
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.564972291228741e-06,
+      "loss": 0.7253,
+      "step": 15069
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56487674374164e-06,
+      "loss": 1.1836,
+      "step": 15070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5647811962545386e-06,
+      "loss": 1.1945,
+      "step": 15071
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5646856487674377e-06,
+      "loss": 0.668,
+      "step": 15072
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5645901012803367e-06,
+      "loss": 0.7462,
+      "step": 15073
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5644945537932357e-06,
+      "loss": 0.8333,
+      "step": 15074
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5643990063061343e-06,
+      "loss": 0.578,
+      "step": 15075
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5643034588190334e-06,
+      "loss": 1.445,
+      "step": 15076
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5642079113319324e-06,
+      "loss": 0.698,
+      "step": 15077
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.564112363844831e-06,
+      "loss": 0.7946,
+      "step": 15078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.56401681635773e-06,
+      "loss": 0.9632,
+      "step": 15079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5639212688706287e-06,
+      "loss": 0.894,
+      "step": 15080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563825721383528e-06,
+      "loss": 0.9102,
+      "step": 15081
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5637301738964268e-06,
+      "loss": 0.5623,
+      "step": 15082
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563634626409326e-06,
+      "loss": 0.6269,
+      "step": 15083
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5635390789222244e-06,
+      "loss": 1.4708,
+      "step": 15084
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5634435314351235e-06,
+      "loss": 0.8217,
+      "step": 15085
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563347983948022e-06,
+      "loss": 1.0088,
+      "step": 15086
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563252436460921e-06,
+      "loss": 0.7492,
+      "step": 15087
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5631568889738206e-06,
+      "loss": 0.9971,
+      "step": 15088
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563061341486719e-06,
+      "loss": 1.0245,
+      "step": 15089
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5629657939996182e-06,
+      "loss": 1.1592,
+      "step": 15090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.562870246512517e-06,
+      "loss": 0.9952,
+      "step": 15091
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.562774699025416e-06,
+      "loss": 0.7623,
+      "step": 15092
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5626791515383145e-06,
+      "loss": 0.5452,
+      "step": 15093
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5625836040512135e-06,
+      "loss": 0.7448,
+      "step": 15094
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.562488056564113e-06,
+      "loss": 0.4625,
+      "step": 15095
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5623925090770116e-06,
+      "loss": 0.644,
+      "step": 15096
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5622969615899107e-06,
+      "loss": 1.0226,
+      "step": 15097
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5622014141028093e-06,
+      "loss": 1.0612,
+      "step": 15098
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5621058666157083e-06,
+      "loss": 0.757,
+      "step": 15099
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.562010319128607e-06,
+      "loss": 0.644,
+      "step": 15100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561914771641506e-06,
+      "loss": 0.7632,
+      "step": 15101
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5618192241544054e-06,
+      "loss": 1.0485,
+      "step": 15102
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561723676667304e-06,
+      "loss": 0.6202,
+      "step": 15103
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561628129180203e-06,
+      "loss": 0.7944,
+      "step": 15104
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5615325816931017e-06,
+      "loss": 0.9823,
+      "step": 15105
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5614370342060007e-06,
+      "loss": 0.6231,
+      "step": 15106
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5613414867188993e-06,
+      "loss": 0.562,
+      "step": 15107
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5612459392317984e-06,
+      "loss": 1.0216,
+      "step": 15108
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.561150391744698e-06,
+      "loss": 1.1767,
+      "step": 15109
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5610548442575965e-06,
+      "loss": 1.1151,
+      "step": 15110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5609592967704955e-06,
+      "loss": 0.6795,
+      "step": 15111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560863749283394e-06,
+      "loss": 0.5676,
+      "step": 15112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560768201796293e-06,
+      "loss": 1.2456,
+      "step": 15113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5606726543091918e-06,
+      "loss": 0.8921,
+      "step": 15114
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560577106822091e-06,
+      "loss": 0.7177,
+      "step": 15115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5604815593349894e-06,
+      "loss": 0.9036,
+      "step": 15116
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560386011847889e-06,
+      "loss": 1.103,
+      "step": 15117
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560290464360788e-06,
+      "loss": 1.1796,
+      "step": 15118
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5601949168736865e-06,
+      "loss": 1.1605,
+      "step": 15119
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5600993693865856e-06,
+      "loss": 1.2675,
+      "step": 15120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.560003821899484e-06,
+      "loss": 0.7959,
+      "step": 15121
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5599082744123832e-06,
+      "loss": 0.8966,
+      "step": 15122
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.559812726925282e-06,
+      "loss": 1.277,
+      "step": 15123
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5597171794381813e-06,
+      "loss": 0.8482,
+      "step": 15124
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.55962163195108e-06,
+      "loss": 1.5295,
+      "step": 15125
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.559526084463979e-06,
+      "loss": 1.1652,
+      "step": 15126
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.559430536976878e-06,
+      "loss": 0.8675,
+      "step": 15127
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5593349894897766e-06,
+      "loss": 1.3088,
+      "step": 15128
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5592394420026757e-06,
+      "loss": 0.9842,
+      "step": 15129
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5591438945155743e-06,
+      "loss": 1.211,
+      "step": 15130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5590483470284737e-06,
+      "loss": 1.2307,
+      "step": 15131
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5589527995413723e-06,
+      "loss": 0.9342,
+      "step": 15132
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5588572520542714e-06,
+      "loss": 1.2751,
+      "step": 15133
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.55876170456717e-06,
+      "loss": 1.3591,
+      "step": 15134
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.558666157080069e-06,
+      "loss": 0.9882,
+      "step": 15135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5585706095929677e-06,
+      "loss": 0.8707,
+      "step": 15136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5584750621058667e-06,
+      "loss": 1.3846,
+      "step": 15137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.558379514618766e-06,
+      "loss": 0.9666,
+      "step": 15138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5582839671316648e-06,
+      "loss": 1.0494,
+      "step": 15139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.558188419644564e-06,
+      "loss": 0.7976,
+      "step": 15140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5580928721574624e-06,
+      "loss": 1.3511,
+      "step": 15141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5579973246703615e-06,
+      "loss": 1.0695,
+      "step": 15142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.55790177718326e-06,
+      "loss": 0.9713,
+      "step": 15143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.557806229696159e-06,
+      "loss": 1.1232,
+      "step": 15144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5577106822090586e-06,
+      "loss": 0.7953,
+      "step": 15145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.557615134721957e-06,
+      "loss": 0.5984,
+      "step": 15146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5575195872348562e-06,
+      "loss": 1.193,
+      "step": 15147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.557424039747755e-06,
+      "loss": 0.7647,
+      "step": 15148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.557328492260654e-06,
+      "loss": 0.6331,
+      "step": 15149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5572329447735525e-06,
+      "loss": 0.8946,
+      "step": 15150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5571373972864515e-06,
+      "loss": 0.7789,
+      "step": 15151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.55704184979935e-06,
+      "loss": 0.7224,
+      "step": 15152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5569463023122496e-06,
+      "loss": 0.7054,
+      "step": 15153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5568507548251487e-06,
+      "loss": 1.2283,
+      "step": 15154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5567552073380473e-06,
+      "loss": 1.3767,
+      "step": 15155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5566596598509463e-06,
+      "loss": 1.3093,
+      "step": 15156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556564112363845e-06,
+      "loss": 0.7403,
+      "step": 15157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556468564876744e-06,
+      "loss": 1.0315,
+      "step": 15158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5563730173896426e-06,
+      "loss": 1.1053,
+      "step": 15159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556277469902542e-06,
+      "loss": 1.2332,
+      "step": 15160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556181922415441e-06,
+      "loss": 1.2753,
+      "step": 15161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5560863749283397e-06,
+      "loss": 1.4037,
+      "step": 15162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5559908274412387e-06,
+      "loss": 0.6749,
+      "step": 15163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5558952799541373e-06,
+      "loss": 1.1945,
+      "step": 15164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5557997324670364e-06,
+      "loss": 1.2204,
+      "step": 15165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555704184979935e-06,
+      "loss": 1.2992,
+      "step": 15166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5556086374928345e-06,
+      "loss": 1.023,
+      "step": 15167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5555130900057335e-06,
+      "loss": 0.9046,
+      "step": 15168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555417542518632e-06,
+      "loss": 0.9326,
+      "step": 15169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555321995031531e-06,
+      "loss": 1.2632,
+      "step": 15170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5552264475444298e-06,
+      "loss": 1.2274,
+      "step": 15171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.555130900057329e-06,
+      "loss": 1.3482,
+      "step": 15172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5550353525702274e-06,
+      "loss": 1.0773,
+      "step": 15173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.554939805083127e-06,
+      "loss": 0.9869,
+      "step": 15174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5548442575960255e-06,
+      "loss": 1.2182,
+      "step": 15175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5547487101089245e-06,
+      "loss": 1.3337,
+      "step": 15176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.554653162621823e-06,
+      "loss": 0.5417,
+      "step": 15177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.554557615134722e-06,
+      "loss": 0.9174,
+      "step": 15178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5544620676476212e-06,
+      "loss": 0.761,
+      "step": 15179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.55436652016052e-06,
+      "loss": 0.6936,
+      "step": 15180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5542709726734193e-06,
+      "loss": 1.2129,
+      "step": 15181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.554175425186318e-06,
+      "loss": 1.1304,
+      "step": 15182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.554079877699217e-06,
+      "loss": 0.7998,
+      "step": 15183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5539843302121156e-06,
+      "loss": 1.1785,
+      "step": 15184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5538887827250146e-06,
+      "loss": 1.288,
+      "step": 15185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5537932352379132e-06,
+      "loss": 0.6514,
+      "step": 15186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5536976877508123e-06,
+      "loss": 1.2526,
+      "step": 15187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.553602140263711e-06,
+      "loss": 0.6002,
+      "step": 15188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5535065927766103e-06,
+      "loss": 1.7549,
+      "step": 15189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5534110452895094e-06,
+      "loss": 1.0038,
+      "step": 15190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.553315497802408e-06,
+      "loss": 0.5985,
+      "step": 15191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.553219950315307e-06,
+      "loss": 0.9837,
+      "step": 15192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5531244028282057e-06,
+      "loss": 1.1129,
+      "step": 15193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5530288553411047e-06,
+      "loss": 1.0841,
+      "step": 15194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5529333078540033e-06,
+      "loss": 0.9445,
+      "step": 15195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5528377603669028e-06,
+      "loss": 1.549,
+      "step": 15196
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.552742212879802e-06,
+      "loss": 0.7327,
+      "step": 15197
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5526466653927004e-06,
+      "loss": 0.8636,
+      "step": 15198
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5525511179055995e-06,
+      "loss": 1.0567,
+      "step": 15199
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.552455570418498e-06,
+      "loss": 1.1181,
+      "step": 15200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.552360022931397e-06,
+      "loss": 1.0963,
+      "step": 15201
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5522644754442957e-06,
+      "loss": 0.7846,
+      "step": 15202
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.552168927957195e-06,
+      "loss": 0.8471,
+      "step": 15203
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5520733804700942e-06,
+      "loss": 0.5609,
+      "step": 15204
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551977832982993e-06,
+      "loss": 1.1706,
+      "step": 15205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551882285495892e-06,
+      "loss": 0.9269,
+      "step": 15206
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5517867380087905e-06,
+      "loss": 1.2394,
+      "step": 15207
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5516911905216895e-06,
+      "loss": 1.1835,
+      "step": 15208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551595643034588e-06,
+      "loss": 1.1854,
+      "step": 15209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5515000955474876e-06,
+      "loss": 0.6285,
+      "step": 15210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5514045480603867e-06,
+      "loss": 1.0228,
+      "step": 15211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5513090005732853e-06,
+      "loss": 0.9168,
+      "step": 15212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5512134530861843e-06,
+      "loss": 0.9951,
+      "step": 15213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551117905599083e-06,
+      "loss": 0.664,
+      "step": 15214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551022358111982e-06,
+      "loss": 0.7125,
+      "step": 15215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5509268106248806e-06,
+      "loss": 0.7205,
+      "step": 15216
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.55083126313778e-06,
+      "loss": 1.0415,
+      "step": 15217
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550735715650679e-06,
+      "loss": 1.1399,
+      "step": 15218
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5506401681635777e-06,
+      "loss": 1.0948,
+      "step": 15219
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5505446206764767e-06,
+      "loss": 1.093,
+      "step": 15220
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5504490731893753e-06,
+      "loss": 0.674,
+      "step": 15221
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5503535257022744e-06,
+      "loss": 1.0347,
+      "step": 15222
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550257978215173e-06,
+      "loss": 0.8605,
+      "step": 15223
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550162430728072e-06,
+      "loss": 0.9221,
+      "step": 15224
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.550066883240971e-06,
+      "loss": 1.6371,
+      "step": 15225
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.54997133575387e-06,
+      "loss": 1.1508,
+      "step": 15226
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5498757882667687e-06,
+      "loss": 1.2122,
+      "step": 15227
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5497802407796678e-06,
+      "loss": 0.8644,
+      "step": 15228
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.549684693292567e-06,
+      "loss": 0.53,
+      "step": 15229
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5495891458054654e-06,
+      "loss": 0.9234,
+      "step": 15230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5494935983183645e-06,
+      "loss": 0.9634,
+      "step": 15231
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5493980508312635e-06,
+      "loss": 1.3301,
+      "step": 15232
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5493025033441625e-06,
+      "loss": 1.0064,
+      "step": 15233
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.549206955857061e-06,
+      "loss": 1.1659,
+      "step": 15234
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.54911140836996e-06,
+      "loss": 1.1492,
+      "step": 15235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.549015860882859e-06,
+      "loss": 1.3155,
+      "step": 15236
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.548920313395758e-06,
+      "loss": 1.1271,
+      "step": 15237
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5488247659086565e-06,
+      "loss": 1.1895,
+      "step": 15238
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.548729218421556e-06,
+      "loss": 1.0405,
+      "step": 15239
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.548633670934455e-06,
+      "loss": 0.8487,
+      "step": 15240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5485381234473536e-06,
+      "loss": 0.9467,
+      "step": 15241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5484425759602526e-06,
+      "loss": 0.9415,
+      "step": 15242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5483470284731512e-06,
+      "loss": 0.91,
+      "step": 15243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5482514809860503e-06,
+      "loss": 0.8534,
+      "step": 15244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.548155933498949e-06,
+      "loss": 1.1791,
+      "step": 15245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5480603860118483e-06,
+      "loss": 0.9416,
+      "step": 15246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5479648385247474e-06,
+      "loss": 0.7311,
+      "step": 15247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.547869291037646e-06,
+      "loss": 0.8356,
+      "step": 15248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.547773743550545e-06,
+      "loss": 1.1087,
+      "step": 15249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5476781960634436e-06,
+      "loss": 0.7362,
+      "step": 15250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5475826485763427e-06,
+      "loss": 0.9966,
+      "step": 15251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5474871010892413e-06,
+      "loss": 0.9211,
+      "step": 15252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5473915536021408e-06,
+      "loss": 1.0542,
+      "step": 15253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.54729600611504e-06,
+      "loss": 1.2117,
+      "step": 15254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5472004586279384e-06,
+      "loss": 1.0813,
+      "step": 15255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5471049111408375e-06,
+      "loss": 1.0181,
+      "step": 15256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.547009363653736e-06,
+      "loss": 0.9266,
+      "step": 15257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.546913816166635e-06,
+      "loss": 0.5966,
+      "step": 15258
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5468182686795337e-06,
+      "loss": 1.3898,
+      "step": 15259
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5467227211924328e-06,
+      "loss": 1.046,
+      "step": 15260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5466271737053322e-06,
+      "loss": 1.369,
+      "step": 15261
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.546531626218231e-06,
+      "loss": 0.7218,
+      "step": 15262
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.54643607873113e-06,
+      "loss": 1.2741,
+      "step": 15263
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5463405312440285e-06,
+      "loss": 1.4187,
+      "step": 15264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5462449837569275e-06,
+      "loss": 0.8912,
+      "step": 15265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.546149436269826e-06,
+      "loss": 0.5686,
+      "step": 15266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.546053888782725e-06,
+      "loss": 1.1728,
+      "step": 15267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5459583412956247e-06,
+      "loss": 0.9513,
+      "step": 15268
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5458627938085233e-06,
+      "loss": 0.5785,
+      "step": 15269
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5457672463214223e-06,
+      "loss": 0.6627,
+      "step": 15270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.545671698834321e-06,
+      "loss": 0.8473,
+      "step": 15271
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54557615134722e-06,
+      "loss": 0.8892,
+      "step": 15272
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5454806038601186e-06,
+      "loss": 1.1301,
+      "step": 15273
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5453850563730176e-06,
+      "loss": 1.0082,
+      "step": 15274
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5452895088859166e-06,
+      "loss": 0.8484,
+      "step": 15275
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5451939613988157e-06,
+      "loss": 1.6541,
+      "step": 15276
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5450984139117143e-06,
+      "loss": 0.8107,
+      "step": 15277
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5450028664246133e-06,
+      "loss": 0.5019,
+      "step": 15278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.544907318937512e-06,
+      "loss": 0.6308,
+      "step": 15279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.544811771450411e-06,
+      "loss": 1.1323,
+      "step": 15280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54471622396331e-06,
+      "loss": 0.6911,
+      "step": 15281
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.544620676476209e-06,
+      "loss": 1.1406,
+      "step": 15282
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.544525128989108e-06,
+      "loss": 0.8359,
+      "step": 15283
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5444295815020067e-06,
+      "loss": 0.9262,
+      "step": 15284
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5443340340149058e-06,
+      "loss": 0.4851,
+      "step": 15285
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5442384865278044e-06,
+      "loss": 0.677,
+      "step": 15286
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5441429390407034e-06,
+      "loss": 1.0603,
+      "step": 15287
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.544047391553602e-06,
+      "loss": 0.8019,
+      "step": 15288
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5439518440665015e-06,
+      "loss": 0.7221,
+      "step": 15289
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5438562965794005e-06,
+      "loss": 0.8227,
+      "step": 15290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.543760749092299e-06,
+      "loss": 1.0647,
+      "step": 15291
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.543665201605198e-06,
+      "loss": 0.6178,
+      "step": 15292
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.543569654118097e-06,
+      "loss": 1.2599,
+      "step": 15293
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.543474106630996e-06,
+      "loss": 1.4372,
+      "step": 15294
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5433785591438945e-06,
+      "loss": 0.8555,
+      "step": 15295
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5432830116567935e-06,
+      "loss": 0.8123,
+      "step": 15296
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.543187464169693e-06,
+      "loss": 1.1974,
+      "step": 15297
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5430919166825916e-06,
+      "loss": 1.2197,
+      "step": 15298
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5429963691954906e-06,
+      "loss": 0.7931,
+      "step": 15299
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5429008217083892e-06,
+      "loss": 1.0693,
+      "step": 15300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5428052742212883e-06,
+      "loss": 0.9071,
+      "step": 15301
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.542709726734187e-06,
+      "loss": 1.3814,
+      "step": 15302
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.542614179247086e-06,
+      "loss": 1.1952,
+      "step": 15303
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5425186317599854e-06,
+      "loss": 0.9054,
+      "step": 15304
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.542423084272884e-06,
+      "loss": 1.2828,
+      "step": 15305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.542327536785783e-06,
+      "loss": 0.6259,
+      "step": 15306
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5422319892986816e-06,
+      "loss": 1.229,
+      "step": 15307
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5421364418115807e-06,
+      "loss": 0.9713,
+      "step": 15308
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5420408943244793e-06,
+      "loss": 1.121,
+      "step": 15309
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5419453468373783e-06,
+      "loss": 0.787,
+      "step": 15310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541849799350278e-06,
+      "loss": 0.4906,
+      "step": 15311
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5417542518631764e-06,
+      "loss": 0.631,
+      "step": 15312
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5416587043760755e-06,
+      "loss": 1.1836,
+      "step": 15313
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541563156888974e-06,
+      "loss": 0.8872,
+      "step": 15314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541467609401873e-06,
+      "loss": 0.846,
+      "step": 15315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5413720619147717e-06,
+      "loss": 0.8249,
+      "step": 15316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5412765144276708e-06,
+      "loss": 0.5514,
+      "step": 15317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.54118096694057e-06,
+      "loss": 0.5891,
+      "step": 15318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541085419453469e-06,
+      "loss": 1.0719,
+      "step": 15319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.540989871966368e-06,
+      "loss": 0.9512,
+      "step": 15320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5408943244792665e-06,
+      "loss": 1.1709,
+      "step": 15321
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5407987769921655e-06,
+      "loss": 0.9612,
+      "step": 15322
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.540703229505064e-06,
+      "loss": 1.6227,
+      "step": 15323
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.540607682017963e-06,
+      "loss": 1.0202,
+      "step": 15324
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5405121345308622e-06,
+      "loss": 1.3045,
+      "step": 15325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5404165870437613e-06,
+      "loss": 0.9163,
+      "step": 15326
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.54032103955666e-06,
+      "loss": 0.7479,
+      "step": 15327
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.540225492069559e-06,
+      "loss": 1.4196,
+      "step": 15328
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5401299445824575e-06,
+      "loss": 0.7412,
+      "step": 15329
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5400343970953566e-06,
+      "loss": 0.8974,
+      "step": 15330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5399388496082556e-06,
+      "loss": 1.1967,
+      "step": 15331
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5398433021211542e-06,
+      "loss": 0.5388,
+      "step": 15332
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5397477546340537e-06,
+      "loss": 1.1875,
+      "step": 15333
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5396522071469523e-06,
+      "loss": 0.8559,
+      "step": 15334
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5395566596598513e-06,
+      "loss": 0.8523,
+      "step": 15335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.53946111217275e-06,
+      "loss": 1.5204,
+      "step": 15336
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.539365564685649e-06,
+      "loss": 0.7218,
+      "step": 15337
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5392700171985476e-06,
+      "loss": 0.9911,
+      "step": 15338
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5391744697114466e-06,
+      "loss": 0.7327,
+      "step": 15339
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.539078922224346e-06,
+      "loss": 1.0682,
+      "step": 15340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5389833747372447e-06,
+      "loss": 1.0475,
+      "step": 15341
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5388878272501438e-06,
+      "loss": 1.0672,
+      "step": 15342
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5387922797630424e-06,
+      "loss": 0.7878,
+      "step": 15343
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5386967322759414e-06,
+      "loss": 1.0872,
+      "step": 15344
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.53860118478884e-06,
+      "loss": 1.0923,
+      "step": 15345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.538505637301739e-06,
+      "loss": 0.8721,
+      "step": 15346
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5384100898146385e-06,
+      "loss": 1.0012,
+      "step": 15347
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.538314542327537e-06,
+      "loss": 0.9966,
+      "step": 15348
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.538218994840436e-06,
+      "loss": 0.9024,
+      "step": 15349
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.538123447353335e-06,
+      "loss": 0.9053,
+      "step": 15350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.538027899866234e-06,
+      "loss": 0.8215,
+      "step": 15351
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5379323523791325e-06,
+      "loss": 0.8803,
+      "step": 15352
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5378368048920315e-06,
+      "loss": 0.9246,
+      "step": 15353
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537741257404931e-06,
+      "loss": 1.186,
+      "step": 15354
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5376457099178296e-06,
+      "loss": 0.8498,
+      "step": 15355
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5375501624307286e-06,
+      "loss": 1.0999,
+      "step": 15356
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5374546149436272e-06,
+      "loss": 0.7074,
+      "step": 15357
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5373590674565263e-06,
+      "loss": 0.927,
+      "step": 15358
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537263519969425e-06,
+      "loss": 1.0754,
+      "step": 15359
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537167972482324e-06,
+      "loss": 0.7741,
+      "step": 15360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5370724249952234e-06,
+      "loss": 1.0299,
+      "step": 15361
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536976877508122e-06,
+      "loss": 1.1846,
+      "step": 15362
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536881330021021e-06,
+      "loss": 1.0494,
+      "step": 15363
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5367857825339196e-06,
+      "loss": 1.2795,
+      "step": 15364
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5366902350468187e-06,
+      "loss": 0.7593,
+      "step": 15365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5365946875597173e-06,
+      "loss": 1.2155,
+      "step": 15366
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5364991400726163e-06,
+      "loss": 1.1464,
+      "step": 15367
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536403592585515e-06,
+      "loss": 1.2469,
+      "step": 15368
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5363080450984144e-06,
+      "loss": 0.8155,
+      "step": 15369
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5362124976113135e-06,
+      "loss": 1.4565,
+      "step": 15370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536116950124212e-06,
+      "loss": 1.2761,
+      "step": 15371
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536021402637111e-06,
+      "loss": 1.2573,
+      "step": 15372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5359258551500097e-06,
+      "loss": 1.0409,
+      "step": 15373
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5358303076629088e-06,
+      "loss": 0.7894,
+      "step": 15374
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5357347601758074e-06,
+      "loss": 0.9391,
+      "step": 15375
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.535639212688707e-06,
+      "loss": 1.0095,
+      "step": 15376
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5355436652016055e-06,
+      "loss": 0.5453,
+      "step": 15377
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5354481177145045e-06,
+      "loss": 0.4171,
+      "step": 15378
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.535352570227403e-06,
+      "loss": 0.506,
+      "step": 15379
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.535257022740302e-06,
+      "loss": 1.1287,
+      "step": 15380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5351614752532008e-06,
+      "loss": 1.625,
+      "step": 15381
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5350659277661e-06,
+      "loss": 0.6579,
+      "step": 15382
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5349703802789993e-06,
+      "loss": 0.7053,
+      "step": 15383
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.534874832791898e-06,
+      "loss": 0.9674,
+      "step": 15384
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.534779285304797e-06,
+      "loss": 0.8505,
+      "step": 15385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5346837378176955e-06,
+      "loss": 0.9074,
+      "step": 15386
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5345881903305946e-06,
+      "loss": 1.0045,
+      "step": 15387
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.534492642843493e-06,
+      "loss": 0.6462,
+      "step": 15388
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5343970953563922e-06,
+      "loss": 0.9386,
+      "step": 15389
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5343015478692917e-06,
+      "loss": 0.9339,
+      "step": 15390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5342060003821903e-06,
+      "loss": 0.8434,
+      "step": 15391
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5341104528950893e-06,
+      "loss": 0.8309,
+      "step": 15392
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.534014905407988e-06,
+      "loss": 0.9114,
+      "step": 15393
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.533919357920887e-06,
+      "loss": 1.1902,
+      "step": 15394
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5338238104337856e-06,
+      "loss": 0.9037,
+      "step": 15395
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5337282629466846e-06,
+      "loss": 1.0348,
+      "step": 15396
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.533632715459584e-06,
+      "loss": 1.0411,
+      "step": 15397
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5335371679724827e-06,
+      "loss": 1.0966,
+      "step": 15398
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5334416204853818e-06,
+      "loss": 0.7671,
+      "step": 15399
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5333460729982804e-06,
+      "loss": 1.2371,
+      "step": 15400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5332505255111794e-06,
+      "loss": 1.3294,
+      "step": 15401
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.533154978024078e-06,
+      "loss": 1.1934,
+      "step": 15402
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.533059430536977e-06,
+      "loss": 0.9819,
+      "step": 15403
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5329638830498757e-06,
+      "loss": 1.098,
+      "step": 15404
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532868335562775e-06,
+      "loss": 0.7321,
+      "step": 15405
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532772788075674e-06,
+      "loss": 0.9005,
+      "step": 15406
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532677240588573e-06,
+      "loss": 0.6973,
+      "step": 15407
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532581693101472e-06,
+      "loss": 0.6921,
+      "step": 15408
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5324861456143705e-06,
+      "loss": 0.7651,
+      "step": 15409
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5323905981272695e-06,
+      "loss": 0.5729,
+      "step": 15410
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532295050640168e-06,
+      "loss": 1.2322,
+      "step": 15411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5321995031530676e-06,
+      "loss": 0.9397,
+      "step": 15412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5321039556659666e-06,
+      "loss": 1.655,
+      "step": 15413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5320084081788652e-06,
+      "loss": 0.8124,
+      "step": 15414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5319128606917643e-06,
+      "loss": 1.3524,
+      "step": 15415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.531817313204663e-06,
+      "loss": 1.1085,
+      "step": 15416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.531721765717562e-06,
+      "loss": 0.6114,
+      "step": 15417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5316262182304605e-06,
+      "loss": 0.6248,
+      "step": 15418
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.53153067074336e-06,
+      "loss": 0.6973,
+      "step": 15419
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5314351232562586e-06,
+      "loss": 0.9237,
+      "step": 15420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5313395757691576e-06,
+      "loss": 0.6103,
+      "step": 15421
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5312440282820567e-06,
+      "loss": 1.284,
+      "step": 15422
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5311484807949553e-06,
+      "loss": 1.0156,
+      "step": 15423
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5310529333078543e-06,
+      "loss": 0.603,
+      "step": 15424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.530957385820753e-06,
+      "loss": 0.9528,
+      "step": 15425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5308618383336524e-06,
+      "loss": 0.393,
+      "step": 15426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.530766290846551e-06,
+      "loss": 1.314,
+      "step": 15427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.53067074335945e-06,
+      "loss": 0.8423,
+      "step": 15428
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5305751958723487e-06,
+      "loss": 1.2551,
+      "step": 15429
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5304796483852477e-06,
+      "loss": 1.0026,
+      "step": 15430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5303841008981463e-06,
+      "loss": 0.6306,
+      "step": 15431
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5302885534110454e-06,
+      "loss": 1.0396,
+      "step": 15432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.530193005923945e-06,
+      "loss": 1.0863,
+      "step": 15433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5300974584368435e-06,
+      "loss": 0.886,
+      "step": 15434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5300019109497425e-06,
+      "loss": 0.6429,
+      "step": 15435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.529906363462641e-06,
+      "loss": 0.6441,
+      "step": 15436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.52981081597554e-06,
+      "loss": 0.8253,
+      "step": 15437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5297152684884388e-06,
+      "loss": 0.7614,
+      "step": 15438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.529619721001338e-06,
+      "loss": 1.1288,
+      "step": 15439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5295241735142364e-06,
+      "loss": 0.7162,
+      "step": 15440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.529428626027136e-06,
+      "loss": 1.165,
+      "step": 15441
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.529333078540035e-06,
+      "loss": 0.9686,
+      "step": 15442
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5292375310529335e-06,
+      "loss": 0.65,
+      "step": 15443
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5291419835658326e-06,
+      "loss": 1.1022,
+      "step": 15444
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.529046436078731e-06,
+      "loss": 0.5928,
+      "step": 15445
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5289508885916302e-06,
+      "loss": 1.3847,
+      "step": 15446
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.528855341104529e-06,
+      "loss": 1.1882,
+      "step": 15447
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5287597936174283e-06,
+      "loss": 0.9164,
+      "step": 15448
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5286642461303273e-06,
+      "loss": 0.8949,
+      "step": 15449
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.528568698643226e-06,
+      "loss": 0.9999,
+      "step": 15450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.528473151156125e-06,
+      "loss": 0.839,
+      "step": 15451
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5283776036690236e-06,
+      "loss": 0.6846,
+      "step": 15452
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5282820561819226e-06,
+      "loss": 0.8368,
+      "step": 15453
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5281865086948213e-06,
+      "loss": 1.4316,
+      "step": 15454
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5280909612077207e-06,
+      "loss": 0.8382,
+      "step": 15455
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5279954137206198e-06,
+      "loss": 0.7422,
+      "step": 15456
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5278998662335184e-06,
+      "loss": 1.4859,
+      "step": 15457
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5278043187464174e-06,
+      "loss": 0.7887,
+      "step": 15458
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527708771259316e-06,
+      "loss": 1.1393,
+      "step": 15459
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527613223772215e-06,
+      "loss": 1.2141,
+      "step": 15460
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5275176762851137e-06,
+      "loss": 0.8525,
+      "step": 15461
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527422128798013e-06,
+      "loss": 1.4428,
+      "step": 15462
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527326581310912e-06,
+      "loss": 0.7211,
+      "step": 15463
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527231033823811e-06,
+      "loss": 0.9503,
+      "step": 15464
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.52713548633671e-06,
+      "loss": 0.5837,
+      "step": 15465
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5270399388496085e-06,
+      "loss": 0.8104,
+      "step": 15466
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5269443913625075e-06,
+      "loss": 0.7737,
+      "step": 15467
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526848843875406e-06,
+      "loss": 1.0359,
+      "step": 15468
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5267532963883056e-06,
+      "loss": 1.174,
+      "step": 15469
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526657748901204e-06,
+      "loss": 1.3175,
+      "step": 15470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5265622014141032e-06,
+      "loss": 1.2014,
+      "step": 15471
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526466653927002e-06,
+      "loss": 1.62,
+      "step": 15472
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526371106439901e-06,
+      "loss": 0.7637,
+      "step": 15473
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5262755589528e-06,
+      "loss": 0.7654,
+      "step": 15474
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5261800114656985e-06,
+      "loss": 1.014,
+      "step": 15475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5260844639785976e-06,
+      "loss": 0.8778,
+      "step": 15476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5259889164914966e-06,
+      "loss": 1.0391,
+      "step": 15477
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5258933690043956e-06,
+      "loss": 0.9414,
+      "step": 15478
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5257978215172943e-06,
+      "loss": 0.713,
+      "step": 15479
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5257022740301933e-06,
+      "loss": 1.3074,
+      "step": 15480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.525606726543092e-06,
+      "loss": 0.708,
+      "step": 15481
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.525511179055991e-06,
+      "loss": 1.2356,
+      "step": 15482
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5254156315688896e-06,
+      "loss": 0.562,
+      "step": 15483
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.525320084081789e-06,
+      "loss": 0.8647,
+      "step": 15484
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.525224536594688e-06,
+      "loss": 0.6027,
+      "step": 15485
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5251289891075867e-06,
+      "loss": 0.9201,
+      "step": 15486
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5250334416204857e-06,
+      "loss": 1.0247,
+      "step": 15487
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5249378941333843e-06,
+      "loss": 0.7378,
+      "step": 15488
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5248423466462834e-06,
+      "loss": 1.1622,
+      "step": 15489
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.524746799159182e-06,
+      "loss": 1.1191,
+      "step": 15490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5246512516720815e-06,
+      "loss": 0.8634,
+      "step": 15491
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5245557041849805e-06,
+      "loss": 1.1399,
+      "step": 15492
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.524460156697879e-06,
+      "loss": 1.1102,
+      "step": 15493
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.524364609210778e-06,
+      "loss": 0.5872,
+      "step": 15494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5242690617236768e-06,
+      "loss": 0.9567,
+      "step": 15495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.524173514236576e-06,
+      "loss": 0.8871,
+      "step": 15496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5240779667494744e-06,
+      "loss": 0.6962,
+      "step": 15497
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523982419262374e-06,
+      "loss": 0.4704,
+      "step": 15498
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523886871775273e-06,
+      "loss": 1.1568,
+      "step": 15499
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5237913242881715e-06,
+      "loss": 1.0914,
+      "step": 15500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5236957768010706e-06,
+      "loss": 1.1456,
+      "step": 15501
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523600229313969e-06,
+      "loss": 1.19,
+      "step": 15502
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5235046818268682e-06,
+      "loss": 0.7968,
+      "step": 15503
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523409134339767e-06,
+      "loss": 0.8173,
+      "step": 15504
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5233135868526663e-06,
+      "loss": 1.0342,
+      "step": 15505
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5232180393655653e-06,
+      "loss": 0.5315,
+      "step": 15506
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523122491878464e-06,
+      "loss": 1.273,
+      "step": 15507
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.523026944391363e-06,
+      "loss": 0.6227,
+      "step": 15508
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5229313969042616e-06,
+      "loss": 0.6754,
+      "step": 15509
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5228358494171606e-06,
+      "loss": 0.5919,
+      "step": 15510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5227403019300593e-06,
+      "loss": 0.7332,
+      "step": 15511
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5226447544429583e-06,
+      "loss": 1.0919,
+      "step": 15512
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5225492069558578e-06,
+      "loss": 0.7738,
+      "step": 15513
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5224536594687564e-06,
+      "loss": 0.854,
+      "step": 15514
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5223581119816554e-06,
+      "loss": 0.6039,
+      "step": 15515
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.522262564494554e-06,
+      "loss": 0.9942,
+      "step": 15516
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.522167017007453e-06,
+      "loss": 0.7921,
+      "step": 15517
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5220714695203517e-06,
+      "loss": 0.9672,
+      "step": 15518
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5219759220332507e-06,
+      "loss": 0.9514,
+      "step": 15519
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5218803745461498e-06,
+      "loss": 1.0096,
+      "step": 15520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521784827059049e-06,
+      "loss": 1.1241,
+      "step": 15521
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5216892795719474e-06,
+      "loss": 1.2104,
+      "step": 15522
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5215937320848465e-06,
+      "loss": 0.9379,
+      "step": 15523
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5214981845977455e-06,
+      "loss": 0.6578,
+      "step": 15524
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521402637110644e-06,
+      "loss": 0.8063,
+      "step": 15525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521307089623543e-06,
+      "loss": 1.204,
+      "step": 15526
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521211542136442e-06,
+      "loss": 1.1771,
+      "step": 15527
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5211159946493412e-06,
+      "loss": 1.1327,
+      "step": 15528
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.52102044716224e-06,
+      "loss": 0.8872,
+      "step": 15529
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.520924899675139e-06,
+      "loss": 1.3495,
+      "step": 15530
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5208293521880375e-06,
+      "loss": 1.0269,
+      "step": 15531
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5207338047009365e-06,
+      "loss": 0.665,
+      "step": 15532
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.520638257213835e-06,
+      "loss": 0.7676,
+      "step": 15533
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5205427097267346e-06,
+      "loss": 0.9651,
+      "step": 15534
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5204471622396336e-06,
+      "loss": 0.8668,
+      "step": 15535
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5203516147525323e-06,
+      "loss": 0.9992,
+      "step": 15536
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5202560672654313e-06,
+      "loss": 1.1531,
+      "step": 15537
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.52016051977833e-06,
+      "loss": 0.9466,
+      "step": 15538
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.520064972291229e-06,
+      "loss": 0.9969,
+      "step": 15539
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5199694248041276e-06,
+      "loss": 1.1346,
+      "step": 15540
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.519873877317027e-06,
+      "loss": 0.8101,
+      "step": 15541
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.519778329829926e-06,
+      "loss": 1.1697,
+      "step": 15542
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5196827823428247e-06,
+      "loss": 0.9273,
+      "step": 15543
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5195872348557237e-06,
+      "loss": 0.6497,
+      "step": 15544
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5194916873686223e-06,
+      "loss": 1.3131,
+      "step": 15545
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5193961398815214e-06,
+      "loss": 1.1615,
+      "step": 15546
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.51930059239442e-06,
+      "loss": 0.7478,
+      "step": 15547
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.519205044907319e-06,
+      "loss": 1.6517,
+      "step": 15548
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5191094974202185e-06,
+      "loss": 0.4653,
+      "step": 15549
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.519013949933117e-06,
+      "loss": 0.7808,
+      "step": 15550
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.518918402446016e-06,
+      "loss": 0.985,
+      "step": 15551
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5188228549589148e-06,
+      "loss": 1.1018,
+      "step": 15552
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.518727307471814e-06,
+      "loss": 1.0095,
+      "step": 15553
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5186317599847124e-06,
+      "loss": 1.1146,
+      "step": 15554
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5185362124976115e-06,
+      "loss": 0.9401,
+      "step": 15555
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.518440665010511e-06,
+      "loss": 0.8971,
+      "step": 15556
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5183451175234095e-06,
+      "loss": 1.2974,
+      "step": 15557
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5182495700363086e-06,
+      "loss": 0.6649,
+      "step": 15558
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.518154022549207e-06,
+      "loss": 1.1838,
+      "step": 15559
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5180584750621062e-06,
+      "loss": 1.168,
+      "step": 15560
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517962927575005e-06,
+      "loss": 1.0422,
+      "step": 15561
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517867380087904e-06,
+      "loss": 0.9098,
+      "step": 15562
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5177718326008033e-06,
+      "loss": 0.9251,
+      "step": 15563
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517676285113702e-06,
+      "loss": 1.2228,
+      "step": 15564
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517580737626601e-06,
+      "loss": 1.353,
+      "step": 15565
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5174851901394996e-06,
+      "loss": 0.7189,
+      "step": 15566
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5173896426523986e-06,
+      "loss": 0.8382,
+      "step": 15567
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5172940951652973e-06,
+      "loss": 0.9726,
+      "step": 15568
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5171985476781963e-06,
+      "loss": 1.1504,
+      "step": 15569
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5171030001910953e-06,
+      "loss": 1.5245,
+      "step": 15570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5170074527039944e-06,
+      "loss": 0.8707,
+      "step": 15571
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516911905216893e-06,
+      "loss": 0.8347,
+      "step": 15572
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516816357729792e-06,
+      "loss": 0.7596,
+      "step": 15573
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5167208102426906e-06,
+      "loss": 1.0177,
+      "step": 15574
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5166252627555897e-06,
+      "loss": 0.7417,
+      "step": 15575
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5165297152684887e-06,
+      "loss": 1.3724,
+      "step": 15576
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5164341677813878e-06,
+      "loss": 0.8371,
+      "step": 15577
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516338620294287e-06,
+      "loss": 0.5383,
+      "step": 15578
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5162430728071854e-06,
+      "loss": 1.0145,
+      "step": 15579
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5161475253200845e-06,
+      "loss": 1.0765,
+      "step": 15580
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516051977832983e-06,
+      "loss": 0.879,
+      "step": 15581
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.515956430345882e-06,
+      "loss": 1.3034,
+      "step": 15582
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5158608828587807e-06,
+      "loss": 1.6837,
+      "step": 15583
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5157653353716798e-06,
+      "loss": 0.9957,
+      "step": 15584
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5156697878845792e-06,
+      "loss": 0.9084,
+      "step": 15585
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515574240397478e-06,
+      "loss": 0.8956,
+      "step": 15586
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515478692910377e-06,
+      "loss": 0.9962,
+      "step": 15587
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5153831454232755e-06,
+      "loss": 1.206,
+      "step": 15588
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5152875979361745e-06,
+      "loss": 0.9932,
+      "step": 15589
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515192050449073e-06,
+      "loss": 0.8769,
+      "step": 15590
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.515096502961972e-06,
+      "loss": 0.9188,
+      "step": 15591
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5150009554748716e-06,
+      "loss": 0.9287,
+      "step": 15592
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5149054079877703e-06,
+      "loss": 1.1287,
+      "step": 15593
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5148098605006693e-06,
+      "loss": 0.7177,
+      "step": 15594
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514714313013568e-06,
+      "loss": 1.1916,
+      "step": 15595
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514618765526467e-06,
+      "loss": 0.9145,
+      "step": 15596
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5145232180393656e-06,
+      "loss": 0.8938,
+      "step": 15597
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5144276705522646e-06,
+      "loss": 0.8335,
+      "step": 15598
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.514332123065164e-06,
+      "loss": 0.835,
+      "step": 15599
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5142365755780627e-06,
+      "loss": 0.9569,
+      "step": 15600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5141410280909617e-06,
+      "loss": 1.1344,
+      "step": 15601
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5140454806038603e-06,
+      "loss": 1.1024,
+      "step": 15602
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5139499331167594e-06,
+      "loss": 0.8739,
+      "step": 15603
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.513854385629658e-06,
+      "loss": 1.2804,
+      "step": 15604
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.513758838142557e-06,
+      "loss": 0.9659,
+      "step": 15605
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5136632906554565e-06,
+      "loss": 1.0239,
+      "step": 15606
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.513567743168355e-06,
+      "loss": 1.1645,
+      "step": 15607
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.513472195681254e-06,
+      "loss": 0.8197,
+      "step": 15608
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5133766481941528e-06,
+      "loss": 0.8932,
+      "step": 15609
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.513281100707052e-06,
+      "loss": 1.4045,
+      "step": 15610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5131855532199504e-06,
+      "loss": 1.0516,
+      "step": 15611
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5130900057328495e-06,
+      "loss": 1.2841,
+      "step": 15612
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5129944582457485e-06,
+      "loss": 0.8591,
+      "step": 15613
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5128989107586475e-06,
+      "loss": 1.3483,
+      "step": 15614
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5128033632715466e-06,
+      "loss": 0.653,
+      "step": 15615
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.512707815784445e-06,
+      "loss": 0.6638,
+      "step": 15616
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5126122682973442e-06,
+      "loss": 0.7237,
+      "step": 15617
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.512516720810243e-06,
+      "loss": 1.0984,
+      "step": 15618
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.512421173323142e-06,
+      "loss": 0.8764,
+      "step": 15619
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5123256258360405e-06,
+      "loss": 1.4192,
+      "step": 15620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.51223007834894e-06,
+      "loss": 0.9587,
+      "step": 15621
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5121345308618386e-06,
+      "loss": 0.5609,
+      "step": 15622
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5120389833747376e-06,
+      "loss": 0.6628,
+      "step": 15623
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5119434358876362e-06,
+      "loss": 0.8156,
+      "step": 15624
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5118478884005353e-06,
+      "loss": 1.0988,
+      "step": 15625
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5117523409134343e-06,
+      "loss": 1.1337,
+      "step": 15626
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511656793426333e-06,
+      "loss": 1.2092,
+      "step": 15627
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5115612459392324e-06,
+      "loss": 0.628,
+      "step": 15628
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511465698452131e-06,
+      "loss": 0.9569,
+      "step": 15629
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.51137015096503e-06,
+      "loss": 0.8892,
+      "step": 15630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5112746034779286e-06,
+      "loss": 0.9089,
+      "step": 15631
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5111790559908277e-06,
+      "loss": 1.0782,
+      "step": 15632
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5110835085037263e-06,
+      "loss": 1.0829,
+      "step": 15633
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5109879610166253e-06,
+      "loss": 1.3032,
+      "step": 15634
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510892413529525e-06,
+      "loss": 1.3985,
+      "step": 15635
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5107968660424234e-06,
+      "loss": 0.8583,
+      "step": 15636
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5107013185553225e-06,
+      "loss": 1.1756,
+      "step": 15637
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510605771068221e-06,
+      "loss": 0.78,
+      "step": 15638
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.51051022358112e-06,
+      "loss": 0.8126,
+      "step": 15639
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5104146760940187e-06,
+      "loss": 0.5776,
+      "step": 15640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5103191286069178e-06,
+      "loss": 0.789,
+      "step": 15641
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5102235811198172e-06,
+      "loss": 1.1933,
+      "step": 15642
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510128033632716e-06,
+      "loss": 1.0098,
+      "step": 15643
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.510032486145615e-06,
+      "loss": 0.6204,
+      "step": 15644
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5099369386585135e-06,
+      "loss": 1.2154,
+      "step": 15645
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5098413911714125e-06,
+      "loss": 1.2011,
+      "step": 15646
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509745843684311e-06,
+      "loss": 1.1237,
+      "step": 15647
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.50965029619721e-06,
+      "loss": 0.9119,
+      "step": 15648
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5095547487101096e-06,
+      "loss": 0.9466,
+      "step": 15649
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5094592012230083e-06,
+      "loss": 0.9512,
+      "step": 15650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5093636537359073e-06,
+      "loss": 1.0627,
+      "step": 15651
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509268106248806e-06,
+      "loss": 0.8785,
+      "step": 15652
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.509172558761705e-06,
+      "loss": 0.7462,
+      "step": 15653
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5090770112746036e-06,
+      "loss": 0.8752,
+      "step": 15654
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5089814637875026e-06,
+      "loss": 1.1178,
+      "step": 15655
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5088859163004012e-06,
+      "loss": 1.1,
+      "step": 15656
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5087903688133007e-06,
+      "loss": 1.4304,
+      "step": 15657
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5086948213261997e-06,
+      "loss": 0.8664,
+      "step": 15658
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5085992738390983e-06,
+      "loss": 1.0946,
+      "step": 15659
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5085037263519974e-06,
+      "loss": 1.1265,
+      "step": 15660
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508408178864896e-06,
+      "loss": 0.7575,
+      "step": 15661
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508312631377795e-06,
+      "loss": 0.9066,
+      "step": 15662
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5082170838906936e-06,
+      "loss": 1.3771,
+      "step": 15663
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508121536403593e-06,
+      "loss": 0.9872,
+      "step": 15664
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.508025988916492e-06,
+      "loss": 0.8599,
+      "step": 15665
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5079304414293908e-06,
+      "loss": 0.9978,
+      "step": 15666
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.50783489394229e-06,
+      "loss": 0.7863,
+      "step": 15667
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5077393464551884e-06,
+      "loss": 0.915,
+      "step": 15668
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5076437989680874e-06,
+      "loss": 1.4078,
+      "step": 15669
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507548251480986e-06,
+      "loss": 0.9092,
+      "step": 15670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5074527039938855e-06,
+      "loss": 1.3954,
+      "step": 15671
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507357156506784e-06,
+      "loss": 1.0033,
+      "step": 15672
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507261609019683e-06,
+      "loss": 1.1415,
+      "step": 15673
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507166061532582e-06,
+      "loss": 0.853,
+      "step": 15674
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.507070514045481e-06,
+      "loss": 1.4011,
+      "step": 15675
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5069749665583794e-06,
+      "loss": 0.9429,
+      "step": 15676
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5068794190712785e-06,
+      "loss": 1.1395,
+      "step": 15677
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506783871584178e-06,
+      "loss": 0.7561,
+      "step": 15678
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5066883240970766e-06,
+      "loss": 1.4287,
+      "step": 15679
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5065927766099756e-06,
+      "loss": 0.9388,
+      "step": 15680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5064972291228742e-06,
+      "loss": 1.0659,
+      "step": 15681
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5064016816357733e-06,
+      "loss": 1.0753,
+      "step": 15682
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506306134148672e-06,
+      "loss": 1.8231,
+      "step": 15683
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506210586661571e-06,
+      "loss": 1.2319,
+      "step": 15684
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5061150391744704e-06,
+      "loss": 0.6922,
+      "step": 15685
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506019491687369e-06,
+      "loss": 1.2826,
+      "step": 15686
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.505923944200268e-06,
+      "loss": 0.8551,
+      "step": 15687
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5058283967131666e-06,
+      "loss": 0.8706,
+      "step": 15688
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5057328492260657e-06,
+      "loss": 0.7964,
+      "step": 15689
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5056373017389643e-06,
+      "loss": 0.6892,
+      "step": 15690
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5055417542518633e-06,
+      "loss": 0.7886,
+      "step": 15691
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505446206764762e-06,
+      "loss": 1.5348,
+      "step": 15692
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5053506592776614e-06,
+      "loss": 1.1676,
+      "step": 15693
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5052551117905604e-06,
+      "loss": 0.8228,
+      "step": 15694
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505159564303459e-06,
+      "loss": 1.1419,
+      "step": 15695
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.505064016816358e-06,
+      "loss": 0.8107,
+      "step": 15696
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5049684693292567e-06,
+      "loss": 0.8549,
+      "step": 15697
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5048729218421558e-06,
+      "loss": 0.9554,
+      "step": 15698
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5047773743550544e-06,
+      "loss": 0.8233,
+      "step": 15699
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504681826867954e-06,
+      "loss": 0.6196,
+      "step": 15700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504586279380853e-06,
+      "loss": 1.1755,
+      "step": 15701
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5044907318937515e-06,
+      "loss": 0.923,
+      "step": 15702
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5043951844066505e-06,
+      "loss": 1.0472,
+      "step": 15703
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504299636919549e-06,
+      "loss": 1.0816,
+      "step": 15704
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504204089432448e-06,
+      "loss": 0.9654,
+      "step": 15705
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.504108541945347e-06,
+      "loss": 0.6709,
+      "step": 15706
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5040129944582463e-06,
+      "loss": 1.2146,
+      "step": 15707
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5039174469711453e-06,
+      "loss": 0.7627,
+      "step": 15708
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.503821899484044e-06,
+      "loss": 0.9231,
+      "step": 15709
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.503726351996943e-06,
+      "loss": 1.1752,
+      "step": 15710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5036308045098416e-06,
+      "loss": 0.8783,
+      "step": 15711
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5035352570227406e-06,
+      "loss": 0.9528,
+      "step": 15712
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5034397095356392e-06,
+      "loss": 1.0372,
+      "step": 15713
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5033441620485387e-06,
+      "loss": 0.8503,
+      "step": 15714
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.6977066081187504,
+      "eval_loss": 1.053205132484436,
+      "eval_runtime": 140.7261,
+      "eval_samples_per_second": 297.763,
+      "eval_steps_per_second": 2.331,
+      "step": 15714
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5032486145614373e-06,
+      "loss": 0.734,
+      "step": 15715
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5031530670743363e-06,
+      "loss": 0.3516,
+      "step": 15716
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5030575195872354e-06,
+      "loss": 0.7033,
+      "step": 15717
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502961972100134e-06,
+      "loss": 0.728,
+      "step": 15718
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502866424613033e-06,
+      "loss": 1.1423,
+      "step": 15719
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5027708771259316e-06,
+      "loss": 0.757,
+      "step": 15720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502675329638831e-06,
+      "loss": 0.7251,
+      "step": 15721
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5025797821517297e-06,
+      "loss": 0.7003,
+      "step": 15722
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5024842346646288e-06,
+      "loss": 0.8422,
+      "step": 15723
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5023886871775274e-06,
+      "loss": 0.7665,
+      "step": 15724
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5022931396904264e-06,
+      "loss": 1.2654,
+      "step": 15725
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502197592203325e-06,
+      "loss": 0.7376,
+      "step": 15726
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502102044716224e-06,
+      "loss": 0.7926,
+      "step": 15727
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502006497229123e-06,
+      "loss": 1.0439,
+      "step": 15728
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501910949742022e-06,
+      "loss": 0.4542,
+      "step": 15729
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501815402254921e-06,
+      "loss": 0.929,
+      "step": 15730
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50171985476782e-06,
+      "loss": 0.8497,
+      "step": 15731
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501624307280719e-06,
+      "loss": 1.215,
+      "step": 15732
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5015287597936174e-06,
+      "loss": 0.7672,
+      "step": 15733
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5014332123065165e-06,
+      "loss": 0.6117,
+      "step": 15734
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501337664819415e-06,
+      "loss": 1.2818,
+      "step": 15735
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5012421173323146e-06,
+      "loss": 0.888,
+      "step": 15736
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5011465698452136e-06,
+      "loss": 0.6084,
+      "step": 15737
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5010510223581122e-06,
+      "loss": 0.5497,
+      "step": 15738
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5009554748710113e-06,
+      "loss": 1.0759,
+      "step": 15739
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.50085992738391e-06,
+      "loss": 0.5997,
+      "step": 15740
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.500764379896809e-06,
+      "loss": 1.019,
+      "step": 15741
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5006688324097075e-06,
+      "loss": 1.1305,
+      "step": 15742
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.500573284922607e-06,
+      "loss": 1.117,
+      "step": 15743
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.500477737435506e-06,
+      "loss": 0.8913,
+      "step": 15744
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5003821899484046e-06,
+      "loss": 1.0775,
+      "step": 15745
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5002866424613037e-06,
+      "loss": 0.6976,
+      "step": 15746
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5001910949742023e-06,
+      "loss": 0.7766,
+      "step": 15747
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5000955474871013e-06,
+      "loss": 0.8501,
+      "step": 15748
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.5e-06,
+      "loss": 0.9702,
+      "step": 15749
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4999044525128994e-06,
+      "loss": 0.7437,
+      "step": 15750
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4998089050257984e-06,
+      "loss": 1.077,
+      "step": 15751
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499713357538697e-06,
+      "loss": 0.9271,
+      "step": 15752
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499617810051596e-06,
+      "loss": 0.992,
+      "step": 15753
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4995222625644947e-06,
+      "loss": 1.1898,
+      "step": 15754
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4994267150773938e-06,
+      "loss": 0.8203,
+      "step": 15755
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4993311675902924e-06,
+      "loss": 0.7125,
+      "step": 15756
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499235620103192e-06,
+      "loss": 0.7023,
+      "step": 15757
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.499140072616091e-06,
+      "loss": 1.0007,
+      "step": 15758
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4990445251289895e-06,
+      "loss": 1.272,
+      "step": 15759
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4989489776418885e-06,
+      "loss": 0.7816,
+      "step": 15760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498853430154787e-06,
+      "loss": 1.7514,
+      "step": 15761
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498757882667686e-06,
+      "loss": 1.0715,
+      "step": 15762
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498662335180585e-06,
+      "loss": 0.8336,
+      "step": 15763
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498566787693484e-06,
+      "loss": 0.9903,
+      "step": 15764
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498471240206383e-06,
+      "loss": 1.1298,
+      "step": 15765
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.498375692719282e-06,
+      "loss": 0.8276,
+      "step": 15766
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4982801452321805e-06,
+      "loss": 0.7772,
+      "step": 15767
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4981845977450796e-06,
+      "loss": 0.8653,
+      "step": 15768
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4980890502579786e-06,
+      "loss": 1.146,
+      "step": 15769
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4979935027708772e-06,
+      "loss": 1.103,
+      "step": 15770
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4978979552837763e-06,
+      "loss": 1.0682,
+      "step": 15771
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4978024077966753e-06,
+      "loss": 1.2587,
+      "step": 15772
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4977068603095743e-06,
+      "loss": 0.4993,
+      "step": 15773
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497611312822473e-06,
+      "loss": 0.7327,
+      "step": 15774
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497515765335372e-06,
+      "loss": 0.8782,
+      "step": 15775
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4974202178482706e-06,
+      "loss": 0.7487,
+      "step": 15776
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4973246703611696e-06,
+      "loss": 0.6331,
+      "step": 15777
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4972291228740683e-06,
+      "loss": 1.0217,
+      "step": 15778
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4971335753869677e-06,
+      "loss": 1.1632,
+      "step": 15779
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4970380278998668e-06,
+      "loss": 0.5434,
+      "step": 15780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4969424804127654e-06,
+      "loss": 1.0414,
+      "step": 15781
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4968469329256644e-06,
+      "loss": 1.0732,
+      "step": 15782
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496751385438563e-06,
+      "loss": 0.8913,
+      "step": 15783
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496655837951462e-06,
+      "loss": 1.1147,
+      "step": 15784
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4965602904643607e-06,
+      "loss": 0.8457,
+      "step": 15785
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.49646474297726e-06,
+      "loss": 1.2343,
+      "step": 15786
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496369195490159e-06,
+      "loss": 1.2342,
+      "step": 15787
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496273648003058e-06,
+      "loss": 0.7177,
+      "step": 15788
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496178100515957e-06,
+      "loss": 0.8767,
+      "step": 15789
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4960825530288554e-06,
+      "loss": 0.9134,
+      "step": 15790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4959870055417545e-06,
+      "loss": 1.1328,
+      "step": 15791
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495891458054653e-06,
+      "loss": 1.0618,
+      "step": 15792
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4957959105675526e-06,
+      "loss": 0.9677,
+      "step": 15793
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4957003630804516e-06,
+      "loss": 0.6054,
+      "step": 15794
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4956048155933502e-06,
+      "loss": 1.5213,
+      "step": 15795
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4955092681062493e-06,
+      "loss": 0.6617,
+      "step": 15796
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.495413720619148e-06,
+      "loss": 1.3537,
+      "step": 15797
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.495318173132047e-06,
+      "loss": 0.8268,
+      "step": 15798
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4952226256449455e-06,
+      "loss": 0.551,
+      "step": 15799
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4951270781578446e-06,
+      "loss": 0.9439,
+      "step": 15800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.495031530670744e-06,
+      "loss": 1.0221,
+      "step": 15801
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4949359831836426e-06,
+      "loss": 0.6785,
+      "step": 15802
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4948404356965417e-06,
+      "loss": 0.8088,
+      "step": 15803
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4947448882094403e-06,
+      "loss": 0.7334,
+      "step": 15804
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4946493407223393e-06,
+      "loss": 0.9041,
+      "step": 15805
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.494553793235238e-06,
+      "loss": 1.1881,
+      "step": 15806
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.494458245748137e-06,
+      "loss": 1.0975,
+      "step": 15807
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4943626982610364e-06,
+      "loss": 1.3046,
+      "step": 15808
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.494267150773935e-06,
+      "loss": 0.7659,
+      "step": 15809
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.494171603286834e-06,
+      "loss": 0.6901,
+      "step": 15810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4940760557997327e-06,
+      "loss": 0.862,
+      "step": 15811
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4939805083126318e-06,
+      "loss": 0.6861,
+      "step": 15812
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4938849608255304e-06,
+      "loss": 1.0745,
+      "step": 15813
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4937894133384294e-06,
+      "loss": 0.8116,
+      "step": 15814
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4936938658513284e-06,
+      "loss": 0.272,
+      "step": 15815
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4935983183642275e-06,
+      "loss": 1.0809,
+      "step": 15816
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493502770877126e-06,
+      "loss": 1.1282,
+      "step": 15817
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493407223390025e-06,
+      "loss": 1.3862,
+      "step": 15818
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493311675902924e-06,
+      "loss": 1.0776,
+      "step": 15819
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493216128415823e-06,
+      "loss": 0.4965,
+      "step": 15820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493120580928722e-06,
+      "loss": 0.8622,
+      "step": 15821
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.493025033441621e-06,
+      "loss": 0.519,
+      "step": 15822
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.49292948595452e-06,
+      "loss": 1.272,
+      "step": 15823
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4928339384674185e-06,
+      "loss": 0.7936,
+      "step": 15824
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4927383909803176e-06,
+      "loss": 0.8361,
+      "step": 15825
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492642843493216e-06,
+      "loss": 1.0627,
+      "step": 15826
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4925472960061152e-06,
+      "loss": 0.7119,
+      "step": 15827
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492451748519014e-06,
+      "loss": 1.0696,
+      "step": 15828
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4923562010319133e-06,
+      "loss": 1.5774,
+      "step": 15829
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4922606535448123e-06,
+      "loss": 1.1239,
+      "step": 15830
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492165106057711e-06,
+      "loss": 0.7563,
+      "step": 15831
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.49206955857061e-06,
+      "loss": 1.3444,
+      "step": 15832
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4919740110835086e-06,
+      "loss": 0.9694,
+      "step": 15833
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4918784635964076e-06,
+      "loss": 0.9609,
+      "step": 15834
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4917829161093063e-06,
+      "loss": 0.7657,
+      "step": 15835
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4916873686222053e-06,
+      "loss": 1.0377,
+      "step": 15836
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4915918211351048e-06,
+      "loss": 1.0949,
+      "step": 15837
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4914962736480034e-06,
+      "loss": 1.343,
+      "step": 15838
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4914007261609024e-06,
+      "loss": 0.9961,
+      "step": 15839
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491305178673801e-06,
+      "loss": 0.5778,
+      "step": 15840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4912096311867e-06,
+      "loss": 0.9794,
+      "step": 15841
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4911140836995987e-06,
+      "loss": 0.9716,
+      "step": 15842
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4910185362124977e-06,
+      "loss": 1.1216,
+      "step": 15843
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.490922988725397e-06,
+      "loss": 1.0188,
+      "step": 15844
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490827441238296e-06,
+      "loss": 1.1543,
+      "step": 15845
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490731893751195e-06,
+      "loss": 1.5868,
+      "step": 15846
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4906363462640934e-06,
+      "loss": 1.1649,
+      "step": 15847
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4905407987769925e-06,
+      "loss": 1.0262,
+      "step": 15848
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.490445251289891e-06,
+      "loss": 1.1645,
+      "step": 15849
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.49034970380279e-06,
+      "loss": 0.7772,
+      "step": 15850
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4902541563156896e-06,
+      "loss": 0.9929,
+      "step": 15851
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4901586088285882e-06,
+      "loss": 0.6448,
+      "step": 15852
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4900630613414873e-06,
+      "loss": 0.9477,
+      "step": 15853
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.489967513854386e-06,
+      "loss": 0.6198,
+      "step": 15854
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.489871966367285e-06,
+      "loss": 1.1128,
+      "step": 15855
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4897764188801835e-06,
+      "loss": 1.134,
+      "step": 15856
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4896808713930826e-06,
+      "loss": 0.5394,
+      "step": 15857
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.489585323905982e-06,
+      "loss": 0.7657,
+      "step": 15858
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4894897764188806e-06,
+      "loss": 0.7534,
+      "step": 15859
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4893942289317797e-06,
+      "loss": 0.534,
+      "step": 15860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4892986814446783e-06,
+      "loss": 0.9205,
+      "step": 15861
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4892031339575773e-06,
+      "loss": 0.812,
+      "step": 15862
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.489107586470476e-06,
+      "loss": 0.7425,
+      "step": 15863
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.489012038983375e-06,
+      "loss": 0.7253,
+      "step": 15864
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.488916491496274e-06,
+      "loss": 0.7228,
+      "step": 15865
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.488820944009173e-06,
+      "loss": 0.6085,
+      "step": 15866
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4887253965220717e-06,
+      "loss": 0.899,
+      "step": 15867
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4886298490349707e-06,
+      "loss": 1.0502,
+      "step": 15868
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4885343015478693e-06,
+      "loss": 0.8505,
+      "step": 15869
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4884387540607684e-06,
+      "loss": 1.0145,
+      "step": 15870
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4883432065736674e-06,
+      "loss": 0.703,
+      "step": 15871
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.488247659086566e-06,
+      "loss": 0.9505,
+      "step": 15872
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4881521115994655e-06,
+      "loss": 0.8858,
+      "step": 15873
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.488056564112364e-06,
+      "loss": 0.7384,
+      "step": 15874
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487961016625263e-06,
+      "loss": 0.6557,
+      "step": 15875
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4878654691381618e-06,
+      "loss": 1.3016,
+      "step": 15876
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487769921651061e-06,
+      "loss": 0.6039,
+      "step": 15877
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4876743741639594e-06,
+      "loss": 1.0571,
+      "step": 15878
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4875788266768584e-06,
+      "loss": 0.5819,
+      "step": 15879
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487483279189758e-06,
+      "loss": 0.8375,
+      "step": 15880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4873877317026565e-06,
+      "loss": 0.629,
+      "step": 15881
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4872921842155556e-06,
+      "loss": 1.1811,
+      "step": 15882
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487196636728454e-06,
+      "loss": 1.1121,
+      "step": 15883
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4871010892413532e-06,
+      "loss": 0.6903,
+      "step": 15884
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.487005541754252e-06,
+      "loss": 0.925,
+      "step": 15885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.486909994267151e-06,
+      "loss": 0.9504,
+      "step": 15886
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4868144467800503e-06,
+      "loss": 0.944,
+      "step": 15887
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.486718899292949e-06,
+      "loss": 0.6283,
+      "step": 15888
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.486623351805848e-06,
+      "loss": 1.3789,
+      "step": 15889
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4865278043187466e-06,
+      "loss": 1.036,
+      "step": 15890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4864322568316456e-06,
+      "loss": 0.7193,
+      "step": 15891
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4863367093445443e-06,
+      "loss": 0.5582,
+      "step": 15892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4862411618574433e-06,
+      "loss": 0.5891,
+      "step": 15893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4861456143703428e-06,
+      "loss": 0.9473,
+      "step": 15894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4860500668832414e-06,
+      "loss": 0.6955,
+      "step": 15895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4859545193961404e-06,
+      "loss": 1.3089,
+      "step": 15896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485858971909039e-06,
+      "loss": 0.7746,
+      "step": 15897
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.485763424421938e-06,
+      "loss": 0.5462,
+      "step": 15898
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4856678769348367e-06,
+      "loss": 0.8627,
+      "step": 15899
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4855723294477357e-06,
+      "loss": 0.9745,
+      "step": 15900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.485476781960635e-06,
+      "loss": 0.5759,
+      "step": 15901
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.485381234473534e-06,
+      "loss": 0.7702,
+      "step": 15902
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.485285686986433e-06,
+      "loss": 1.1304,
+      "step": 15903
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4851901394993314e-06,
+      "loss": 1.0597,
+      "step": 15904
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4850945920122305e-06,
+      "loss": 1.1206,
+      "step": 15905
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.484999044525129e-06,
+      "loss": 0.9978,
+      "step": 15906
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.484903497038028e-06,
+      "loss": 0.4487,
+      "step": 15907
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4848079495509268e-06,
+      "loss": 0.819,
+      "step": 15908
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4847124020638262e-06,
+      "loss": 1.2432,
+      "step": 15909
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4846168545767253e-06,
+      "loss": 0.8069,
+      "step": 15910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.484521307089624e-06,
+      "loss": 1.0951,
+      "step": 15911
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.484425759602523e-06,
+      "loss": 0.5601,
+      "step": 15912
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4843302121154215e-06,
+      "loss": 1.026,
+      "step": 15913
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4842346646283206e-06,
+      "loss": 1.037,
+      "step": 15914
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.484139117141219e-06,
+      "loss": 0.754,
+      "step": 15915
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4840435696541186e-06,
+      "loss": 0.825,
+      "step": 15916
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4839480221670173e-06,
+      "loss": 0.9712,
+      "step": 15917
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4838524746799163e-06,
+      "loss": 1.1408,
+      "step": 15918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.483756927192815e-06,
+      "loss": 0.8001,
+      "step": 15919
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.483661379705714e-06,
+      "loss": 0.7101,
+      "step": 15920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.483565832218613e-06,
+      "loss": 1.1515,
+      "step": 15921
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4834702847315116e-06,
+      "loss": 1.0801,
+      "step": 15922
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.483374737244411e-06,
+      "loss": 0.943,
+      "step": 15923
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4832791897573097e-06,
+      "loss": 0.337,
+      "step": 15924
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4831836422702087e-06,
+      "loss": 1.0511,
+      "step": 15925
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4830880947831073e-06,
+      "loss": 0.8807,
+      "step": 15926
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4829925472960064e-06,
+      "loss": 0.984,
+      "step": 15927
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482896999808905e-06,
+      "loss": 0.7353,
+      "step": 15928
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482801452321804e-06,
+      "loss": 0.8031,
+      "step": 15929
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4827059048347035e-06,
+      "loss": 1.0823,
+      "step": 15930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482610357347602e-06,
+      "loss": 0.7582,
+      "step": 15931
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482514809860501e-06,
+      "loss": 0.6741,
+      "step": 15932
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4824192623733998e-06,
+      "loss": 1.0436,
+      "step": 15933
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482323714886299e-06,
+      "loss": 1.657,
+      "step": 15934
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4822281673991974e-06,
+      "loss": 0.8944,
+      "step": 15935
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4821326199120964e-06,
+      "loss": 0.6901,
+      "step": 15936
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482037072424996e-06,
+      "loss": 0.7162,
+      "step": 15937
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4819415249378945e-06,
+      "loss": 1.4408,
+      "step": 15938
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4818459774507936e-06,
+      "loss": 0.9068,
+      "step": 15939
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481750429963692e-06,
+      "loss": 0.6767,
+      "step": 15940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4816548824765912e-06,
+      "loss": 1.1746,
+      "step": 15941
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.48155933498949e-06,
+      "loss": 1.2305,
+      "step": 15942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481463787502389e-06,
+      "loss": 0.8159,
+      "step": 15943
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4813682400152875e-06,
+      "loss": 1.2873,
+      "step": 15944
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481272692528187e-06,
+      "loss": 0.8322,
+      "step": 15945
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481177145041086e-06,
+      "loss": 0.4341,
+      "step": 15946
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4810815975539846e-06,
+      "loss": 1.5756,
+      "step": 15947
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4809860500668836e-06,
+      "loss": 1.0802,
+      "step": 15948
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4808905025797823e-06,
+      "loss": 1.074,
+      "step": 15949
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4807949550926813e-06,
+      "loss": 0.6881,
+      "step": 15950
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.48069940760558e-06,
+      "loss": 0.7861,
+      "step": 15951
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4806038601184794e-06,
+      "loss": 0.5045,
+      "step": 15952
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4805083126313784e-06,
+      "loss": 1.1969,
+      "step": 15953
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480412765144277e-06,
+      "loss": 0.7492,
+      "step": 15954
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480317217657176e-06,
+      "loss": 1.1467,
+      "step": 15955
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4802216701700747e-06,
+      "loss": 0.9626,
+      "step": 15956
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4801261226829737e-06,
+      "loss": 0.8589,
+      "step": 15957
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4800305751958723e-06,
+      "loss": 0.803,
+      "step": 15958
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479935027708772e-06,
+      "loss": 0.7608,
+      "step": 15959
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479839480221671e-06,
+      "loss": 0.8643,
+      "step": 15960
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4797439327345694e-06,
+      "loss": 0.6207,
+      "step": 15961
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4796483852474685e-06,
+      "loss": 1.2502,
+      "step": 15962
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479552837760367e-06,
+      "loss": 0.7355,
+      "step": 15963
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479457290273266e-06,
+      "loss": 0.9625,
+      "step": 15964
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4793617427861648e-06,
+      "loss": 0.7979,
+      "step": 15965
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4792661952990642e-06,
+      "loss": 0.9568,
+      "step": 15966
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479170647811963e-06,
+      "loss": 0.5338,
+      "step": 15967
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.479075100324862e-06,
+      "loss": 1.0693,
+      "step": 15968
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4789795528377605e-06,
+      "loss": 0.8377,
+      "step": 15969
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4788840053506595e-06,
+      "loss": 0.8158,
+      "step": 15970
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.478788457863558e-06,
+      "loss": 1.1824,
+      "step": 15971
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.478692910376457e-06,
+      "loss": 0.7635,
+      "step": 15972
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4785973628893566e-06,
+      "loss": 0.7115,
+      "step": 15973
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4785018154022553e-06,
+      "loss": 0.8873,
+      "step": 15974
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4784062679151543e-06,
+      "loss": 0.5689,
+      "step": 15975
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.478310720428053e-06,
+      "loss": 0.6676,
+      "step": 15976
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.478215172940952e-06,
+      "loss": 0.9066,
+      "step": 15977
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4781196254538506e-06,
+      "loss": 1.3459,
+      "step": 15978
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4780240779667496e-06,
+      "loss": 0.9503,
+      "step": 15979
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4779285304796482e-06,
+      "loss": 1.1007,
+      "step": 15980
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4778329829925477e-06,
+      "loss": 0.8412,
+      "step": 15981
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4777374355054467e-06,
+      "loss": 1.342,
+      "step": 15982
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4776418880183453e-06,
+      "loss": 0.6363,
+      "step": 15983
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4775463405312444e-06,
+      "loss": 0.9062,
+      "step": 15984
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.477450793044143e-06,
+      "loss": 1.0767,
+      "step": 15985
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.477355245557042e-06,
+      "loss": 0.7208,
+      "step": 15986
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4772596980699406e-06,
+      "loss": 1.3068,
+      "step": 15987
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.47716415058284e-06,
+      "loss": 1.4862,
+      "step": 15988
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.477068603095739e-06,
+      "loss": 1.3133,
+      "step": 15989
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4769730556086378e-06,
+      "loss": 0.9648,
+      "step": 15990
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476877508121537e-06,
+      "loss": 0.8249,
+      "step": 15991
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4767819606344354e-06,
+      "loss": 1.1834,
+      "step": 15992
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4766864131473344e-06,
+      "loss": 1.3217,
+      "step": 15993
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476590865660233e-06,
+      "loss": 1.1271,
+      "step": 15994
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4764953181731325e-06,
+      "loss": 0.9611,
+      "step": 15995
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4763997706860316e-06,
+      "loss": 0.969,
+      "step": 15996
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.47630422319893e-06,
+      "loss": 1.0091,
+      "step": 15997
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4762086757118292e-06,
+      "loss": 1.077,
+      "step": 15998
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476113128224728e-06,
+      "loss": 0.6456,
+      "step": 15999
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476017580737627e-06,
+      "loss": 0.9683,
+      "step": 16000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4759220332505255e-06,
+      "loss": 0.7038,
+      "step": 16001
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.475826485763425e-06,
+      "loss": 1.0973,
+      "step": 16002
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475730938276324e-06,
+      "loss": 0.9683,
+      "step": 16003
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4756353907892226e-06,
+      "loss": 0.6578,
+      "step": 16004
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4755398433021216e-06,
+      "loss": 0.8035,
+      "step": 16005
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4754442958150203e-06,
+      "loss": 1.1159,
+      "step": 16006
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4753487483279193e-06,
+      "loss": 0.9656,
+      "step": 16007
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475253200840818e-06,
+      "loss": 0.7999,
+      "step": 16008
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4751576533537174e-06,
+      "loss": 1.4276,
+      "step": 16009
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.475062105866616e-06,
+      "loss": 0.5548,
+      "step": 16010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.474966558379515e-06,
+      "loss": 0.5718,
+      "step": 16011
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.474871010892414e-06,
+      "loss": 1.0066,
+      "step": 16012
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4747754634053127e-06,
+      "loss": 1.3616,
+      "step": 16013
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4746799159182117e-06,
+      "loss": 1.0362,
+      "step": 16014
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4745843684311103e-06,
+      "loss": 1.1193,
+      "step": 16015
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4744888209440094e-06,
+      "loss": 0.969,
+      "step": 16016
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4743932734569084e-06,
+      "loss": 1.2237,
+      "step": 16017
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4742977259698074e-06,
+      "loss": 0.7859,
+      "step": 16018
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.474202178482706e-06,
+      "loss": 0.9109,
+      "step": 16019
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.474106630995605e-06,
+      "loss": 1.2075,
+      "step": 16020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4740110835085037e-06,
+      "loss": 0.6367,
+      "step": 16021
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4739155360214028e-06,
+      "loss": 1.0238,
+      "step": 16022
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.473819988534302e-06,
+      "loss": 0.7273,
+      "step": 16023
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.473724441047201e-06,
+      "loss": 0.7115,
+      "step": 16024
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4736288935601e-06,
+      "loss": 0.8296,
+      "step": 16025
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4735333460729985e-06,
+      "loss": 0.6997,
+      "step": 16026
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4734377985858975e-06,
+      "loss": 0.7621,
+      "step": 16027
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.473342251098796e-06,
+      "loss": 1.0238,
+      "step": 16028
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.473246703611695e-06,
+      "loss": 1.3233,
+      "step": 16029
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4731511561245938e-06,
+      "loss": 1.0997,
+      "step": 16030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4730556086374933e-06,
+      "loss": 0.9191,
+      "step": 16031
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4729600611503923e-06,
+      "loss": 0.9613,
+      "step": 16032
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.472864513663291e-06,
+      "loss": 1.0475,
+      "step": 16033
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47276896617619e-06,
+      "loss": 1.1788,
+      "step": 16034
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4726734186890886e-06,
+      "loss": 1.1606,
+      "step": 16035
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4725778712019876e-06,
+      "loss": 0.9925,
+      "step": 16036
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.472482323714886e-06,
+      "loss": 1.0062,
+      "step": 16037
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4723867762277857e-06,
+      "loss": 0.8379,
+      "step": 16038
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4722912287406847e-06,
+      "loss": 0.9699,
+      "step": 16039
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4721956812535833e-06,
+      "loss": 0.8771,
+      "step": 16040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4721001337664824e-06,
+      "loss": 0.7329,
+      "step": 16041
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.472004586279381e-06,
+      "loss": 1.3127,
+      "step": 16042
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47190903879228e-06,
+      "loss": 1.0229,
+      "step": 16043
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4718134913051786e-06,
+      "loss": 1.0796,
+      "step": 16044
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471717943818078e-06,
+      "loss": 1.1402,
+      "step": 16045
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471622396330977e-06,
+      "loss": 0.9728,
+      "step": 16046
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4715268488438757e-06,
+      "loss": 0.7774,
+      "step": 16047
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471431301356775e-06,
+      "loss": 0.6679,
+      "step": 16048
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4713357538696734e-06,
+      "loss": 1.1724,
+      "step": 16049
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4712402063825724e-06,
+      "loss": 0.7591,
+      "step": 16050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471144658895471e-06,
+      "loss": 0.7571,
+      "step": 16051
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.47104911140837e-06,
+      "loss": 1.202,
+      "step": 16052
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4709535639212696e-06,
+      "loss": 0.4962,
+      "step": 16053
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.470858016434168e-06,
+      "loss": 1.0774,
+      "step": 16054
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470762468947067e-06,
+      "loss": 0.8431,
+      "step": 16055
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470666921459966e-06,
+      "loss": 0.7405,
+      "step": 16056
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470571373972865e-06,
+      "loss": 1.0546,
+      "step": 16057
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4704758264857635e-06,
+      "loss": 0.7478,
+      "step": 16058
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4703802789986625e-06,
+      "loss": 0.7384,
+      "step": 16059
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4702847315115616e-06,
+      "loss": 0.9935,
+      "step": 16060
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4701891840244606e-06,
+      "loss": 1.0284,
+      "step": 16061
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.470093636537359e-06,
+      "loss": 0.8392,
+      "step": 16062
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4699980890502582e-06,
+      "loss": 1.1053,
+      "step": 16063
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4699025415631573e-06,
+      "loss": 1.056,
+      "step": 16064
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.469806994076056e-06,
+      "loss": 1.0099,
+      "step": 16065
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.469711446588955e-06,
+      "loss": 1.2816,
+      "step": 16066
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.469615899101854e-06,
+      "loss": 1.2935,
+      "step": 16067
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.469520351614753e-06,
+      "loss": 1.2218,
+      "step": 16068
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4694248041276516e-06,
+      "loss": 0.5448,
+      "step": 16069
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4693292566405507e-06,
+      "loss": 0.878,
+      "step": 16070
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4692337091534493e-06,
+      "loss": 1.0753,
+      "step": 16071
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4691381616663483e-06,
+      "loss": 0.5843,
+      "step": 16072
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.469042614179247e-06,
+      "loss": 0.9417,
+      "step": 16073
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4689470666921464e-06,
+      "loss": 0.4444,
+      "step": 16074
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4688515192050454e-06,
+      "loss": 0.6379,
+      "step": 16075
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.468755971717944e-06,
+      "loss": 0.6144,
+      "step": 16076
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.468660424230843e-06,
+      "loss": 0.8283,
+      "step": 16077
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4685648767437417e-06,
+      "loss": 0.8424,
+      "step": 16078
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4684693292566407e-06,
+      "loss": 0.8207,
+      "step": 16079
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4683737817695394e-06,
+      "loss": 1.1892,
+      "step": 16080
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.468278234282439e-06,
+      "loss": 0.6959,
+      "step": 16081
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.468182686795338e-06,
+      "loss": 0.7918,
+      "step": 16082
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4680871393082365e-06,
+      "loss": 0.716,
+      "step": 16083
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4679915918211355e-06,
+      "loss": 0.6818,
+      "step": 16084
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467896044334034e-06,
+      "loss": 0.6037,
+      "step": 16085
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467800496846933e-06,
+      "loss": 0.7809,
+      "step": 16086
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4677049493598318e-06,
+      "loss": 1.5645,
+      "step": 16087
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467609401872731e-06,
+      "loss": 0.775,
+      "step": 16088
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4675138543856303e-06,
+      "loss": 0.8418,
+      "step": 16089
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467418306898529e-06,
+      "loss": 0.9199,
+      "step": 16090
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467322759411428e-06,
+      "loss": 0.8423,
+      "step": 16091
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4672272119243266e-06,
+      "loss": 0.6273,
+      "step": 16092
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4671316644372256e-06,
+      "loss": 1.2195,
+      "step": 16093
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467036116950124e-06,
+      "loss": 1.1121,
+      "step": 16094
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4669405694630232e-06,
+      "loss": 0.7511,
+      "step": 16095
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4668450219759227e-06,
+      "loss": 0.9753,
+      "step": 16096
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4667494744888213e-06,
+      "loss": 0.9492,
+      "step": 16097
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4666539270017204e-06,
+      "loss": 0.8933,
+      "step": 16098
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466558379514619e-06,
+      "loss": 0.6608,
+      "step": 16099
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466462832027518e-06,
+      "loss": 0.9227,
+      "step": 16100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4663672845404166e-06,
+      "loss": 0.7025,
+      "step": 16101
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4662717370533157e-06,
+      "loss": 0.5444,
+      "step": 16102
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466176189566215e-06,
+      "loss": 0.9474,
+      "step": 16103
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4660806420791137e-06,
+      "loss": 0.9227,
+      "step": 16104
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4659850945920128e-06,
+      "loss": 1.2078,
+      "step": 16105
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4658895471049114e-06,
+      "loss": 0.7682,
+      "step": 16106
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4657939996178104e-06,
+      "loss": 1.2768,
+      "step": 16107
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465698452130709e-06,
+      "loss": 0.8269,
+      "step": 16108
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465602904643608e-06,
+      "loss": 0.8908,
+      "step": 16109
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465507357156507e-06,
+      "loss": 0.7019,
+      "step": 16110
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465411809669406e-06,
+      "loss": 0.7848,
+      "step": 16111
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4653162621823048e-06,
+      "loss": 1.0422,
+      "step": 16112
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465220714695204e-06,
+      "loss": 1.1954,
+      "step": 16113
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.465125167208103e-06,
+      "loss": 1.4222,
+      "step": 16114
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4650296197210015e-06,
+      "loss": 0.833,
+      "step": 16115
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4649340722339005e-06,
+      "loss": 0.9493,
+      "step": 16116
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4648385247467996e-06,
+      "loss": 0.7635,
+      "step": 16117
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4647429772596986e-06,
+      "loss": 0.7755,
+      "step": 16118
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.464647429772597e-06,
+      "loss": 1.0093,
+      "step": 16119
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4645518822854962e-06,
+      "loss": 1.2357,
+      "step": 16120
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.464456334798395e-06,
+      "loss": 0.736,
+      "step": 16121
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.464360787311294e-06,
+      "loss": 0.9872,
+      "step": 16122
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4642652398241925e-06,
+      "loss": 0.8057,
+      "step": 16123
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4641696923370916e-06,
+      "loss": 1.1422,
+      "step": 16124
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.464074144849991e-06,
+      "loss": 0.8594,
+      "step": 16125
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4639785973628896e-06,
+      "loss": 1.0446,
+      "step": 16126
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4638830498757887e-06,
+      "loss": 1.5303,
+      "step": 16127
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4637875023886873e-06,
+      "loss": 0.8621,
+      "step": 16128
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4636919549015863e-06,
+      "loss": 0.6541,
+      "step": 16129
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.463596407414485e-06,
+      "loss": 0.7574,
+      "step": 16130
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.463500859927384e-06,
+      "loss": 1.3113,
+      "step": 16131
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4634053124402834e-06,
+      "loss": 1.021,
+      "step": 16132
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.463309764953182e-06,
+      "loss": 1.4128,
+      "step": 16133
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.463214217466081e-06,
+      "loss": 0.9262,
+      "step": 16134
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4631186699789797e-06,
+      "loss": 0.9906,
+      "step": 16135
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4630231224918787e-06,
+      "loss": 0.5402,
+      "step": 16136
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4629275750047774e-06,
+      "loss": 0.8055,
+      "step": 16137
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4628320275176764e-06,
+      "loss": 0.8526,
+      "step": 16138
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462736480030576e-06,
+      "loss": 0.7583,
+      "step": 16139
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4626409325434745e-06,
+      "loss": 0.6492,
+      "step": 16140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4625453850563735e-06,
+      "loss": 0.6045,
+      "step": 16141
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462449837569272e-06,
+      "loss": 0.7608,
+      "step": 16142
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462354290082171e-06,
+      "loss": 1.1933,
+      "step": 16143
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4622587425950698e-06,
+      "loss": 0.9687,
+      "step": 16144
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462163195107969e-06,
+      "loss": 1.2121,
+      "step": 16145
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4620676476208683e-06,
+      "loss": 0.8189,
+      "step": 16146
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461972100133767e-06,
+      "loss": 0.7318,
+      "step": 16147
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461876552646666e-06,
+      "loss": 0.4696,
+      "step": 16148
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4617810051595646e-06,
+      "loss": 1.0514,
+      "step": 16149
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4616854576724636e-06,
+      "loss": 0.7706,
+      "step": 16150
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461589910185362e-06,
+      "loss": 0.5596,
+      "step": 16151
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4614943626982612e-06,
+      "loss": 0.9534,
+      "step": 16152
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4613988152111607e-06,
+      "loss": 0.8457,
+      "step": 16153
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4613032677240593e-06,
+      "loss": 0.8566,
+      "step": 16154
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4612077202369584e-06,
+      "loss": 0.8856,
+      "step": 16155
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461112172749857e-06,
+      "loss": 1.0847,
+      "step": 16156
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461016625262756e-06,
+      "loss": 0.7982,
+      "step": 16157
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4609210777756546e-06,
+      "loss": 1.0471,
+      "step": 16158
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4608255302885537e-06,
+      "loss": 1.3091,
+      "step": 16159
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4607299828014523e-06,
+      "loss": 0.9378,
+      "step": 16160
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4606344353143517e-06,
+      "loss": 0.6735,
+      "step": 16161
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4605388878272504e-06,
+      "loss": 0.7991,
+      "step": 16162
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4604433403401494e-06,
+      "loss": 1.3652,
+      "step": 16163
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.460347792853048e-06,
+      "loss": 1.2159,
+      "step": 16164
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.460252245365947e-06,
+      "loss": 0.9688,
+      "step": 16165
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.460156697878846e-06,
+      "loss": 1.1061,
+      "step": 16166
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4600611503917447e-06,
+      "loss": 0.8176,
+      "step": 16167
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459965602904644e-06,
+      "loss": 0.7987,
+      "step": 16168
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4598700554175428e-06,
+      "loss": 1.1051,
+      "step": 16169
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459774507930442e-06,
+      "loss": 0.6513,
+      "step": 16170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4596789604433404e-06,
+      "loss": 0.8078,
+      "step": 16171
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4595834129562395e-06,
+      "loss": 1.4079,
+      "step": 16172
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459487865469138e-06,
+      "loss": 0.7515,
+      "step": 16173
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459392317982037e-06,
+      "loss": 0.9056,
+      "step": 16174
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4592967704949366e-06,
+      "loss": 1.2296,
+      "step": 16175
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459201223007835e-06,
+      "loss": 0.5511,
+      "step": 16176
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4591056755207342e-06,
+      "loss": 0.9466,
+      "step": 16177
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.459010128033633e-06,
+      "loss": 1.0586,
+      "step": 16178
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.458914580546532e-06,
+      "loss": 1.1847,
+      "step": 16179
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4588190330594305e-06,
+      "loss": 0.7407,
+      "step": 16180
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4587234855723296e-06,
+      "loss": 1.0659,
+      "step": 16181
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.458627938085229e-06,
+      "loss": 1.3057,
+      "step": 16182
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4585323905981276e-06,
+      "loss": 0.6194,
+      "step": 16183
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4584368431110267e-06,
+      "loss": 0.8404,
+      "step": 16184
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4583412956239253e-06,
+      "loss": 0.736,
+      "step": 16185
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4582457481368243e-06,
+      "loss": 1.0854,
+      "step": 16186
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.458150200649723e-06,
+      "loss": 1.3293,
+      "step": 16187
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.458054653162622e-06,
+      "loss": 0.7359,
+      "step": 16188
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4579591056755214e-06,
+      "loss": 0.9347,
+      "step": 16189
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.45786355818842e-06,
+      "loss": 0.8863,
+      "step": 16190
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.457768010701319e-06,
+      "loss": 1.0696,
+      "step": 16191
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4576724632142177e-06,
+      "loss": 0.7831,
+      "step": 16192
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4575769157271167e-06,
+      "loss": 0.8288,
+      "step": 16193
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4574813682400154e-06,
+      "loss": 1.1037,
+      "step": 16194
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4573858207529144e-06,
+      "loss": 0.9968,
+      "step": 16195
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.457290273265813e-06,
+      "loss": 0.8324,
+      "step": 16196
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4571947257787125e-06,
+      "loss": 0.9789,
+      "step": 16197
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4570991782916115e-06,
+      "loss": 0.7514,
+      "step": 16198
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.45700363080451e-06,
+      "loss": 0.6637,
+      "step": 16199
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456908083317409e-06,
+      "loss": 0.976,
+      "step": 16200
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4568125358303078e-06,
+      "loss": 1.2999,
+      "step": 16201
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456716988343207e-06,
+      "loss": 1.0234,
+      "step": 16202
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4566214408561054e-06,
+      "loss": 1.1434,
+      "step": 16203
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456525893369005e-06,
+      "loss": 1.0779,
+      "step": 16204
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.456430345881904e-06,
+      "loss": 0.8979,
+      "step": 16205
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4563347983948026e-06,
+      "loss": 0.8267,
+      "step": 16206
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4562392509077016e-06,
+      "loss": 1.0091,
+      "step": 16207
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4561437034206e-06,
+      "loss": 1.1951,
+      "step": 16208
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4560481559334992e-06,
+      "loss": 1.0609,
+      "step": 16209
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.455952608446398e-06,
+      "loss": 1.1177,
+      "step": 16210
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4558570609592973e-06,
+      "loss": 1.3316,
+      "step": 16211
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.455761513472196e-06,
+      "loss": 1.4638,
+      "step": 16212
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.455665965985095e-06,
+      "loss": 0.8967,
+      "step": 16213
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4555704184979936e-06,
+      "loss": 1.0898,
+      "step": 16214
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4554748710108926e-06,
+      "loss": 0.8895,
+      "step": 16215
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4553793235237917e-06,
+      "loss": 1.2335,
+      "step": 16216
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4552837760366903e-06,
+      "loss": 1.2027,
+      "step": 16217
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4551882285495897e-06,
+      "loss": 0.6012,
+      "step": 16218
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4550926810624884e-06,
+      "loss": 0.7804,
+      "step": 16219
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4549971335753874e-06,
+      "loss": 0.9709,
+      "step": 16220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.454901586088286e-06,
+      "loss": 0.644,
+      "step": 16221
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.454806038601185e-06,
+      "loss": 0.9472,
+      "step": 16222
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4547104911140837e-06,
+      "loss": 0.6838,
+      "step": 16223
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4546149436269827e-06,
+      "loss": 1.3397,
+      "step": 16224
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.454519396139882e-06,
+      "loss": 1.0492,
+      "step": 16225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4544238486527808e-06,
+      "loss": 0.9435,
+      "step": 16226
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45432830116568e-06,
+      "loss": 0.9955,
+      "step": 16227
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4542327536785784e-06,
+      "loss": 1.0701,
+      "step": 16228
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4541372061914775e-06,
+      "loss": 1.0839,
+      "step": 16229
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.454041658704376e-06,
+      "loss": 0.7757,
+      "step": 16230
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.453946111217275e-06,
+      "loss": 0.7966,
+      "step": 16231
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4538505637301737e-06,
+      "loss": 0.8668,
+      "step": 16232
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.453755016243073e-06,
+      "loss": 1.1827,
+      "step": 16233
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4536594687559722e-06,
+      "loss": 0.659,
+      "step": 16234
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.453563921268871e-06,
+      "loss": 0.6651,
+      "step": 16235
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45346837378177e-06,
+      "loss": 0.7718,
+      "step": 16236
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4533728262946685e-06,
+      "loss": 0.7356,
+      "step": 16237
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4532772788075676e-06,
+      "loss": 1.2081,
+      "step": 16238
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.453181731320466e-06,
+      "loss": 1.1038,
+      "step": 16239
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4530861838333656e-06,
+      "loss": 0.7206,
+      "step": 16240
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4529906363462647e-06,
+      "loss": 0.5793,
+      "step": 16241
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4528950888591633e-06,
+      "loss": 1.1116,
+      "step": 16242
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4527995413720623e-06,
+      "loss": 0.542,
+      "step": 16243
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.452703993884961e-06,
+      "loss": 1.3596,
+      "step": 16244
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.45260844639786e-06,
+      "loss": 0.9916,
+      "step": 16245
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4525128989107586e-06,
+      "loss": 0.8124,
+      "step": 16246
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.452417351423658e-06,
+      "loss": 0.6067,
+      "step": 16247
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.452321803936557e-06,
+      "loss": 1.0418,
+      "step": 16248
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4522262564494557e-06,
+      "loss": 0.5421,
+      "step": 16249
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4521307089623547e-06,
+      "loss": 0.4942,
+      "step": 16250
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4520351614752534e-06,
+      "loss": 0.8716,
+      "step": 16251
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4519396139881524e-06,
+      "loss": 0.7013,
+      "step": 16252
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451844066501051e-06,
+      "loss": 1.0454,
+      "step": 16253
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4517485190139505e-06,
+      "loss": 1.1799,
+      "step": 16254
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4516529715268495e-06,
+      "loss": 1.1842,
+      "step": 16255
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451557424039748e-06,
+      "loss": 0.8842,
+      "step": 16256
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451461876552647e-06,
+      "loss": 0.869,
+      "step": 16257
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4513663290655458e-06,
+      "loss": 0.6979,
+      "step": 16258
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451270781578445e-06,
+      "loss": 0.8995,
+      "step": 16259
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4511752340913434e-06,
+      "loss": 0.8738,
+      "step": 16260
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451079686604243e-06,
+      "loss": 0.6719,
+      "step": 16261
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4509841391171415e-06,
+      "loss": 1.3949,
+      "step": 16262
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4508885916300406e-06,
+      "loss": 0.813,
+      "step": 16263
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450793044142939e-06,
+      "loss": 0.977,
+      "step": 16264
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450697496655838e-06,
+      "loss": 1.007,
+      "step": 16265
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450601949168737e-06,
+      "loss": 0.4262,
+      "step": 16266
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450506401681636e-06,
+      "loss": 1.2937,
+      "step": 16267
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450410854194535e-06,
+      "loss": 0.9227,
+      "step": 16268
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450315306707434e-06,
+      "loss": 0.8208,
+      "step": 16269
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.450219759220333e-06,
+      "loss": 0.8785,
+      "step": 16270
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4501242117332316e-06,
+      "loss": 1.0967,
+      "step": 16271
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4500286642461306e-06,
+      "loss": 1.1299,
+      "step": 16272
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4499331167590292e-06,
+      "loss": 1.116,
+      "step": 16273
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4498375692719283e-06,
+      "loss": 1.3226,
+      "step": 16274
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.449742021784827e-06,
+      "loss": 0.968,
+      "step": 16275
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4496464742977264e-06,
+      "loss": 1.0901,
+      "step": 16276
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4495509268106254e-06,
+      "loss": 0.6219,
+      "step": 16277
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.449455379323524e-06,
+      "loss": 0.8663,
+      "step": 16278
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.449359831836423e-06,
+      "loss": 0.4894,
+      "step": 16279
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4492642843493217e-06,
+      "loss": 0.5569,
+      "step": 16280
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4491687368622207e-06,
+      "loss": 0.8566,
+      "step": 16281
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4490731893751193e-06,
+      "loss": 0.795,
+      "step": 16282
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4489776418880188e-06,
+      "loss": 0.9186,
+      "step": 16283
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448882094400918e-06,
+      "loss": 0.9368,
+      "step": 16284
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4487865469138164e-06,
+      "loss": 0.8642,
+      "step": 16285
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4486909994267155e-06,
+      "loss": 0.52,
+      "step": 16286
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448595451939614e-06,
+      "loss": 1.0336,
+      "step": 16287
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448499904452513e-06,
+      "loss": 0.9108,
+      "step": 16288
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4484043569654117e-06,
+      "loss": 1.1229,
+      "step": 16289
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448308809478311e-06,
+      "loss": 1.2867,
+      "step": 16290
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4482132619912102e-06,
+      "loss": 0.8353,
+      "step": 16291
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448117714504109e-06,
+      "loss": 1.2611,
+      "step": 16292
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.448022167017008e-06,
+      "loss": 0.979,
+      "step": 16293
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4479266195299065e-06,
+      "loss": 1.0566,
+      "step": 16294
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4478310720428056e-06,
+      "loss": 1.1552,
+      "step": 16295
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.447735524555704e-06,
+      "loss": 1.0304,
+      "step": 16296
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4476399770686036e-06,
+      "loss": 0.961,
+      "step": 16297
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4475444295815027e-06,
+      "loss": 0.8191,
+      "step": 16298
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4474488820944013e-06,
+      "loss": 1.1771,
+      "step": 16299
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4473533346073003e-06,
+      "loss": 0.9826,
+      "step": 16300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.447257787120199e-06,
+      "loss": 0.6697,
+      "step": 16301
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.447162239633098e-06,
+      "loss": 0.7712,
+      "step": 16302
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4470666921459966e-06,
+      "loss": 0.5807,
+      "step": 16303
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4469711446588956e-06,
+      "loss": 1.2054,
+      "step": 16304
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4468755971717947e-06,
+      "loss": 0.8752,
+      "step": 16305
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4467800496846937e-06,
+      "loss": 0.7199,
+      "step": 16306
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4466845021975927e-06,
+      "loss": 0.9184,
+      "step": 16307
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4465889547104914e-06,
+      "loss": 1.2082,
+      "step": 16308
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4464934072233904e-06,
+      "loss": 0.7462,
+      "step": 16309
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446397859736289e-06,
+      "loss": 0.7807,
+      "step": 16310
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446302312249188e-06,
+      "loss": 0.7421,
+      "step": 16311
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446206764762087e-06,
+      "loss": 0.6303,
+      "step": 16312
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446111217274986e-06,
+      "loss": 0.5387,
+      "step": 16313
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4460156697878847e-06,
+      "loss": 1.1387,
+      "step": 16314
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4459201223007838e-06,
+      "loss": 1.326,
+      "step": 16315
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4458245748136824e-06,
+      "loss": 0.7205,
+      "step": 16316
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4457290273265814e-06,
+      "loss": 0.9782,
+      "step": 16317
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4456334798394805e-06,
+      "loss": 0.5146,
+      "step": 16318
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4455379323523795e-06,
+      "loss": 0.7829,
+      "step": 16319
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4454423848652786e-06,
+      "loss": 1.2401,
+      "step": 16320
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445346837378177e-06,
+      "loss": 1.0336,
+      "step": 16321
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445251289891076e-06,
+      "loss": 1.3179,
+      "step": 16322
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445155742403975e-06,
+      "loss": 0.6276,
+      "step": 16323
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.445060194916874e-06,
+      "loss": 1.227,
+      "step": 16324
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4449646474297725e-06,
+      "loss": 1.1383,
+      "step": 16325
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.444869099942672e-06,
+      "loss": 0.5944,
+      "step": 16326
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.444773552455571e-06,
+      "loss": 0.6522,
+      "step": 16327
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4446780049684696e-06,
+      "loss": 0.9876,
+      "step": 16328
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4445824574813686e-06,
+      "loss": 0.974,
+      "step": 16329
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4444869099942672e-06,
+      "loss": 1.0277,
+      "step": 16330
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4443913625071663e-06,
+      "loss": 0.665,
+      "step": 16331
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.444295815020065e-06,
+      "loss": 1.4492,
+      "step": 16332
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4442002675329644e-06,
+      "loss": 1.3037,
+      "step": 16333
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4441047200458634e-06,
+      "loss": 1.2131,
+      "step": 16334
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.444009172558762e-06,
+      "loss": 0.8786,
+      "step": 16335
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.443913625071661e-06,
+      "loss": 0.8825,
+      "step": 16336
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4438180775845597e-06,
+      "loss": 0.9124,
+      "step": 16337
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4437225300974587e-06,
+      "loss": 0.9293,
+      "step": 16338
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4436269826103573e-06,
+      "loss": 0.9642,
+      "step": 16339
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4435314351232564e-06,
+      "loss": 1.0634,
+      "step": 16340
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.443435887636156e-06,
+      "loss": 0.8079,
+      "step": 16341
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4433403401490544e-06,
+      "loss": 0.8175,
+      "step": 16342
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4432447926619535e-06,
+      "loss": 0.795,
+      "step": 16343
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.443149245174852e-06,
+      "loss": 0.8925,
+      "step": 16344
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.443053697687751e-06,
+      "loss": 0.6774,
+      "step": 16345
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4429581502006497e-06,
+      "loss": 0.7858,
+      "step": 16346
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4428626027135488e-06,
+      "loss": 0.819,
+      "step": 16347
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4427670552264482e-06,
+      "loss": 0.9734,
+      "step": 16348
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.442671507739347e-06,
+      "loss": 0.8146,
+      "step": 16349
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.442575960252246e-06,
+      "loss": 0.9428,
+      "step": 16350
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4424804127651445e-06,
+      "loss": 0.8597,
+      "step": 16351
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4423848652780436e-06,
+      "loss": 1.1262,
+      "step": 16352
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.442289317790942e-06,
+      "loss": 0.9881,
+      "step": 16353
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.442193770303841e-06,
+      "loss": 1.4653,
+      "step": 16354
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4420982228167402e-06,
+      "loss": 0.8176,
+      "step": 16355
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4420026753296393e-06,
+      "loss": 0.8928,
+      "step": 16356
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4419071278425383e-06,
+      "loss": 0.6987,
+      "step": 16357
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441811580355437e-06,
+      "loss": 0.4108,
+      "step": 16358
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441716032868336e-06,
+      "loss": 0.8933,
+      "step": 16359
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4416204853812346e-06,
+      "loss": 0.9848,
+      "step": 16360
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4415249378941336e-06,
+      "loss": 0.791,
+      "step": 16361
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4414293904070327e-06,
+      "loss": 1.0054,
+      "step": 16362
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4413338429199317e-06,
+      "loss": 0.4945,
+      "step": 16363
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4412382954328303e-06,
+      "loss": 1.0153,
+      "step": 16364
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4411427479457294e-06,
+      "loss": 1.0542,
+      "step": 16365
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441047200458628e-06,
+      "loss": 1.0068,
+      "step": 16366
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.440951652971527e-06,
+      "loss": 0.9885,
+      "step": 16367
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4408561054844256e-06,
+      "loss": 1.2061,
+      "step": 16368
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440760557997325e-06,
+      "loss": 0.8087,
+      "step": 16369
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440665010510224e-06,
+      "loss": 0.5931,
+      "step": 16370
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4405694630231227e-06,
+      "loss": 0.7838,
+      "step": 16371
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4404739155360218e-06,
+      "loss": 0.8878,
+      "step": 16372
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4403783680489204e-06,
+      "loss": 1.2672,
+      "step": 16373
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4402828205618194e-06,
+      "loss": 0.7239,
+      "step": 16374
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440187273074718e-06,
+      "loss": 0.9905,
+      "step": 16375
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.440091725587617e-06,
+      "loss": 1.1588,
+      "step": 16376
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4399961781005166e-06,
+      "loss": 0.6554,
+      "step": 16377
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.439900630613415e-06,
+      "loss": 0.9364,
+      "step": 16378
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.439805083126314e-06,
+      "loss": 0.5735,
+      "step": 16379
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.439709535639213e-06,
+      "loss": 0.9127,
+      "step": 16380
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.439613988152112e-06,
+      "loss": 1.1065,
+      "step": 16381
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4395184406650105e-06,
+      "loss": 1.0023,
+      "step": 16382
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4394228931779095e-06,
+      "loss": 0.5919,
+      "step": 16383
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.439327345690809e-06,
+      "loss": 0.6307,
+      "step": 16384
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4392317982037076e-06,
+      "loss": 1.1071,
+      "step": 16385
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4391362507166066e-06,
+      "loss": 0.8945,
+      "step": 16386
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4390407032295052e-06,
+      "loss": 1.0504,
+      "step": 16387
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4389451557424043e-06,
+      "loss": 0.5914,
+      "step": 16388
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.438849608255303e-06,
+      "loss": 0.7064,
+      "step": 16389
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.438754060768202e-06,
+      "loss": 0.7601,
+      "step": 16390
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4386585132811014e-06,
+      "loss": 1.0292,
+      "step": 16391
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.438562965794e-06,
+      "loss": 0.968,
+      "step": 16392
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.438467418306899e-06,
+      "loss": 1.0781,
+      "step": 16393
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4383718708197977e-06,
+      "loss": 0.8865,
+      "step": 16394
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4382763233326967e-06,
+      "loss": 0.6641,
+      "step": 16395
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4381807758455953e-06,
+      "loss": 1.3546,
+      "step": 16396
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4380852283584944e-06,
+      "loss": 0.7201,
+      "step": 16397
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437989680871394e-06,
+      "loss": 1.318,
+      "step": 16398
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4378941333842924e-06,
+      "loss": 1.0293,
+      "step": 16399
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4377985858971915e-06,
+      "loss": 0.7594,
+      "step": 16400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.43770303841009e-06,
+      "loss": 0.9328,
+      "step": 16401
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437607490922989e-06,
+      "loss": 0.5823,
+      "step": 16402
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4375119434358877e-06,
+      "loss": 0.8757,
+      "step": 16403
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4374163959487868e-06,
+      "loss": 1.1823,
+      "step": 16404
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437320848461686e-06,
+      "loss": 0.6542,
+      "step": 16405
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437225300974585e-06,
+      "loss": 0.8421,
+      "step": 16406
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4371297534874835e-06,
+      "loss": 0.7829,
+      "step": 16407
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4370342060003825e-06,
+      "loss": 0.8206,
+      "step": 16408
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4369386585132816e-06,
+      "loss": 0.9471,
+      "step": 16409
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.43684311102618e-06,
+      "loss": 0.6147,
+      "step": 16410
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436747563539079e-06,
+      "loss": 1.0555,
+      "step": 16411
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436652016051978e-06,
+      "loss": 1.1935,
+      "step": 16412
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4365564685648773e-06,
+      "loss": 0.7658,
+      "step": 16413
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436460921077776e-06,
+      "loss": 0.8385,
+      "step": 16414
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436365373590675e-06,
+      "loss": 1.0711,
+      "step": 16415
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4362698261035735e-06,
+      "loss": 0.8633,
+      "step": 16416
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4361742786164726e-06,
+      "loss": 0.9002,
+      "step": 16417
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.436078731129371e-06,
+      "loss": 1.3776,
+      "step": 16418
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4359831836422702e-06,
+      "loss": 0.8685,
+      "step": 16419
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4358876361551697e-06,
+      "loss": 1.4034,
+      "step": 16420
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4357920886680683e-06,
+      "loss": 0.8391,
+      "step": 16421
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4356965411809674e-06,
+      "loss": 0.8709,
+      "step": 16422
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435600993693866e-06,
+      "loss": 0.7992,
+      "step": 16423
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435505446206765e-06,
+      "loss": 0.947,
+      "step": 16424
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4354098987196636e-06,
+      "loss": 0.9307,
+      "step": 16425
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4353143512325627e-06,
+      "loss": 1.2495,
+      "step": 16426
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.435218803745462e-06,
+      "loss": 0.8542,
+      "step": 16427
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4351232562583607e-06,
+      "loss": 0.7939,
+      "step": 16428
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4350277087712598e-06,
+      "loss": 0.6342,
+      "step": 16429
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4349321612841584e-06,
+      "loss": 1.12,
+      "step": 16430
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4348366137970574e-06,
+      "loss": 1.1107,
+      "step": 16431
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.434741066309956e-06,
+      "loss": 0.6307,
+      "step": 16432
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.434645518822855e-06,
+      "loss": 0.8011,
+      "step": 16433
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4345499713357546e-06,
+      "loss": 0.9282,
+      "step": 16434
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.434454423848653e-06,
+      "loss": 0.777,
+      "step": 16435
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.434358876361552e-06,
+      "loss": 0.8971,
+      "step": 16436
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.434263328874451e-06,
+      "loss": 1.2005,
+      "step": 16437
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43416778138735e-06,
+      "loss": 0.804,
+      "step": 16438
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4340722339002485e-06,
+      "loss": 1.1916,
+      "step": 16439
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4339766864131475e-06,
+      "loss": 1.001,
+      "step": 16440
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.433881138926047e-06,
+      "loss": 1.1493,
+      "step": 16441
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4337855914389456e-06,
+      "loss": 0.7996,
+      "step": 16442
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4336900439518446e-06,
+      "loss": 0.4689,
+      "step": 16443
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4335944964647432e-06,
+      "loss": 0.7198,
+      "step": 16444
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4334989489776423e-06,
+      "loss": 0.656,
+      "step": 16445
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.433403401490541e-06,
+      "loss": 0.5913,
+      "step": 16446
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43330785400344e-06,
+      "loss": 0.5567,
+      "step": 16447
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4332123065163385e-06,
+      "loss": 0.9302,
+      "step": 16448
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.433116759029238e-06,
+      "loss": 0.7637,
+      "step": 16449
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.433021211542137e-06,
+      "loss": 0.915,
+      "step": 16450
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4329256640550357e-06,
+      "loss": 1.1679,
+      "step": 16451
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4328301165679347e-06,
+      "loss": 0.5711,
+      "step": 16452
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4327345690808333e-06,
+      "loss": 1.0748,
+      "step": 16453
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4326390215937324e-06,
+      "loss": 0.7939,
+      "step": 16454
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.432543474106631e-06,
+      "loss": 0.7885,
+      "step": 16455
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4324479266195304e-06,
+      "loss": 0.7358,
+      "step": 16456
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.432352379132429e-06,
+      "loss": 0.7373,
+      "step": 16457
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.432256831645328e-06,
+      "loss": 0.7328,
+      "step": 16458
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4321612841582267e-06,
+      "loss": 1.1954,
+      "step": 16459
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4320657366711257e-06,
+      "loss": 0.7149,
+      "step": 16460
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4319701891840248e-06,
+      "loss": 0.6471,
+      "step": 16461
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4318746416969234e-06,
+      "loss": 0.8221,
+      "step": 16462
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431779094209823e-06,
+      "loss": 0.7102,
+      "step": 16463
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4316835467227215e-06,
+      "loss": 0.6313,
+      "step": 16464
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4315879992356205e-06,
+      "loss": 0.8304,
+      "step": 16465
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431492451748519e-06,
+      "loss": 0.6345,
+      "step": 16466
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431396904261418e-06,
+      "loss": 1.1123,
+      "step": 16467
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4313013567743168e-06,
+      "loss": 1.2065,
+      "step": 16468
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431205809287216e-06,
+      "loss": 0.7366,
+      "step": 16469
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4311102618001153e-06,
+      "loss": 1.0491,
+      "step": 16470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.431014714313014e-06,
+      "loss": 0.9955,
+      "step": 16471
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.430919166825913e-06,
+      "loss": 0.7412,
+      "step": 16472
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4308236193388115e-06,
+      "loss": 0.9655,
+      "step": 16473
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4307280718517106e-06,
+      "loss": 0.5803,
+      "step": 16474
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430632524364609e-06,
+      "loss": 1.1682,
+      "step": 16475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4305369768775082e-06,
+      "loss": 0.5312,
+      "step": 16476
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4304414293904077e-06,
+      "loss": 0.7089,
+      "step": 16477
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4303458819033063e-06,
+      "loss": 1.5632,
+      "step": 16478
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4302503344162054e-06,
+      "loss": 0.8813,
+      "step": 16479
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430154786929104e-06,
+      "loss": 0.8353,
+      "step": 16480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.430059239442003e-06,
+      "loss": 1.1145,
+      "step": 16481
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4299636919549016e-06,
+      "loss": 0.6589,
+      "step": 16482
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4298681444678007e-06,
+      "loss": 0.9863,
+      "step": 16483
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4297725969806993e-06,
+      "loss": 0.7538,
+      "step": 16484
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4296770494935987e-06,
+      "loss": 1.048,
+      "step": 16485
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4295815020064978e-06,
+      "loss": 0.6177,
+      "step": 16486
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4294859545193964e-06,
+      "loss": 0.7098,
+      "step": 16487
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4293904070322954e-06,
+      "loss": 0.8274,
+      "step": 16488
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.429294859545194e-06,
+      "loss": 0.6539,
+      "step": 16489
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.429199312058093e-06,
+      "loss": 1.0073,
+      "step": 16490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4291037645709917e-06,
+      "loss": 1.4341,
+      "step": 16491
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.429008217083891e-06,
+      "loss": 0.9917,
+      "step": 16492
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.42891266959679e-06,
+      "loss": 0.9838,
+      "step": 16493
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.428817122109689e-06,
+      "loss": 0.8271,
+      "step": 16494
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.428721574622588e-06,
+      "loss": 0.7385,
+      "step": 16495
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4286260271354865e-06,
+      "loss": 1.0806,
+      "step": 16496
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4285304796483855e-06,
+      "loss": 0.5701,
+      "step": 16497
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.428434932161284e-06,
+      "loss": 1.0419,
+      "step": 16498
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4283393846741836e-06,
+      "loss": 0.936,
+      "step": 16499
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4282438371870826e-06,
+      "loss": 0.8412,
+      "step": 16500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4281482896999812e-06,
+      "loss": 1.221,
+      "step": 16501
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4280527422128803e-06,
+      "loss": 0.5968,
+      "step": 16502
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427957194725779e-06,
+      "loss": 0.8208,
+      "step": 16503
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427861647238678e-06,
+      "loss": 1.2769,
+      "step": 16504
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4277660997515765e-06,
+      "loss": 0.856,
+      "step": 16505
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427670552264476e-06,
+      "loss": 0.7776,
+      "step": 16506
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4275750047773746e-06,
+      "loss": 1.1235,
+      "step": 16507
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4274794572902737e-06,
+      "loss": 0.6404,
+      "step": 16508
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4273839098031723e-06,
+      "loss": 1.1488,
+      "step": 16509
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4272883623160713e-06,
+      "loss": 0.8329,
+      "step": 16510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4271928148289704e-06,
+      "loss": 0.6279,
+      "step": 16511
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.427097267341869e-06,
+      "loss": 1.0851,
+      "step": 16512
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4270017198547684e-06,
+      "loss": 0.7886,
+      "step": 16513
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426906172367667e-06,
+      "loss": 0.9584,
+      "step": 16514
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426810624880566e-06,
+      "loss": 1.2632,
+      "step": 16515
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4267150773934647e-06,
+      "loss": 0.5233,
+      "step": 16516
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4266195299063637e-06,
+      "loss": 1.2842,
+      "step": 16517
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4265239824192624e-06,
+      "loss": 1.0033,
+      "step": 16518
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4264284349321614e-06,
+      "loss": 0.8502,
+      "step": 16519
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.42633288744506e-06,
+      "loss": 0.6189,
+      "step": 16520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4262373399579595e-06,
+      "loss": 0.9673,
+      "step": 16521
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4261417924708585e-06,
+      "loss": 0.8203,
+      "step": 16522
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426046244983757e-06,
+      "loss": 1.0028,
+      "step": 16523
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.425950697496656e-06,
+      "loss": 1.1109,
+      "step": 16524
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4258551500095548e-06,
+      "loss": 0.7306,
+      "step": 16525
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425759602522454e-06,
+      "loss": 0.6657,
+      "step": 16526
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4256640550353524e-06,
+      "loss": 1.1084,
+      "step": 16527
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425568507548252e-06,
+      "loss": 0.6962,
+      "step": 16528
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425472960061151e-06,
+      "loss": 0.5599,
+      "step": 16529
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4253774125740495e-06,
+      "loss": 1.0521,
+      "step": 16530
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4252818650869486e-06,
+      "loss": 0.5791,
+      "step": 16531
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.425186317599847e-06,
+      "loss": 0.8682,
+      "step": 16532
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4250907701127462e-06,
+      "loss": 0.824,
+      "step": 16533
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.424995222625645e-06,
+      "loss": 0.7456,
+      "step": 16534
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4248996751385443e-06,
+      "loss": 0.6993,
+      "step": 16535
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4248041276514434e-06,
+      "loss": 1.0061,
+      "step": 16536
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.424708580164342e-06,
+      "loss": 0.7555,
+      "step": 16537
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.424613032677241e-06,
+      "loss": 0.8501,
+      "step": 16538
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4245174851901396e-06,
+      "loss": 1.0911,
+      "step": 16539
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4244219377030387e-06,
+      "loss": 0.8364,
+      "step": 16540
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4243263902159373e-06,
+      "loss": 0.9069,
+      "step": 16541
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4242308427288367e-06,
+      "loss": 0.7894,
+      "step": 16542
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4241352952417358e-06,
+      "loss": 0.8653,
+      "step": 16543
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4240397477546344e-06,
+      "loss": 0.7427,
+      "step": 16544
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4239442002675334e-06,
+      "loss": 0.985,
+      "step": 16545
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423848652780432e-06,
+      "loss": 0.98,
+      "step": 16546
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423753105293331e-06,
+      "loss": 0.873,
+      "step": 16547
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4236575578062297e-06,
+      "loss": 1.1449,
+      "step": 16548
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423562010319129e-06,
+      "loss": 0.8822,
+      "step": 16549
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423466462832028e-06,
+      "loss": 0.9738,
+      "step": 16550
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423370915344927e-06,
+      "loss": 1.028,
+      "step": 16551
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.423275367857826e-06,
+      "loss": 0.8499,
+      "step": 16552
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4231798203707245e-06,
+      "loss": 0.6764,
+      "step": 16553
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4230842728836235e-06,
+      "loss": 0.9468,
+      "step": 16554
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.422988725396522e-06,
+      "loss": 0.9344,
+      "step": 16555
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.422893177909421e-06,
+      "loss": 0.8294,
+      "step": 16556
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.42279763042232e-06,
+      "loss": 0.7526,
+      "step": 16557
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4227020829352192e-06,
+      "loss": 0.7474,
+      "step": 16558
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.422606535448118e-06,
+      "loss": 1.0421,
+      "step": 16559
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.422510987961017e-06,
+      "loss": 0.6532,
+      "step": 16560
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4224154404739155e-06,
+      "loss": 1.0812,
+      "step": 16561
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4223198929868145e-06,
+      "loss": 0.7207,
+      "step": 16562
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4222243454997136e-06,
+      "loss": 0.7199,
+      "step": 16563
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4221287980126126e-06,
+      "loss": 0.5475,
+      "step": 16564
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4220332505255117e-06,
+      "loss": 0.8402,
+      "step": 16565
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4219377030384103e-06,
+      "loss": 0.8083,
+      "step": 16566
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4218421555513093e-06,
+      "loss": 1.1914,
+      "step": 16567
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421746608064208e-06,
+      "loss": 1.35,
+      "step": 16568
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421651060577107e-06,
+      "loss": 0.8499,
+      "step": 16569
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4215555130900056e-06,
+      "loss": 0.694,
+      "step": 16570
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421459965602905e-06,
+      "loss": 1.0942,
+      "step": 16571
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421364418115804e-06,
+      "loss": 0.8295,
+      "step": 16572
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4212688706287027e-06,
+      "loss": 1.0902,
+      "step": 16573
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4211733231416017e-06,
+      "loss": 1.493,
+      "step": 16574
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4210777756545004e-06,
+      "loss": 0.9531,
+      "step": 16575
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4209822281673994e-06,
+      "loss": 1.0107,
+      "step": 16576
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.420886680680298e-06,
+      "loss": 1.0174,
+      "step": 16577
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4207911331931975e-06,
+      "loss": 0.7746,
+      "step": 16578
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4206955857060965e-06,
+      "loss": 0.4568,
+      "step": 16579
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420600038218995e-06,
+      "loss": 0.9195,
+      "step": 16580
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420504490731894e-06,
+      "loss": 0.9474,
+      "step": 16581
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4204089432447928e-06,
+      "loss": 0.8711,
+      "step": 16582
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420313395757692e-06,
+      "loss": 1.0758,
+      "step": 16583
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4202178482705904e-06,
+      "loss": 0.8626,
+      "step": 16584
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.42012230078349e-06,
+      "loss": 0.4755,
+      "step": 16585
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.420026753296389e-06,
+      "loss": 0.7581,
+      "step": 16586
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4199312058092875e-06,
+      "loss": 1.2494,
+      "step": 16587
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4198356583221866e-06,
+      "loss": 1.4265,
+      "step": 16588
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.419740110835085e-06,
+      "loss": 0.8366,
+      "step": 16589
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4196445633479842e-06,
+      "loss": 0.8938,
+      "step": 16590
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.419549015860883e-06,
+      "loss": 0.8615,
+      "step": 16591
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.419453468373782e-06,
+      "loss": 0.9322,
+      "step": 16592
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4193579208866814e-06,
+      "loss": 0.7799,
+      "step": 16593
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.41926237339958e-06,
+      "loss": 0.7309,
+      "step": 16594
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.419166825912479e-06,
+      "loss": 0.9611,
+      "step": 16595
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4190712784253776e-06,
+      "loss": 0.8847,
+      "step": 16596
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4189757309382767e-06,
+      "loss": 0.8119,
+      "step": 16597
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4188801834511753e-06,
+      "loss": 1.3543,
+      "step": 16598
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4187846359640743e-06,
+      "loss": 1.1733,
+      "step": 16599
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4186890884769734e-06,
+      "loss": 0.771,
+      "step": 16600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4185935409898724e-06,
+      "loss": 0.9528,
+      "step": 16601
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4184979935027714e-06,
+      "loss": 0.6954,
+      "step": 16602
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.41840244601567e-06,
+      "loss": 1.5511,
+      "step": 16603
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.418306898528569e-06,
+      "loss": 0.7116,
+      "step": 16604
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4182113510414677e-06,
+      "loss": 0.7989,
+      "step": 16605
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4181158035543667e-06,
+      "loss": 0.9167,
+      "step": 16606
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4180202560672658e-06,
+      "loss": 0.6395,
+      "step": 16607
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417924708580165e-06,
+      "loss": 1.124,
+      "step": 16608
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4178291610930634e-06,
+      "loss": 0.8971,
+      "step": 16609
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4177336136059625e-06,
+      "loss": 1.1721,
+      "step": 16610
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417638066118861e-06,
+      "loss": 0.8521,
+      "step": 16611
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.41754251863176e-06,
+      "loss": 0.86,
+      "step": 16612
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417446971144659e-06,
+      "loss": 0.9876,
+      "step": 16613
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417351423657558e-06,
+      "loss": 0.8228,
+      "step": 16614
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4172558761704572e-06,
+      "loss": 1.2753,
+      "step": 16615
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417160328683356e-06,
+      "loss": 0.8938,
+      "step": 16616
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417064781196255e-06,
+      "loss": 0.9479,
+      "step": 16617
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4169692337091535e-06,
+      "loss": 0.7023,
+      "step": 16618
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4168736862220525e-06,
+      "loss": 0.9756,
+      "step": 16619
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416778138734951e-06,
+      "loss": 0.7429,
+      "step": 16620
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4166825912478506e-06,
+      "loss": 0.8513,
+      "step": 16621
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4165870437607497e-06,
+      "loss": 0.5844,
+      "step": 16622
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4164914962736483e-06,
+      "loss": 1.0941,
+      "step": 16623
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4163959487865473e-06,
+      "loss": 0.7705,
+      "step": 16624
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416300401299446e-06,
+      "loss": 1.4279,
+      "step": 16625
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.416204853812345e-06,
+      "loss": 0.9105,
+      "step": 16626
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4161093063252436e-06,
+      "loss": 0.9686,
+      "step": 16627
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4160137588381426e-06,
+      "loss": 1.0487,
+      "step": 16628
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.415918211351042e-06,
+      "loss": 0.5534,
+      "step": 16629
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4158226638639407e-06,
+      "loss": 1.2194,
+      "step": 16630
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4157271163768397e-06,
+      "loss": 1.1507,
+      "step": 16631
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4156315688897384e-06,
+      "loss": 1.0901,
+      "step": 16632
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4155360214026374e-06,
+      "loss": 1.0435,
+      "step": 16633
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415440473915536e-06,
+      "loss": 0.9485,
+      "step": 16634
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415344926428435e-06,
+      "loss": 1.03,
+      "step": 16635
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4152493789413345e-06,
+      "loss": 1.0927,
+      "step": 16636
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415153831454233e-06,
+      "loss": 0.9583,
+      "step": 16637
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.415058283967132e-06,
+      "loss": 1.2405,
+      "step": 16638
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4149627364800308e-06,
+      "loss": 0.6477,
+      "step": 16639
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.41486718899293e-06,
+      "loss": 0.7102,
+      "step": 16640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4147716415058284e-06,
+      "loss": 0.9469,
+      "step": 16641
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4146760940187275e-06,
+      "loss": 0.9456,
+      "step": 16642
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.414580546531627e-06,
+      "loss": 1.5537,
+      "step": 16643
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4144849990445255e-06,
+      "loss": 0.7503,
+      "step": 16644
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4143894515574246e-06,
+      "loss": 0.5095,
+      "step": 16645
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.414293904070323e-06,
+      "loss": 0.5871,
+      "step": 16646
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4141983565832222e-06,
+      "loss": 0.7711,
+      "step": 16647
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.414102809096121e-06,
+      "loss": 0.9911,
+      "step": 16648
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.41400726160902e-06,
+      "loss": 0.9786,
+      "step": 16649
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.413911714121919e-06,
+      "loss": 0.6157,
+      "step": 16650
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.413816166634818e-06,
+      "loss": 0.8629,
+      "step": 16651
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.413720619147717e-06,
+      "loss": 1.1955,
+      "step": 16652
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4136250716606156e-06,
+      "loss": 0.6541,
+      "step": 16653
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4135295241735147e-06,
+      "loss": 0.7791,
+      "step": 16654
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4134339766864133e-06,
+      "loss": 0.723,
+      "step": 16655
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4133384291993123e-06,
+      "loss": 0.9655,
+      "step": 16656
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4132428817122114e-06,
+      "loss": 0.91,
+      "step": 16657
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4131473342251104e-06,
+      "loss": 0.6555,
+      "step": 16658
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.413051786738009e-06,
+      "loss": 0.8565,
+      "step": 16659
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412956239250908e-06,
+      "loss": 0.7048,
+      "step": 16660
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4128606917638067e-06,
+      "loss": 1.1514,
+      "step": 16661
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4127651442767057e-06,
+      "loss": 1.1431,
+      "step": 16662
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4126695967896043e-06,
+      "loss": 0.9834,
+      "step": 16663
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4125740493025034e-06,
+      "loss": 0.9712,
+      "step": 16664
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412478501815403e-06,
+      "loss": 0.707,
+      "step": 16665
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4123829543283014e-06,
+      "loss": 0.8436,
+      "step": 16666
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4122874068412005e-06,
+      "loss": 1.1668,
+      "step": 16667
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412191859354099e-06,
+      "loss": 0.6838,
+      "step": 16668
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412096311866998e-06,
+      "loss": 0.9803,
+      "step": 16669
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4120007643798967e-06,
+      "loss": 1.0256,
+      "step": 16670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4119052168927958e-06,
+      "loss": 0.5918,
+      "step": 16671
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4118096694056952e-06,
+      "loss": 1.0381,
+      "step": 16672
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411714121918594e-06,
+      "loss": 0.7719,
+      "step": 16673
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411618574431493e-06,
+      "loss": 1.1878,
+      "step": 16674
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4115230269443915e-06,
+      "loss": 0.9684,
+      "step": 16675
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4114274794572905e-06,
+      "loss": 1.1256,
+      "step": 16676
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411331931970189e-06,
+      "loss": 1.2574,
+      "step": 16677
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.411236384483088e-06,
+      "loss": 0.8025,
+      "step": 16678
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4111408369959877e-06,
+      "loss": 0.7242,
+      "step": 16679
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4110452895088863e-06,
+      "loss": 0.7839,
+      "step": 16680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4109497420217853e-06,
+      "loss": 0.8581,
+      "step": 16681
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.410854194534684e-06,
+      "loss": 1.3946,
+      "step": 16682
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.410758647047583e-06,
+      "loss": 1.0674,
+      "step": 16683
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4106630995604816e-06,
+      "loss": 1.1719,
+      "step": 16684
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4105675520733806e-06,
+      "loss": 0.8462,
+      "step": 16685
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.41047200458628e-06,
+      "loss": 0.9697,
+      "step": 16686
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4103764570991787e-06,
+      "loss": 1.2969,
+      "step": 16687
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4102809096120777e-06,
+      "loss": 0.8989,
+      "step": 16688
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4101853621249764e-06,
+      "loss": 0.6562,
+      "step": 16689
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4100898146378754e-06,
+      "loss": 1.0676,
+      "step": 16690
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.409994267150774e-06,
+      "loss": 0.909,
+      "step": 16691
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.409898719663673e-06,
+      "loss": 0.4475,
+      "step": 16692
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4098031721765725e-06,
+      "loss": 1.1088,
+      "step": 16693
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.409707624689471e-06,
+      "loss": 0.8807,
+      "step": 16694
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.40961207720237e-06,
+      "loss": 1.1637,
+      "step": 16695
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4095165297152688e-06,
+      "loss": 0.7181,
+      "step": 16696
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.409420982228168e-06,
+      "loss": 1.1289,
+      "step": 16697
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4093254347410664e-06,
+      "loss": 0.8768,
+      "step": 16698
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4092298872539655e-06,
+      "loss": 0.7801,
+      "step": 16699
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.409134339766864e-06,
+      "loss": 1.0002,
+      "step": 16700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4090387922797635e-06,
+      "loss": 1.005,
+      "step": 16701
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.408943244792662e-06,
+      "loss": 0.6886,
+      "step": 16702
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.408847697305561e-06,
+      "loss": 1.3442,
+      "step": 16703
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4087521498184602e-06,
+      "loss": 0.8338,
+      "step": 16704
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.408656602331359e-06,
+      "loss": 0.7282,
+      "step": 16705
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.408561054844258e-06,
+      "loss": 0.8434,
+      "step": 16706
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4084655073571565e-06,
+      "loss": 1.171,
+      "step": 16707
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.408369959870056e-06,
+      "loss": 0.9458,
+      "step": 16708
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4082744123829546e-06,
+      "loss": 0.9679,
+      "step": 16709
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4081788648958536e-06,
+      "loss": 0.7216,
+      "step": 16710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4080833174087522e-06,
+      "loss": 0.7183,
+      "step": 16711
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4079877699216513e-06,
+      "loss": 0.8703,
+      "step": 16712
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.40789222243455e-06,
+      "loss": 0.7883,
+      "step": 16713
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407796674947449e-06,
+      "loss": 0.9003,
+      "step": 16714
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4077011274603484e-06,
+      "loss": 1.0616,
+      "step": 16715
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407605579973247e-06,
+      "loss": 0.9047,
+      "step": 16716
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407510032486146e-06,
+      "loss": 1.1703,
+      "step": 16717
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4074144849990447e-06,
+      "loss": 0.7642,
+      "step": 16718
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4073189375119437e-06,
+      "loss": 0.6795,
+      "step": 16719
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4072233900248423e-06,
+      "loss": 0.9654,
+      "step": 16720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4071278425377414e-06,
+      "loss": 0.9346,
+      "step": 16721
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407032295050641e-06,
+      "loss": 1.2701,
+      "step": 16722
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4069367475635394e-06,
+      "loss": 1.0649,
+      "step": 16723
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4068412000764385e-06,
+      "loss": 0.9698,
+      "step": 16724
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406745652589337e-06,
+      "loss": 0.7802,
+      "step": 16725
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406650105102236e-06,
+      "loss": 0.8594,
+      "step": 16726
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4065545576151347e-06,
+      "loss": 0.7793,
+      "step": 16727
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4064590101280338e-06,
+      "loss": 1.3276,
+      "step": 16728
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4063634626409332e-06,
+      "loss": 0.7481,
+      "step": 16729
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406267915153832e-06,
+      "loss": 0.7503,
+      "step": 16730
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406172367666731e-06,
+      "loss": 0.693,
+      "step": 16731
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4060768201796295e-06,
+      "loss": 1.003,
+      "step": 16732
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4059812726925285e-06,
+      "loss": 1.1098,
+      "step": 16733
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405885725205427e-06,
+      "loss": 0.8733,
+      "step": 16734
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405790177718326e-06,
+      "loss": 1.4172,
+      "step": 16735
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.405694630231225e-06,
+      "loss": 1.2916,
+      "step": 16736
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4055990827441243e-06,
+      "loss": 0.6298,
+      "step": 16737
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4055035352570233e-06,
+      "loss": 1.3013,
+      "step": 16738
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.405407987769922e-06,
+      "loss": 0.7451,
+      "step": 16739
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.405312440282821e-06,
+      "loss": 1.0904,
+      "step": 16740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4052168927957196e-06,
+      "loss": 0.9218,
+      "step": 16741
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4051213453086186e-06,
+      "loss": 0.9316,
+      "step": 16742
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4050257978215172e-06,
+      "loss": 0.6652,
+      "step": 16743
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4049302503344167e-06,
+      "loss": 0.934,
+      "step": 16744
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4048347028473157e-06,
+      "loss": 1.05,
+      "step": 16745
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4047391553602144e-06,
+      "loss": 1.1289,
+      "step": 16746
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4046436078731134e-06,
+      "loss": 1.1787,
+      "step": 16747
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.404548060386012e-06,
+      "loss": 0.8248,
+      "step": 16748
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.404452512898911e-06,
+      "loss": 0.3945,
+      "step": 16749
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4043569654118097e-06,
+      "loss": 0.8228,
+      "step": 16750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.404261417924709e-06,
+      "loss": 1.4293,
+      "step": 16751
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4041658704376077e-06,
+      "loss": 0.8993,
+      "step": 16752
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4040703229505068e-06,
+      "loss": 1.021,
+      "step": 16753
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4039747754634054e-06,
+      "loss": 0.8225,
+      "step": 16754
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4038792279763044e-06,
+      "loss": 0.707,
+      "step": 16755
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4037836804892035e-06,
+      "loss": 1.0316,
+      "step": 16756
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.403688133002102e-06,
+      "loss": 1.1724,
+      "step": 16757
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4035925855150015e-06,
+      "loss": 0.6784,
+      "step": 16758
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4034970380279e-06,
+      "loss": 0.7313,
+      "step": 16759
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.403401490540799e-06,
+      "loss": 0.7514,
+      "step": 16760
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.403305943053698e-06,
+      "loss": 0.7386,
+      "step": 16761
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.403210395566597e-06,
+      "loss": 0.832,
+      "step": 16762
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4031148480794955e-06,
+      "loss": 0.9172,
+      "step": 16763
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4030193005923945e-06,
+      "loss": 0.9287,
+      "step": 16764
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402923753105294e-06,
+      "loss": 0.6992,
+      "step": 16765
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4028282056181926e-06,
+      "loss": 0.5162,
+      "step": 16766
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4027326581310916e-06,
+      "loss": 0.7129,
+      "step": 16767
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4026371106439902e-06,
+      "loss": 0.9082,
+      "step": 16768
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4025415631568893e-06,
+      "loss": 0.9852,
+      "step": 16769
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402446015669788e-06,
+      "loss": 1.4425,
+      "step": 16770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402350468182687e-06,
+      "loss": 0.7106,
+      "step": 16771
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4022549206955855e-06,
+      "loss": 1.0574,
+      "step": 16772
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402159373208485e-06,
+      "loss": 0.924,
+      "step": 16773
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402063825721384e-06,
+      "loss": 0.8651,
+      "step": 16774
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4019682782342827e-06,
+      "loss": 1.0499,
+      "step": 16775
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4018727307471817e-06,
+      "loss": 0.6164,
+      "step": 16776
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4017771832600803e-06,
+      "loss": 1.2509,
+      "step": 16777
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4016816357729794e-06,
+      "loss": 0.7142,
+      "step": 16778
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401586088285878e-06,
+      "loss": 1.0094,
+      "step": 16779
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4014905407987774e-06,
+      "loss": 1.1344,
+      "step": 16780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4013949933116765e-06,
+      "loss": 1.3908,
+      "step": 16781
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401299445824575e-06,
+      "loss": 0.8819,
+      "step": 16782
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401203898337474e-06,
+      "loss": 0.6407,
+      "step": 16783
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4011083508503727e-06,
+      "loss": 1.2635,
+      "step": 16784
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4010128033632718e-06,
+      "loss": 0.8685,
+      "step": 16785
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4009172558761704e-06,
+      "loss": 0.8127,
+      "step": 16786
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.40082170838907e-06,
+      "loss": 0.9051,
+      "step": 16787
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400726160901969e-06,
+      "loss": 0.5524,
+      "step": 16788
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4006306134148675e-06,
+      "loss": 1.4253,
+      "step": 16789
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4005350659277665e-06,
+      "loss": 0.979,
+      "step": 16790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400439518440665e-06,
+      "loss": 0.5913,
+      "step": 16791
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400343970953564e-06,
+      "loss": 0.9496,
+      "step": 16792
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.400248423466463e-06,
+      "loss": 0.6786,
+      "step": 16793
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4001528759793623e-06,
+      "loss": 1.1654,
+      "step": 16794
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.4000573284922613e-06,
+      "loss": 0.827,
+      "step": 16795
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.39996178100516e-06,
+      "loss": 1.0748,
+      "step": 16796
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.399866233518059e-06,
+      "loss": 1.3675,
+      "step": 16797
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3997706860309576e-06,
+      "loss": 0.8326,
+      "step": 16798
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3996751385438566e-06,
+      "loss": 0.9959,
+      "step": 16799
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3995795910567552e-06,
+      "loss": 0.9528,
+      "step": 16800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3994840435696547e-06,
+      "loss": 1.2772,
+      "step": 16801
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3993884960825533e-06,
+      "loss": 0.8755,
+      "step": 16802
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3992929485954524e-06,
+      "loss": 1.1051,
+      "step": 16803
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.399197401108351e-06,
+      "loss": 0.8011,
+      "step": 16804
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.39910185362125e-06,
+      "loss": 0.763,
+      "step": 16805
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.399006306134149e-06,
+      "loss": 0.9333,
+      "step": 16806
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3989107586470477e-06,
+      "loss": 0.6616,
+      "step": 16807
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3988152111599467e-06,
+      "loss": 1.1391,
+      "step": 16808
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3987196636728457e-06,
+      "loss": 0.8201,
+      "step": 16809
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3986241161857448e-06,
+      "loss": 1.3211,
+      "step": 16810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3985285686986434e-06,
+      "loss": 0.7571,
+      "step": 16811
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3984330212115424e-06,
+      "loss": 0.7212,
+      "step": 16812
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.398337473724441e-06,
+      "loss": 0.803,
+      "step": 16813
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.39824192623734e-06,
+      "loss": 0.5199,
+      "step": 16814
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3981463787502387e-06,
+      "loss": 1.2841,
+      "step": 16815
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.398050831263138e-06,
+      "loss": 0.685,
+      "step": 16816
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.397955283776037e-06,
+      "loss": 1.2198,
+      "step": 16817
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.397859736288936e-06,
+      "loss": 0.892,
+      "step": 16818
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.397764188801835e-06,
+      "loss": 1.1409,
+      "step": 16819
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3976686413147335e-06,
+      "loss": 0.6816,
+      "step": 16820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3975730938276325e-06,
+      "loss": 0.6876,
+      "step": 16821
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.397477546340531e-06,
+      "loss": 0.8283,
+      "step": 16822
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3973819988534306e-06,
+      "loss": 1.0185,
+      "step": 16823
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3972864513663296e-06,
+      "loss": 1.1327,
+      "step": 16824
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3971909038792282e-06,
+      "loss": 1.03,
+      "step": 16825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3970953563921273e-06,
+      "loss": 0.9806,
+      "step": 16826
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396999808905026e-06,
+      "loss": 0.6674,
+      "step": 16827
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396904261417925e-06,
+      "loss": 0.9882,
+      "step": 16828
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3968087139308235e-06,
+      "loss": 0.6152,
+      "step": 16829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396713166443723e-06,
+      "loss": 0.7588,
+      "step": 16830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396617618956622e-06,
+      "loss": 0.6965,
+      "step": 16831
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3965220714695207e-06,
+      "loss": 0.7644,
+      "step": 16832
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3964265239824197e-06,
+      "loss": 1.0713,
+      "step": 16833
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3963309764953183e-06,
+      "loss": 0.8804,
+      "step": 16834
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3962354290082173e-06,
+      "loss": 0.7628,
+      "step": 16835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396139881521116e-06,
+      "loss": 0.7594,
+      "step": 16836
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3960443340340154e-06,
+      "loss": 0.4252,
+      "step": 16837
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3959487865469145e-06,
+      "loss": 1.2847,
+      "step": 16838
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395853239059813e-06,
+      "loss": 0.81,
+      "step": 16839
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395757691572712e-06,
+      "loss": 1.0717,
+      "step": 16840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3956621440856107e-06,
+      "loss": 0.8134,
+      "step": 16841
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3955665965985098e-06,
+      "loss": 0.6459,
+      "step": 16842
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3954710491114084e-06,
+      "loss": 0.6983,
+      "step": 16843
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3953755016243074e-06,
+      "loss": 0.918,
+      "step": 16844
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.395279954137207e-06,
+      "loss": 0.6997,
+      "step": 16845
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3951844066501055e-06,
+      "loss": 1.1238,
+      "step": 16846
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3950888591630045e-06,
+      "loss": 1.1123,
+      "step": 16847
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394993311675903e-06,
+      "loss": 1.1044,
+      "step": 16848
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394897764188802e-06,
+      "loss": 0.7836,
+      "step": 16849
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394802216701701e-06,
+      "loss": 1.0934,
+      "step": 16850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3947066692146e-06,
+      "loss": 0.6697,
+      "step": 16851
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394611121727499e-06,
+      "loss": 1.1514,
+      "step": 16852
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394515574240398e-06,
+      "loss": 0.5456,
+      "step": 16853
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3944200267532965e-06,
+      "loss": 0.5708,
+      "step": 16854
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3943244792661956e-06,
+      "loss": 0.5946,
+      "step": 16855
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.394228931779094e-06,
+      "loss": 0.8386,
+      "step": 16856
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3941333842919932e-06,
+      "loss": 1.1742,
+      "step": 16857
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3940378368048923e-06,
+      "loss": 0.8742,
+      "step": 16858
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3939422893177913e-06,
+      "loss": 0.5457,
+      "step": 16859
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3938467418306903e-06,
+      "loss": 0.9203,
+      "step": 16860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.393751194343589e-06,
+      "loss": 0.5638,
+      "step": 16861
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.393655646856488e-06,
+      "loss": 0.9876,
+      "step": 16862
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3935600993693866e-06,
+      "loss": 1.332,
+      "step": 16863
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3934645518822857e-06,
+      "loss": 1.3812,
+      "step": 16864
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3933690043951843e-06,
+      "loss": 0.8411,
+      "step": 16865
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3932734569080837e-06,
+      "loss": 0.9063,
+      "step": 16866
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3931779094209828e-06,
+      "loss": 0.8482,
+      "step": 16867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3930823619338814e-06,
+      "loss": 0.8623,
+      "step": 16868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3929868144467804e-06,
+      "loss": 1.0649,
+      "step": 16869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392891266959679e-06,
+      "loss": 1.033,
+      "step": 16870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392795719472578e-06,
+      "loss": 0.78,
+      "step": 16871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3927001719854767e-06,
+      "loss": 0.6381,
+      "step": 16872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392604624498376e-06,
+      "loss": 0.9842,
+      "step": 16873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392509077011275e-06,
+      "loss": 1.5316,
+      "step": 16874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392413529524174e-06,
+      "loss": 1.2454,
+      "step": 16875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392317982037073e-06,
+      "loss": 0.6585,
+      "step": 16876
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3922224345499715e-06,
+      "loss": 0.6348,
+      "step": 16877
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3921268870628705e-06,
+      "loss": 1.7473,
+      "step": 16878
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.392031339575769e-06,
+      "loss": 1.2228,
+      "step": 16879
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391935792088668e-06,
+      "loss": 0.9107,
+      "step": 16880
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3918402446015676e-06,
+      "loss": 1.281,
+      "step": 16881
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3917446971144662e-06,
+      "loss": 0.9574,
+      "step": 16882
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3916491496273653e-06,
+      "loss": 0.7419,
+      "step": 16883
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391553602140264e-06,
+      "loss": 0.8571,
+      "step": 16884
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391458054653163e-06,
+      "loss": 0.7849,
+      "step": 16885
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3913625071660615e-06,
+      "loss": 1.0135,
+      "step": 16886
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3912669596789606e-06,
+      "loss": 0.916,
+      "step": 16887
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.39117141219186e-06,
+      "loss": 1.0829,
+      "step": 16888
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3910758647047587e-06,
+      "loss": 0.7974,
+      "step": 16889
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3909803172176577e-06,
+      "loss": 1.0565,
+      "step": 16890
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3908847697305563e-06,
+      "loss": 1.2556,
+      "step": 16891
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3907892222434553e-06,
+      "loss": 0.5645,
+      "step": 16892
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390693674756354e-06,
+      "loss": 0.6644,
+      "step": 16893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390598127269253e-06,
+      "loss": 0.5698,
+      "step": 16894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390502579782152e-06,
+      "loss": 0.9023,
+      "step": 16895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.390407032295051e-06,
+      "loss": 1.1854,
+      "step": 16896
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.39031148480795e-06,
+      "loss": 0.3967,
+      "step": 16897
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3902159373208487e-06,
+      "loss": 0.7853,
+      "step": 16898
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3901203898337478e-06,
+      "loss": 1.2085,
+      "step": 16899
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3900248423466464e-06,
+      "loss": 1.2333,
+      "step": 16900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3899292948595454e-06,
+      "loss": 0.9536,
+      "step": 16901
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3898337473724445e-06,
+      "loss": 1.1143,
+      "step": 16902
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3897381998853435e-06,
+      "loss": 1.104,
+      "step": 16903
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389642652398242e-06,
+      "loss": 0.5859,
+      "step": 16904
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389547104911141e-06,
+      "loss": 0.7122,
+      "step": 16905
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3894515574240398e-06,
+      "loss": 0.768,
+      "step": 16906
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389356009936939e-06,
+      "loss": 0.5921,
+      "step": 16907
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389260462449838e-06,
+      "loss": 0.4893,
+      "step": 16908
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389164914962737e-06,
+      "loss": 0.7088,
+      "step": 16909
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.389069367475636e-06,
+      "loss": 0.8205,
+      "step": 16910
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3889738199885345e-06,
+      "loss": 0.7869,
+      "step": 16911
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3888782725014336e-06,
+      "loss": 0.7456,
+      "step": 16912
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.388782725014332e-06,
+      "loss": 0.5771,
+      "step": 16913
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3886871775272312e-06,
+      "loss": 0.9751,
+      "step": 16914
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.38859163004013e-06,
+      "loss": 0.9189,
+      "step": 16915
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.388496082553029e-06,
+      "loss": 1.1585,
+      "step": 16916
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3884005350659283e-06,
+      "loss": 0.7749,
+      "step": 16917
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.388304987578827e-06,
+      "loss": 0.8932,
+      "step": 16918
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.388209440091726e-06,
+      "loss": 0.6696,
+      "step": 16919
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3881138926046246e-06,
+      "loss": 0.6331,
+      "step": 16920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3880183451175237e-06,
+      "loss": 0.9906,
+      "step": 16921
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3879227976304223e-06,
+      "loss": 1.1189,
+      "step": 16922
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3878272501433213e-06,
+      "loss": 1.0151,
+      "step": 16923
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3877317026562208e-06,
+      "loss": 0.7397,
+      "step": 16924
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3876361551691194e-06,
+      "loss": 1.1762,
+      "step": 16925
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3875406076820184e-06,
+      "loss": 0.5732,
+      "step": 16926
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.387445060194917e-06,
+      "loss": 0.4166,
+      "step": 16927
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.387349512707816e-06,
+      "loss": 0.6742,
+      "step": 16928
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3872539652207147e-06,
+      "loss": 0.8127,
+      "step": 16929
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3871584177336137e-06,
+      "loss": 1.1231,
+      "step": 16930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.387062870246513e-06,
+      "loss": 0.9457,
+      "step": 16931
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386967322759412e-06,
+      "loss": 0.7336,
+      "step": 16932
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386871775272311e-06,
+      "loss": 0.932,
+      "step": 16933
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3867762277852095e-06,
+      "loss": 0.7001,
+      "step": 16934
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3866806802981085e-06,
+      "loss": 0.7477,
+      "step": 16935
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386585132811007e-06,
+      "loss": 0.6636,
+      "step": 16936
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386489585323906e-06,
+      "loss": 0.4657,
+      "step": 16937
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3863940378368056e-06,
+      "loss": 1.3795,
+      "step": 16938
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3862984903497042e-06,
+      "loss": 1.0017,
+      "step": 16939
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3862029428626033e-06,
+      "loss": 1.0821,
+      "step": 16940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386107395375502e-06,
+      "loss": 0.8694,
+      "step": 16941
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.386011847888401e-06,
+      "loss": 0.6892,
+      "step": 16942
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3859163004012995e-06,
+      "loss": 0.9664,
+      "step": 16943
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3858207529141986e-06,
+      "loss": 0.9801,
+      "step": 16944
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3857252054270976e-06,
+      "loss": 0.6046,
+      "step": 16945
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3856296579399967e-06,
+      "loss": 1.0031,
+      "step": 16946
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3855341104528957e-06,
+      "loss": 0.5313,
+      "step": 16947
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3854385629657943e-06,
+      "loss": 1.1209,
+      "step": 16948
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3853430154786933e-06,
+      "loss": 0.6565,
+      "step": 16949
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385247467991592e-06,
+      "loss": 1.1196,
+      "step": 16950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.385151920504491e-06,
+      "loss": 0.7195,
+      "step": 16951
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3850563730173896e-06,
+      "loss": 1.0812,
+      "step": 16952
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.384960825530289e-06,
+      "loss": 0.7205,
+      "step": 16953
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3848652780431877e-06,
+      "loss": 0.7437,
+      "step": 16954
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3847697305560867e-06,
+      "loss": 1.1141,
+      "step": 16955
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3846741830689853e-06,
+      "loss": 1.1536,
+      "step": 16956
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3845786355818844e-06,
+      "loss": 1.0972,
+      "step": 16957
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.384483088094783e-06,
+      "loss": 1.3253,
+      "step": 16958
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.384387540607682e-06,
+      "loss": 0.7088,
+      "step": 16959
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3842919931205815e-06,
+      "loss": 1.5869,
+      "step": 16960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38419644563348e-06,
+      "loss": 0.845,
+      "step": 16961
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.384100898146379e-06,
+      "loss": 1.2249,
+      "step": 16962
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3840053506592778e-06,
+      "loss": 0.8119,
+      "step": 16963
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.383909803172177e-06,
+      "loss": 0.6648,
+      "step": 16964
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3838142556850754e-06,
+      "loss": 0.6921,
+      "step": 16965
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3837187081979745e-06,
+      "loss": 1.4504,
+      "step": 16966
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.383623160710874e-06,
+      "loss": 0.958,
+      "step": 16967
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3835276132237725e-06,
+      "loss": 0.7635,
+      "step": 16968
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3834320657366716e-06,
+      "loss": 0.7307,
+      "step": 16969
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38333651824957e-06,
+      "loss": 0.8862,
+      "step": 16970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3832409707624692e-06,
+      "loss": 0.8159,
+      "step": 16971
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.383145423275368e-06,
+      "loss": 0.8236,
+      "step": 16972
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.383049875788267e-06,
+      "loss": 0.837,
+      "step": 16973
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3829543283011663e-06,
+      "loss": 1.0576,
+      "step": 16974
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.382858780814065e-06,
+      "loss": 0.8644,
+      "step": 16975
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.382763233326964e-06,
+      "loss": 0.9276,
+      "step": 16976
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3826676858398626e-06,
+      "loss": 1.2265,
+      "step": 16977
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3825721383527617e-06,
+      "loss": 0.8451,
+      "step": 16978
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3824765908656603e-06,
+      "loss": 1.2826,
+      "step": 16979
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3823810433785593e-06,
+      "loss": 0.746,
+      "step": 16980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3822854958914588e-06,
+      "loss": 1.0615,
+      "step": 16981
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3821899484043574e-06,
+      "loss": 0.6716,
+      "step": 16982
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3820944009172564e-06,
+      "loss": 1.0369,
+      "step": 16983
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381998853430155e-06,
+      "loss": 1.0981,
+      "step": 16984
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381903305943054e-06,
+      "loss": 0.8306,
+      "step": 16985
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3818077584559527e-06,
+      "loss": 0.4242,
+      "step": 16986
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3817122109688517e-06,
+      "loss": 0.9205,
+      "step": 16987
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3816166634817503e-06,
+      "loss": 0.8554,
+      "step": 16988
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.38152111599465e-06,
+      "loss": 1.1743,
+      "step": 16989
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381425568507549e-06,
+      "loss": 0.8283,
+      "step": 16990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3813300210204475e-06,
+      "loss": 0.466,
+      "step": 16991
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3812344735333465e-06,
+      "loss": 0.7445,
+      "step": 16992
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381138926046245e-06,
+      "loss": 0.7908,
+      "step": 16993
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381043378559144e-06,
+      "loss": 0.5814,
+      "step": 16994
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3809478310720428e-06,
+      "loss": 1.1098,
+      "step": 16995
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3808522835849422e-06,
+      "loss": 1.0956,
+      "step": 16996
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.380756736097841e-06,
+      "loss": 0.5962,
+      "step": 16997
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.38066118861074e-06,
+      "loss": 0.8391,
+      "step": 16998
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380565641123639e-06,
+      "loss": 1.0164,
+      "step": 16999
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3804700936365375e-06,
+      "loss": 0.9408,
+      "step": 17000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3803745461494366e-06,
+      "loss": 1.3863,
+      "step": 17001
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.380278998662335e-06,
+      "loss": 0.9169,
+      "step": 17002
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3801834511752347e-06,
+      "loss": 1.1992,
+      "step": 17003
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3800879036881333e-06,
+      "loss": 0.8454,
+      "step": 17004
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3799923562010323e-06,
+      "loss": 0.7043,
+      "step": 17005
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.379896808713931e-06,
+      "loss": 0.8129,
+      "step": 17006
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.37980126122683e-06,
+      "loss": 0.7034,
+      "step": 17007
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3797057137397286e-06,
+      "loss": 0.6828,
+      "step": 17008
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3796101662526276e-06,
+      "loss": 0.8905,
+      "step": 17009
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.379514618765527e-06,
+      "loss": 1.0977,
+      "step": 17010
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3794190712784257e-06,
+      "loss": 0.4116,
+      "step": 17011
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3793235237913247e-06,
+      "loss": 0.9639,
+      "step": 17012
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3792279763042233e-06,
+      "loss": 1.1633,
+      "step": 17013
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3791324288171224e-06,
+      "loss": 0.8726,
+      "step": 17014
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.379036881330021e-06,
+      "loss": 0.7365,
+      "step": 17015
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.37894133384292e-06,
+      "loss": 0.9702,
+      "step": 17016
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3788457863558195e-06,
+      "loss": 0.4827,
+      "step": 17017
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.378750238868718e-06,
+      "loss": 1.1483,
+      "step": 17018
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.378654691381617e-06,
+      "loss": 0.3871,
+      "step": 17019
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3785591438945158e-06,
+      "loss": 0.8293,
+      "step": 17020
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.378463596407415e-06,
+      "loss": 0.92,
+      "step": 17021
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3783680489203134e-06,
+      "loss": 0.7141,
+      "step": 17022
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3782725014332125e-06,
+      "loss": 1.2981,
+      "step": 17023
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.378176953946111e-06,
+      "loss": 0.6591,
+      "step": 17024
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3780814064590105e-06,
+      "loss": 0.7957,
+      "step": 17025
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3779858589719096e-06,
+      "loss": 0.605,
+      "step": 17026
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377890311484808e-06,
+      "loss": 0.8651,
+      "step": 17027
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3777947639977072e-06,
+      "loss": 0.4775,
+      "step": 17028
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377699216510606e-06,
+      "loss": 0.6478,
+      "step": 17029
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377603669023505e-06,
+      "loss": 0.8473,
+      "step": 17030
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3775081215364035e-06,
+      "loss": 0.734,
+      "step": 17031
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377412574049303e-06,
+      "loss": 0.7043,
+      "step": 17032
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.377317026562202e-06,
+      "loss": 0.9661,
+      "step": 17033
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3772214790751006e-06,
+      "loss": 0.91,
+      "step": 17034
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3771259315879997e-06,
+      "loss": 0.7843,
+      "step": 17035
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3770303841008983e-06,
+      "loss": 0.8638,
+      "step": 17036
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3769348366137973e-06,
+      "loss": 0.928,
+      "step": 17037
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376839289126696e-06,
+      "loss": 0.9343,
+      "step": 17038
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3767437416395954e-06,
+      "loss": 1.3597,
+      "step": 17039
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3766481941524944e-06,
+      "loss": 1.1777,
+      "step": 17040
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376552646665393e-06,
+      "loss": 0.7753,
+      "step": 17041
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376457099178292e-06,
+      "loss": 0.499,
+      "step": 17042
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3763615516911907e-06,
+      "loss": 0.7526,
+      "step": 17043
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3762660042040897e-06,
+      "loss": 0.7452,
+      "step": 17044
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3761704567169883e-06,
+      "loss": 0.7545,
+      "step": 17045
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376074909229888e-06,
+      "loss": 0.5629,
+      "step": 17046
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3759793617427864e-06,
+      "loss": 0.9916,
+      "step": 17047
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3758838142556855e-06,
+      "loss": 0.8241,
+      "step": 17048
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3757882667685845e-06,
+      "loss": 0.6306,
+      "step": 17049
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375692719281483e-06,
+      "loss": 1.0495,
+      "step": 17050
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375597171794382e-06,
+      "loss": 0.7243,
+      "step": 17051
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3755016243072808e-06,
+      "loss": 1.0875,
+      "step": 17052
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3754060768201802e-06,
+      "loss": 0.8123,
+      "step": 17053
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375310529333079e-06,
+      "loss": 0.9134,
+      "step": 17054
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.375214981845978e-06,
+      "loss": 0.9268,
+      "step": 17055
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3751194343588765e-06,
+      "loss": 0.973,
+      "step": 17056
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3750238868717755e-06,
+      "loss": 1.335,
+      "step": 17057
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.374928339384674e-06,
+      "loss": 0.5447,
+      "step": 17058
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.374832791897573e-06,
+      "loss": 0.8849,
+      "step": 17059
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.374737244410472e-06,
+      "loss": 0.7175,
+      "step": 17060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3746416969233713e-06,
+      "loss": 0.826,
+      "step": 17061
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3745461494362703e-06,
+      "loss": 0.8102,
+      "step": 17062
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.374450601949169e-06,
+      "loss": 0.7355,
+      "step": 17063
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.374355054462068e-06,
+      "loss": 0.9567,
+      "step": 17064
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3742595069749666e-06,
+      "loss": 1.0311,
+      "step": 17065
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3741639594878656e-06,
+      "loss": 1.1966,
+      "step": 17066
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3740684120007642e-06,
+      "loss": 0.9588,
+      "step": 17067
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3739728645136637e-06,
+      "loss": 1.1896,
+      "step": 17068
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3738773170265627e-06,
+      "loss": 1.2102,
+      "step": 17069
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3737817695394613e-06,
+      "loss": 1.0043,
+      "step": 17070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3736862220523604e-06,
+      "loss": 1.1768,
+      "step": 17071
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373590674565259e-06,
+      "loss": 0.748,
+      "step": 17072
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373495127078158e-06,
+      "loss": 1.0981,
+      "step": 17073
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3733995795910567e-06,
+      "loss": 1.4069,
+      "step": 17074
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373304032103956e-06,
+      "loss": 0.8669,
+      "step": 17075
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373208484616855e-06,
+      "loss": 1.1826,
+      "step": 17076
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3731129371297538e-06,
+      "loss": 1.0261,
+      "step": 17077
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373017389642653e-06,
+      "loss": 1.3099,
+      "step": 17078
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3729218421555514e-06,
+      "loss": 1.0458,
+      "step": 17079
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3728262946684505e-06,
+      "loss": 0.5043,
+      "step": 17080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.372730747181349e-06,
+      "loss": 0.8093,
+      "step": 17081
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3726351996942485e-06,
+      "loss": 0.438,
+      "step": 17082
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3725396522071476e-06,
+      "loss": 0.9174,
+      "step": 17083
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.372444104720046e-06,
+      "loss": 0.9177,
+      "step": 17084
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3723485572329452e-06,
+      "loss": 0.6243,
+      "step": 17085
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.372253009745844e-06,
+      "loss": 1.0503,
+      "step": 17086
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.372157462258743e-06,
+      "loss": 1.7552,
+      "step": 17087
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3720619147716415e-06,
+      "loss": 0.8878,
+      "step": 17088
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371966367284541e-06,
+      "loss": 0.8308,
+      "step": 17089
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37187081979744e-06,
+      "loss": 0.7896,
+      "step": 17090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3717752723103386e-06,
+      "loss": 0.9189,
+      "step": 17091
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3716797248232377e-06,
+      "loss": 0.9882,
+      "step": 17092
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3715841773361363e-06,
+      "loss": 1.1696,
+      "step": 17093
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3714886298490353e-06,
+      "loss": 0.6083,
+      "step": 17094
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371393082361934e-06,
+      "loss": 1.3293,
+      "step": 17095
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371297534874833e-06,
+      "loss": 0.7862,
+      "step": 17096
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371201987387732e-06,
+      "loss": 0.8229,
+      "step": 17097
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371106439900631e-06,
+      "loss": 0.9395,
+      "step": 17098
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3710108924135297e-06,
+      "loss": 1.0863,
+      "step": 17099
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3709153449264287e-06,
+      "loss": 0.8858,
+      "step": 17100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3708197974393277e-06,
+      "loss": 0.6985,
+      "step": 17101
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3707242499522263e-06,
+      "loss": 1.1341,
+      "step": 17102
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3706287024651254e-06,
+      "loss": 0.9561,
+      "step": 17103
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3705331549780244e-06,
+      "loss": 0.8811,
+      "step": 17104
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3704376074909235e-06,
+      "loss": 0.6402,
+      "step": 17105
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370342060003822e-06,
+      "loss": 0.7045,
+      "step": 17106
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.370246512516721e-06,
+      "loss": 1.4971,
+      "step": 17107
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3701509650296197e-06,
+      "loss": 1.1464,
+      "step": 17108
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3700554175425188e-06,
+      "loss": 1.039,
+      "step": 17109
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3699598700554174e-06,
+      "loss": 1.038,
+      "step": 17110
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.369864322568317e-06,
+      "loss": 0.9753,
+      "step": 17111
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.369768775081216e-06,
+      "loss": 0.5862,
+      "step": 17112
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3696732275941145e-06,
+      "loss": 1.0554,
+      "step": 17113
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3695776801070135e-06,
+      "loss": 1.0642,
+      "step": 17114
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.369482132619912e-06,
+      "loss": 1.3864,
+      "step": 17115
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.369386585132811e-06,
+      "loss": 0.9916,
+      "step": 17116
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.36929103764571e-06,
+      "loss": 0.6678,
+      "step": 17117
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3691954901586093e-06,
+      "loss": 0.9911,
+      "step": 17118
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3690999426715083e-06,
+      "loss": 0.9845,
+      "step": 17119
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.369004395184407e-06,
+      "loss": 0.6878,
+      "step": 17120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.368908847697306e-06,
+      "loss": 0.7438,
+      "step": 17121
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3688133002102046e-06,
+      "loss": 0.9338,
+      "step": 17122
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3687177527231036e-06,
+      "loss": 1.1208,
+      "step": 17123
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3686222052360022e-06,
+      "loss": 0.7833,
+      "step": 17124
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3685266577489017e-06,
+      "loss": 1.1092,
+      "step": 17125
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3684311102618007e-06,
+      "loss": 0.9158,
+      "step": 17126
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3683355627746993e-06,
+      "loss": 0.7047,
+      "step": 17127
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3682400152875984e-06,
+      "loss": 0.9495,
+      "step": 17128
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.368144467800497e-06,
+      "loss": 1.6316,
+      "step": 17129
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.368048920313396e-06,
+      "loss": 1.4151,
+      "step": 17130
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3679533728262947e-06,
+      "loss": 0.7547,
+      "step": 17131
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3678578253391937e-06,
+      "loss": 1.1029,
+      "step": 17132
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367762277852093e-06,
+      "loss": 0.786,
+      "step": 17133
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3676667303649918e-06,
+      "loss": 1.4226,
+      "step": 17134
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367571182877891e-06,
+      "loss": 0.9431,
+      "step": 17135
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3674756353907894e-06,
+      "loss": 0.982,
+      "step": 17136
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3673800879036885e-06,
+      "loss": 1.3156,
+      "step": 17137
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367284540416587e-06,
+      "loss": 0.848,
+      "step": 17138
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367188992929486e-06,
+      "loss": 0.6232,
+      "step": 17139
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3670934454423856e-06,
+      "loss": 0.7631,
+      "step": 17140
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366997897955284e-06,
+      "loss": 1.172,
+      "step": 17141
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3669023504681832e-06,
+      "loss": 0.8025,
+      "step": 17142
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366806802981082e-06,
+      "loss": 1.2614,
+      "step": 17143
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366711255493981e-06,
+      "loss": 1.0732,
+      "step": 17144
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3666157080068795e-06,
+      "loss": 0.9041,
+      "step": 17145
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3665201605197785e-06,
+      "loss": 0.6713,
+      "step": 17146
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3664246130326776e-06,
+      "loss": 1.4112,
+      "step": 17147
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3663290655455766e-06,
+      "loss": 0.599,
+      "step": 17148
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3662335180584752e-06,
+      "loss": 1.2202,
+      "step": 17149
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3661379705713743e-06,
+      "loss": 0.7646,
+      "step": 17150
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.366042423084273e-06,
+      "loss": 0.5382,
+      "step": 17151
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.365946875597172e-06,
+      "loss": 0.6397,
+      "step": 17152
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.365851328110071e-06,
+      "loss": 0.6319,
+      "step": 17153
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.36575578062297e-06,
+      "loss": 1.0192,
+      "step": 17154
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365660233135869e-06,
+      "loss": 1.2297,
+      "step": 17155
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3655646856487677e-06,
+      "loss": 0.8109,
+      "step": 17156
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3654691381616667e-06,
+      "loss": 1.1257,
+      "step": 17157
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3653735906745653e-06,
+      "loss": 1.4588,
+      "step": 17158
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3652780431874643e-06,
+      "loss": 1.0605,
+      "step": 17159
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.365182495700363e-06,
+      "loss": 0.6799,
+      "step": 17160
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3650869482132624e-06,
+      "loss": 1.1172,
+      "step": 17161
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3649914007261615e-06,
+      "loss": 0.8454,
+      "step": 17162
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36489585323906e-06,
+      "loss": 0.8947,
+      "step": 17163
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.364800305751959e-06,
+      "loss": 0.8566,
+      "step": 17164
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3647047582648577e-06,
+      "loss": 0.6791,
+      "step": 17165
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3646092107777568e-06,
+      "loss": 0.6123,
+      "step": 17166
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3645136632906554e-06,
+      "loss": 0.7036,
+      "step": 17167
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3644181158035544e-06,
+      "loss": 0.5916,
+      "step": 17168
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.364322568316454e-06,
+      "loss": 1.1885,
+      "step": 17169
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3642270208293525e-06,
+      "loss": 0.7528,
+      "step": 17170
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3641314733422515e-06,
+      "loss": 0.5525,
+      "step": 17171
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36403592585515e-06,
+      "loss": 1.5778,
+      "step": 17172
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.363940378368049e-06,
+      "loss": 0.5495,
+      "step": 17173
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.363844830880948e-06,
+      "loss": 1.1192,
+      "step": 17174
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.363749283393847e-06,
+      "loss": 0.8491,
+      "step": 17175
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3636537359067463e-06,
+      "loss": 0.8936,
+      "step": 17176
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.363558188419645e-06,
+      "loss": 0.9267,
+      "step": 17177
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.363462640932544e-06,
+      "loss": 0.9684,
+      "step": 17178
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3633670934454426e-06,
+      "loss": 1.0115,
+      "step": 17179
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3632715459583416e-06,
+      "loss": 1.0725,
+      "step": 17180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3631759984712402e-06,
+      "loss": 1.0207,
+      "step": 17181
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3630804509841393e-06,
+      "loss": 0.6624,
+      "step": 17182
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3629849034970387e-06,
+      "loss": 0.7445,
+      "step": 17183
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3628893560099373e-06,
+      "loss": 0.7136,
+      "step": 17184
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3627938085228364e-06,
+      "loss": 0.9134,
+      "step": 17185
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.362698261035735e-06,
+      "loss": 0.4367,
+      "step": 17186
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.362602713548634e-06,
+      "loss": 0.6026,
+      "step": 17187
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3625071660615327e-06,
+      "loss": 1.0318,
+      "step": 17188
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3624116185744317e-06,
+      "loss": 0.6214,
+      "step": 17189
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3623160710873307e-06,
+      "loss": 0.6649,
+      "step": 17190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3622205236002298e-06,
+      "loss": 0.9142,
+      "step": 17191
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.362124976113129e-06,
+      "loss": 0.622,
+      "step": 17192
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3620294286260274e-06,
+      "loss": 0.7058,
+      "step": 17193
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3619338811389265e-06,
+      "loss": 0.5867,
+      "step": 17194
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361838333651825e-06,
+      "loss": 0.797,
+      "step": 17195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361742786164724e-06,
+      "loss": 0.7937,
+      "step": 17196
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361647238677623e-06,
+      "loss": 1.2788,
+      "step": 17197
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361551691190522e-06,
+      "loss": 1.4404,
+      "step": 17198
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361456143703421e-06,
+      "loss": 0.7614,
+      "step": 17199
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.36136059621632e-06,
+      "loss": 0.9687,
+      "step": 17200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3612650487292185e-06,
+      "loss": 1.1957,
+      "step": 17201
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3611695012421175e-06,
+      "loss": 0.8042,
+      "step": 17202
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3610739537550165e-06,
+      "loss": 0.8543,
+      "step": 17203
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.360978406267915e-06,
+      "loss": 1.1165,
+      "step": 17204
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3608828587808146e-06,
+      "loss": 0.6661,
+      "step": 17205
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3607873112937132e-06,
+      "loss": 0.7055,
+      "step": 17206
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3606917638066123e-06,
+      "loss": 1.3175,
+      "step": 17207
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.360596216319511e-06,
+      "loss": 1.1756,
+      "step": 17208
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.36050066883241e-06,
+      "loss": 1.3176,
+      "step": 17209
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3604051213453085e-06,
+      "loss": 0.8329,
+      "step": 17210
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3603095738582076e-06,
+      "loss": 0.9906,
+      "step": 17211
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.360214026371107e-06,
+      "loss": 0.6314,
+      "step": 17212
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3601184788840056e-06,
+      "loss": 0.8296,
+      "step": 17213
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3600229313969047e-06,
+      "loss": 0.771,
+      "step": 17214
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3599273839098033e-06,
+      "loss": 1.1155,
+      "step": 17215
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3598318364227023e-06,
+      "loss": 1.019,
+      "step": 17216
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.359736288935601e-06,
+      "loss": 1.0027,
+      "step": 17217
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3596407414485e-06,
+      "loss": 0.8169,
+      "step": 17218
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3595451939613995e-06,
+      "loss": 0.6489,
+      "step": 17219
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.359449646474298e-06,
+      "loss": 1.0497,
+      "step": 17220
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.359354098987197e-06,
+      "loss": 0.6884,
+      "step": 17221
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3592585515000957e-06,
+      "loss": 1.1122,
+      "step": 17222
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3591630040129948e-06,
+      "loss": 0.7517,
+      "step": 17223
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3590674565258934e-06,
+      "loss": 1.128,
+      "step": 17224
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3589719090387924e-06,
+      "loss": 0.9741,
+      "step": 17225
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358876361551692e-06,
+      "loss": 0.9473,
+      "step": 17226
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3587808140645905e-06,
+      "loss": 0.9097,
+      "step": 17227
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3586852665774895e-06,
+      "loss": 0.9633,
+      "step": 17228
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358589719090388e-06,
+      "loss": 0.8241,
+      "step": 17229
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358494171603287e-06,
+      "loss": 0.9894,
+      "step": 17230
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358398624116186e-06,
+      "loss": 0.7925,
+      "step": 17231
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358303076629085e-06,
+      "loss": 1.1545,
+      "step": 17232
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3582075291419843e-06,
+      "loss": 1.2261,
+      "step": 17233
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358111981654883e-06,
+      "loss": 1.2829,
+      "step": 17234
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.358016434167782e-06,
+      "loss": 1.1085,
+      "step": 17235
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3579208866806806e-06,
+      "loss": 0.4436,
+      "step": 17236
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3578253391935796e-06,
+      "loss": 1.0357,
+      "step": 17237
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3577297917064782e-06,
+      "loss": 0.9423,
+      "step": 17238
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3576342442193773e-06,
+      "loss": 1.0248,
+      "step": 17239
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.357538696732276e-06,
+      "loss": 0.9407,
+      "step": 17240
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3574431492451753e-06,
+      "loss": 1.512,
+      "step": 17241
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3573476017580744e-06,
+      "loss": 0.7539,
+      "step": 17242
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.357252054270973e-06,
+      "loss": 1.5007,
+      "step": 17243
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.357156506783872e-06,
+      "loss": 0.6834,
+      "step": 17244
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3570609592967706e-06,
+      "loss": 0.9046,
+      "step": 17245
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3569654118096697e-06,
+      "loss": 1.0117,
+      "step": 17246
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3568698643225683e-06,
+      "loss": 0.8249,
+      "step": 17247
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3567743168354678e-06,
+      "loss": 0.8748,
+      "step": 17248
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3566787693483664e-06,
+      "loss": 1.0202,
+      "step": 17249
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3565832218612654e-06,
+      "loss": 1.0713,
+      "step": 17250
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356487674374164e-06,
+      "loss": 1.2212,
+      "step": 17251
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356392126887063e-06,
+      "loss": 0.9909,
+      "step": 17252
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3562965793999617e-06,
+      "loss": 0.6879,
+      "step": 17253
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3562010319128607e-06,
+      "loss": 0.5919,
+      "step": 17254
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.35610548442576e-06,
+      "loss": 1.0407,
+      "step": 17255
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356009936938659e-06,
+      "loss": 0.8466,
+      "step": 17256
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.355914389451558e-06,
+      "loss": 0.9003,
+      "step": 17257
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3558188419644565e-06,
+      "loss": 0.4456,
+      "step": 17258
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3557232944773555e-06,
+      "loss": 0.7972,
+      "step": 17259
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355627746990254e-06,
+      "loss": 1.0111,
+      "step": 17260
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355532199503153e-06,
+      "loss": 0.5543,
+      "step": 17261
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3554366520160526e-06,
+      "loss": 0.9512,
+      "step": 17262
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3553411045289512e-06,
+      "loss": 0.7738,
+      "step": 17263
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3552455570418503e-06,
+      "loss": 0.8495,
+      "step": 17264
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355150009554749e-06,
+      "loss": 0.6995,
+      "step": 17265
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.355054462067648e-06,
+      "loss": 0.5561,
+      "step": 17266
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3549589145805465e-06,
+      "loss": 0.7706,
+      "step": 17267
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3548633670934456e-06,
+      "loss": 0.9949,
+      "step": 17268
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.354767819606345e-06,
+      "loss": 0.9023,
+      "step": 17269
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3546722721192436e-06,
+      "loss": 0.8575,
+      "step": 17270
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3545767246321427e-06,
+      "loss": 0.8254,
+      "step": 17271
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3544811771450413e-06,
+      "loss": 0.7481,
+      "step": 17272
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3543856296579403e-06,
+      "loss": 1.143,
+      "step": 17273
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.354290082170839e-06,
+      "loss": 0.9211,
+      "step": 17274
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.354194534683738e-06,
+      "loss": 0.9285,
+      "step": 17275
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3540989871966366e-06,
+      "loss": 0.7638,
+      "step": 17276
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.354003439709536e-06,
+      "loss": 0.7775,
+      "step": 17277
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.353907892222435e-06,
+      "loss": 0.8842,
+      "step": 17278
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3538123447353337e-06,
+      "loss": 0.7637,
+      "step": 17279
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3537167972482328e-06,
+      "loss": 1.2667,
+      "step": 17280
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3536212497611314e-06,
+      "loss": 1.0026,
+      "step": 17281
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3535257022740304e-06,
+      "loss": 0.8047,
+      "step": 17282
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.353430154786929e-06,
+      "loss": 0.7044,
+      "step": 17283
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3533346072998285e-06,
+      "loss": 0.8802,
+      "step": 17284
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3532390598127275e-06,
+      "loss": 0.9875,
+      "step": 17285
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.353143512325626e-06,
+      "loss": 0.8607,
+      "step": 17286
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.353047964838525e-06,
+      "loss": 0.9302,
+      "step": 17287
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352952417351424e-06,
+      "loss": 1.0054,
+      "step": 17288
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352856869864323e-06,
+      "loss": 0.5894,
+      "step": 17289
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3527613223772215e-06,
+      "loss": 0.9035,
+      "step": 17290
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352665774890121e-06,
+      "loss": 1.1954,
+      "step": 17291
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3525702274030195e-06,
+      "loss": 1.3296,
+      "step": 17292
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3524746799159186e-06,
+      "loss": 0.9369,
+      "step": 17293
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3523791324288176e-06,
+      "loss": 1.1766,
+      "step": 17294
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3522835849417162e-06,
+      "loss": 0.8787,
+      "step": 17295
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3521880374546153e-06,
+      "loss": 0.8168,
+      "step": 17296
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352092489967514e-06,
+      "loss": 0.6288,
+      "step": 17297
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3519969424804133e-06,
+      "loss": 1.1771,
+      "step": 17298
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351901394993312e-06,
+      "loss": 1.0955,
+      "step": 17299
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351805847506211e-06,
+      "loss": 1.0736,
+      "step": 17300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3517103000191096e-06,
+      "loss": 0.6952,
+      "step": 17301
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3516147525320086e-06,
+      "loss": 1.0569,
+      "step": 17302
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3515192050449073e-06,
+      "loss": 0.5636,
+      "step": 17303
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3514236575578063e-06,
+      "loss": 0.7948,
+      "step": 17304
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3513281100707058e-06,
+      "loss": 0.9867,
+      "step": 17305
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3512325625836044e-06,
+      "loss": 0.6428,
+      "step": 17306
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3511370150965034e-06,
+      "loss": 0.5481,
+      "step": 17307
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351041467609402e-06,
+      "loss": 0.9519,
+      "step": 17308
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.350945920122301e-06,
+      "loss": 0.7988,
+      "step": 17309
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3508503726351997e-06,
+      "loss": 0.7154,
+      "step": 17310
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.3507548251480987e-06,
+      "loss": 0.7755,
+      "step": 17311
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3506592776609973e-06,
+      "loss": 0.9717,
+      "step": 17312
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350563730173897e-06,
+      "loss": 1.1228,
+      "step": 17313
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350468182686796e-06,
+      "loss": 0.9504,
+      "step": 17314
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3503726351996945e-06,
+      "loss": 0.6365,
+      "step": 17315
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3502770877125935e-06,
+      "loss": 0.9952,
+      "step": 17316
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350181540225492e-06,
+      "loss": 0.7381,
+      "step": 17317
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.350085992738391e-06,
+      "loss": 0.9438,
+      "step": 17318
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3499904452512898e-06,
+      "loss": 1.4234,
+      "step": 17319
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3498948977641892e-06,
+      "loss": 1.0311,
+      "step": 17320
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3497993502770883e-06,
+      "loss": 0.8598,
+      "step": 17321
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.349703802789987e-06,
+      "loss": 0.9045,
+      "step": 17322
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.349608255302886e-06,
+      "loss": 0.8451,
+      "step": 17323
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3495127078157845e-06,
+      "loss": 0.9012,
+      "step": 17324
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3494171603286836e-06,
+      "loss": 0.8668,
+      "step": 17325
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.349321612841582e-06,
+      "loss": 0.9749,
+      "step": 17326
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3492260653544816e-06,
+      "loss": 0.4983,
+      "step": 17327
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3491305178673807e-06,
+      "loss": 0.5808,
+      "step": 17328
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3490349703802793e-06,
+      "loss": 0.9639,
+      "step": 17329
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3489394228931783e-06,
+      "loss": 1.0293,
+      "step": 17330
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.348843875406077e-06,
+      "loss": 1.1665,
+      "step": 17331
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.348748327918976e-06,
+      "loss": 0.8507,
+      "step": 17332
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3486527804318746e-06,
+      "loss": 1.135,
+      "step": 17333
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.348557232944774e-06,
+      "loss": 0.7056,
+      "step": 17334
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.348461685457673e-06,
+      "loss": 0.7918,
+      "step": 17335
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3483661379705717e-06,
+      "loss": 0.8093,
+      "step": 17336
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3482705904834708e-06,
+      "loss": 1.156,
+      "step": 17337
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3481750429963694e-06,
+      "loss": 0.8255,
+      "step": 17338
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3480794955092684e-06,
+      "loss": 0.8309,
+      "step": 17339
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347983948022167e-06,
+      "loss": 1.2526,
+      "step": 17340
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3478884005350665e-06,
+      "loss": 1.2528,
+      "step": 17341
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347792853047965e-06,
+      "loss": 0.9113,
+      "step": 17342
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347697305560864e-06,
+      "loss": 1.1455,
+      "step": 17343
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347601758073763e-06,
+      "loss": 0.8642,
+      "step": 17344
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347506210586662e-06,
+      "loss": 0.7392,
+      "step": 17345
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347410663099561e-06,
+      "loss": 1.2707,
+      "step": 17346
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3473151156124595e-06,
+      "loss": 0.8078,
+      "step": 17347
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3472195681253585e-06,
+      "loss": 0.8196,
+      "step": 17348
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3471240206382575e-06,
+      "loss": 1.1127,
+      "step": 17349
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3470284731511566e-06,
+      "loss": 0.9749,
+      "step": 17350
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346932925664055e-06,
+      "loss": 0.8285,
+      "step": 17351
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3468373781769542e-06,
+      "loss": 0.8256,
+      "step": 17352
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346741830689853e-06,
+      "loss": 1.2777,
+      "step": 17353
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346646283202752e-06,
+      "loss": 1.2067,
+      "step": 17354
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3465507357156505e-06,
+      "loss": 1.2001,
+      "step": 17355
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.34645518822855e-06,
+      "loss": 0.5923,
+      "step": 17356
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346359640741449e-06,
+      "loss": 0.7503,
+      "step": 17357
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3462640932543476e-06,
+      "loss": 1.0783,
+      "step": 17358
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3461685457672466e-06,
+      "loss": 0.8958,
+      "step": 17359
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3460729982801453e-06,
+      "loss": 1.0076,
+      "step": 17360
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3459774507930443e-06,
+      "loss": 0.9015,
+      "step": 17361
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.345881903305943e-06,
+      "loss": 1.1044,
+      "step": 17362
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3457863558188424e-06,
+      "loss": 0.7657,
+      "step": 17363
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3456908083317414e-06,
+      "loss": 0.9818,
+      "step": 17364
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34559526084464e-06,
+      "loss": 0.7247,
+      "step": 17365
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345499713357539e-06,
+      "loss": 0.8894,
+      "step": 17366
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3454041658704377e-06,
+      "loss": 0.8802,
+      "step": 17367
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3453086183833367e-06,
+      "loss": 1.3796,
+      "step": 17368
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3452130708962353e-06,
+      "loss": 0.8662,
+      "step": 17369
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345117523409135e-06,
+      "loss": 1.0073,
+      "step": 17370
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.345021975922034e-06,
+      "loss": 0.7327,
+      "step": 17371
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3449264284349325e-06,
+      "loss": 1.0801,
+      "step": 17372
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3448308809478315e-06,
+      "loss": 1.2071,
+      "step": 17373
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34473533346073e-06,
+      "loss": 0.654,
+      "step": 17374
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.344639785973629e-06,
+      "loss": 0.5761,
+      "step": 17375
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3445442384865278e-06,
+      "loss": 1.0998,
+      "step": 17376
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3444486909994272e-06,
+      "loss": 0.8562,
+      "step": 17377
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3443531435123263e-06,
+      "loss": 1.0614,
+      "step": 17378
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.344257596025225e-06,
+      "loss": 0.9295,
+      "step": 17379
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.344162048538124e-06,
+      "loss": 0.8723,
+      "step": 17380
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3440665010510225e-06,
+      "loss": 1.167,
+      "step": 17381
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3439709535639216e-06,
+      "loss": 0.7609,
+      "step": 17382
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34387540607682e-06,
+      "loss": 1.0529,
+      "step": 17383
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3437798585897192e-06,
+      "loss": 0.6817,
+      "step": 17384
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3436843111026187e-06,
+      "loss": 1.1381,
+      "step": 17385
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3435887636155173e-06,
+      "loss": 0.7552,
+      "step": 17386
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3434932161284163e-06,
+      "loss": 0.6424,
+      "step": 17387
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.343397668641315e-06,
+      "loss": 0.8755,
+      "step": 17388
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.343302121154214e-06,
+      "loss": 0.8545,
+      "step": 17389
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3432065736671126e-06,
+      "loss": 0.8591,
+      "step": 17390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3431110261800116e-06,
+      "loss": 0.8391,
+      "step": 17391
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3430154786929107e-06,
+      "loss": 1.3266,
+      "step": 17392
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3429199312058097e-06,
+      "loss": 0.8616,
+      "step": 17393
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3428243837187083e-06,
+      "loss": 0.8244,
+      "step": 17394
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3427288362316074e-06,
+      "loss": 1.2796,
+      "step": 17395
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3426332887445064e-06,
+      "loss": 1.0236,
+      "step": 17396
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.342537741257405e-06,
+      "loss": 0.7893,
+      "step": 17397
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.342442193770304e-06,
+      "loss": 0.7337,
+      "step": 17398
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.342346646283203e-06,
+      "loss": 1.1939,
+      "step": 17399
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.342251098796102e-06,
+      "loss": 0.7541,
+      "step": 17400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3421555513090008e-06,
+      "loss": 0.5622,
+      "step": 17401
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3420600038219e-06,
+      "loss": 1.0254,
+      "step": 17402
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3419644563347984e-06,
+      "loss": 0.9654,
+      "step": 17403
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3418689088476975e-06,
+      "loss": 0.78,
+      "step": 17404
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341773361360596e-06,
+      "loss": 0.5209,
+      "step": 17405
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3416778138734955e-06,
+      "loss": 0.9085,
+      "step": 17406
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3415822663863946e-06,
+      "loss": 1.1598,
+      "step": 17407
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341486718899293e-06,
+      "loss": 1.1937,
+      "step": 17408
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3413911714121922e-06,
+      "loss": 0.7621,
+      "step": 17409
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341295623925091e-06,
+      "loss": 0.9406,
+      "step": 17410
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.34120007643799e-06,
+      "loss": 0.813,
+      "step": 17411
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3411045289508885e-06,
+      "loss": 0.6851,
+      "step": 17412
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.341008981463788e-06,
+      "loss": 0.5841,
+      "step": 17413
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.340913433976687e-06,
+      "loss": 0.853,
+      "step": 17414
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3408178864895856e-06,
+      "loss": 0.8773,
+      "step": 17415
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3407223390024846e-06,
+      "loss": 0.6602,
+      "step": 17416
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3406267915153833e-06,
+      "loss": 0.6455,
+      "step": 17417
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3405312440282823e-06,
+      "loss": 0.6956,
+      "step": 17418
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340435696541181e-06,
+      "loss": 0.4667,
+      "step": 17419
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.34034014905408e-06,
+      "loss": 0.7824,
+      "step": 17420
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3402446015669794e-06,
+      "loss": 1.0173,
+      "step": 17421
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340149054079878e-06,
+      "loss": 0.809,
+      "step": 17422
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.340053506592777e-06,
+      "loss": 0.4901,
+      "step": 17423
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3399579591056757e-06,
+      "loss": 0.7656,
+      "step": 17424
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3398624116185747e-06,
+      "loss": 0.8144,
+      "step": 17425
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3397668641314733e-06,
+      "loss": 1.2618,
+      "step": 17426
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3396713166443724e-06,
+      "loss": 0.8233,
+      "step": 17427
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.339575769157272e-06,
+      "loss": 0.7662,
+      "step": 17428
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3394802216701705e-06,
+      "loss": 0.8303,
+      "step": 17429
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3393846741830695e-06,
+      "loss": 1.0298,
+      "step": 17430
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.339289126695968e-06,
+      "loss": 0.6515,
+      "step": 17431
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.339193579208867e-06,
+      "loss": 0.8265,
+      "step": 17432
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3390980317217658e-06,
+      "loss": 1.0645,
+      "step": 17433
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.339002484234665e-06,
+      "loss": 1.1444,
+      "step": 17434
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3389069367475643e-06,
+      "loss": 0.7987,
+      "step": 17435
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.338811389260463e-06,
+      "loss": 1.0036,
+      "step": 17436
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.338715841773362e-06,
+      "loss": 0.587,
+      "step": 17437
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3386202942862605e-06,
+      "loss": 0.9547,
+      "step": 17438
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3385247467991596e-06,
+      "loss": 1.3362,
+      "step": 17439
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.338429199312058e-06,
+      "loss": 0.97,
+      "step": 17440
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3383336518249572e-06,
+      "loss": 0.8461,
+      "step": 17441
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3382381043378563e-06,
+      "loss": 0.7622,
+      "step": 17442
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3381425568507553e-06,
+      "loss": 0.8794,
+      "step": 17443
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.338047009363654e-06,
+      "loss": 1.4668,
+      "step": 17444
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337951461876553e-06,
+      "loss": 1.0265,
+      "step": 17445
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3378559143894516e-06,
+      "loss": 1.5346,
+      "step": 17446
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3377603669023506e-06,
+      "loss": 0.9837,
+      "step": 17447
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3376648194152496e-06,
+      "loss": 0.9514,
+      "step": 17448
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3375692719281487e-06,
+      "loss": 0.8793,
+      "step": 17449
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3374737244410477e-06,
+      "loss": 1.1484,
+      "step": 17450
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3373781769539463e-06,
+      "loss": 0.6652,
+      "step": 17451
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3372826294668454e-06,
+      "loss": 1.1126,
+      "step": 17452
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337187081979744e-06,
+      "loss": 1.159,
+      "step": 17453
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.337091534492643e-06,
+      "loss": 1.0966,
+      "step": 17454
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3369959870055416e-06,
+      "loss": 0.8313,
+      "step": 17455
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3369004395184407e-06,
+      "loss": 0.9636,
+      "step": 17456
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.33680489203134e-06,
+      "loss": 0.7259,
+      "step": 17457
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3367093445442388e-06,
+      "loss": 1.0724,
+      "step": 17458
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336613797057138e-06,
+      "loss": 0.752,
+      "step": 17459
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3365182495700364e-06,
+      "loss": 0.8444,
+      "step": 17460
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3364227020829355e-06,
+      "loss": 1.1431,
+      "step": 17461
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336327154595834e-06,
+      "loss": 1.2892,
+      "step": 17462
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336231607108733e-06,
+      "loss": 1.0736,
+      "step": 17463
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3361360596216326e-06,
+      "loss": 0.813,
+      "step": 17464
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336040512134531e-06,
+      "loss": 1.1964,
+      "step": 17465
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3359449646474302e-06,
+      "loss": 0.5028,
+      "step": 17466
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335849417160329e-06,
+      "loss": 0.8407,
+      "step": 17467
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335753869673228e-06,
+      "loss": 0.8799,
+      "step": 17468
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3356583221861265e-06,
+      "loss": 0.7765,
+      "step": 17469
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3355627746990255e-06,
+      "loss": 0.8463,
+      "step": 17470
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.335467227211925e-06,
+      "loss": 0.6049,
+      "step": 17471
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3353716797248236e-06,
+      "loss": 0.8042,
+      "step": 17472
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3352761322377226e-06,
+      "loss": 1.164,
+      "step": 17473
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3351805847506213e-06,
+      "loss": 0.8857,
+      "step": 17474
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3350850372635203e-06,
+      "loss": 0.8432,
+      "step": 17475
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.334989489776419e-06,
+      "loss": 0.5841,
+      "step": 17476
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.334893942289318e-06,
+      "loss": 0.8107,
+      "step": 17477
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3347983948022174e-06,
+      "loss": 0.6563,
+      "step": 17478
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.334702847315116e-06,
+      "loss": 0.8183,
+      "step": 17479
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.334607299828015e-06,
+      "loss": 0.5737,
+      "step": 17480
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3345117523409137e-06,
+      "loss": 0.9969,
+      "step": 17481
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3344162048538127e-06,
+      "loss": 1.0228,
+      "step": 17482
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3343206573667113e-06,
+      "loss": 1.1614,
+      "step": 17483
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3342251098796104e-06,
+      "loss": 0.6231,
+      "step": 17484
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3341295623925094e-06,
+      "loss": 1.1365,
+      "step": 17485
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3340340149054085e-06,
+      "loss": 1.2857,
+      "step": 17486
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3339384674183075e-06,
+      "loss": 0.887,
+      "step": 17487
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.333842919931206e-06,
+      "loss": 1.0814,
+      "step": 17488
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.333747372444105e-06,
+      "loss": 0.7417,
+      "step": 17489
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3336518249570038e-06,
+      "loss": 0.6664,
+      "step": 17490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.333556277469903e-06,
+      "loss": 0.9469,
+      "step": 17491
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3334607299828014e-06,
+      "loss": 0.7353,
+      "step": 17492
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.333365182495701e-06,
+      "loss": 0.9304,
+      "step": 17493
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3332696350085995e-06,
+      "loss": 1.1532,
+      "step": 17494
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3331740875214985e-06,
+      "loss": 1.0856,
+      "step": 17495
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.333078540034397e-06,
+      "loss": 0.9108,
+      "step": 17496
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332982992547296e-06,
+      "loss": 0.8292,
+      "step": 17497
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3328874450601952e-06,
+      "loss": 0.9204,
+      "step": 17498
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332791897573094e-06,
+      "loss": 0.8213,
+      "step": 17499
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3326963500859933e-06,
+      "loss": 0.8846,
+      "step": 17500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332600802598892e-06,
+      "loss": 1.2395,
+      "step": 17501
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332505255111791e-06,
+      "loss": 0.972,
+      "step": 17502
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3324097076246896e-06,
+      "loss": 0.955,
+      "step": 17503
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3323141601375886e-06,
+      "loss": 1.0436,
+      "step": 17504
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3322186126504872e-06,
+      "loss": 0.7921,
+      "step": 17505
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3321230651633863e-06,
+      "loss": 1.4992,
+      "step": 17506
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3320275176762857e-06,
+      "loss": 0.4233,
+      "step": 17507
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3319319701891843e-06,
+      "loss": 0.6226,
+      "step": 17508
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3318364227020834e-06,
+      "loss": 0.7367,
+      "step": 17509
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331740875214982e-06,
+      "loss": 1.0449,
+      "step": 17510
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331645327727881e-06,
+      "loss": 0.8443,
+      "step": 17511
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3315497802407796e-06,
+      "loss": 1.3683,
+      "step": 17512
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3314542327536787e-06,
+      "loss": 1.3616,
+      "step": 17513
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331358685266578e-06,
+      "loss": 1.3619,
+      "step": 17514
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3312631377794768e-06,
+      "loss": 0.8345,
+      "step": 17515
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.331167590292376e-06,
+      "loss": 0.7412,
+      "step": 17516
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3310720428052744e-06,
+      "loss": 0.956,
+      "step": 17517
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3309764953181735e-06,
+      "loss": 0.5535,
+      "step": 17518
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330880947831072e-06,
+      "loss": 0.6987,
+      "step": 17519
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330785400343971e-06,
+      "loss": 0.5491,
+      "step": 17520
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3306898528568706e-06,
+      "loss": 0.6312,
+      "step": 17521
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330594305369769e-06,
+      "loss": 0.573,
+      "step": 17522
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3304987578826682e-06,
+      "loss": 1.125,
+      "step": 17523
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330403210395567e-06,
+      "loss": 0.4507,
+      "step": 17524
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330307662908466e-06,
+      "loss": 0.9651,
+      "step": 17525
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3302121154213645e-06,
+      "loss": 0.9401,
+      "step": 17526
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3301165679342635e-06,
+      "loss": 1.1318,
+      "step": 17527
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330021020447162e-06,
+      "loss": 0.8367,
+      "step": 17528
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3299254729600616e-06,
+      "loss": 1.085,
+      "step": 17529
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3298299254729606e-06,
+      "loss": 0.5401,
+      "step": 17530
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3297343779858593e-06,
+      "loss": 0.6484,
+      "step": 17531
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3296388304987583e-06,
+      "loss": 0.9324,
+      "step": 17532
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.329543283011657e-06,
+      "loss": 0.9576,
+      "step": 17533
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.329447735524556e-06,
+      "loss": 0.8269,
+      "step": 17534
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3293521880374546e-06,
+      "loss": 1.0361,
+      "step": 17535
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.329256640550354e-06,
+      "loss": 0.9516,
+      "step": 17536
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.329161093063253e-06,
+      "loss": 0.9714,
+      "step": 17537
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3290655455761517e-06,
+      "loss": 0.5875,
+      "step": 17538
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3289699980890507e-06,
+      "loss": 0.9335,
+      "step": 17539
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3288744506019493e-06,
+      "loss": 0.6713,
+      "step": 17540
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3287789031148484e-06,
+      "loss": 0.631,
+      "step": 17541
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.328683355627747e-06,
+      "loss": 1.1695,
+      "step": 17542
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3285878081406465e-06,
+      "loss": 1.1031,
+      "step": 17543
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.328492260653545e-06,
+      "loss": 0.8023,
+      "step": 17544
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.328396713166444e-06,
+      "loss": 1.1758,
+      "step": 17545
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3283011656793427e-06,
+      "loss": 1.2256,
+      "step": 17546
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3282056181922418e-06,
+      "loss": 0.6748,
+      "step": 17547
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3281100707051404e-06,
+      "loss": 0.9824,
+      "step": 17548
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3280145232180394e-06,
+      "loss": 0.9109,
+      "step": 17549
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327918975730939e-06,
+      "loss": 0.8254,
+      "step": 17550
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3278234282438375e-06,
+      "loss": 1.0875,
+      "step": 17551
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3277278807567365e-06,
+      "loss": 1.2215,
+      "step": 17552
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327632333269635e-06,
+      "loss": 0.7227,
+      "step": 17553
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327536785782534e-06,
+      "loss": 0.9001,
+      "step": 17554
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327441238295433e-06,
+      "loss": 0.8638,
+      "step": 17555
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327345690808332e-06,
+      "loss": 0.6582,
+      "step": 17556
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3272501433212313e-06,
+      "loss": 0.854,
+      "step": 17557
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32715459583413e-06,
+      "loss": 0.7526,
+      "step": 17558
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327059048347029e-06,
+      "loss": 0.945,
+      "step": 17559
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3269635008599276e-06,
+      "loss": 0.7527,
+      "step": 17560
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3268679533728266e-06,
+      "loss": 1.5271,
+      "step": 17561
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3267724058857252e-06,
+      "loss": 0.9441,
+      "step": 17562
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3266768583986243e-06,
+      "loss": 1.2366,
+      "step": 17563
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326581310911523e-06,
+      "loss": 0.9349,
+      "step": 17564
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3264857634244223e-06,
+      "loss": 0.7253,
+      "step": 17565
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3263902159373214e-06,
+      "loss": 0.4681,
+      "step": 17566
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32629466845022e-06,
+      "loss": 0.8172,
+      "step": 17567
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326199120963119e-06,
+      "loss": 0.5882,
+      "step": 17568
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3261035734760176e-06,
+      "loss": 0.8936,
+      "step": 17569
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3260080259889167e-06,
+      "loss": 0.8085,
+      "step": 17570
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3259124785018153e-06,
+      "loss": 0.6647,
+      "step": 17571
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3258169310147148e-06,
+      "loss": 0.8365,
+      "step": 17572
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.325721383527614e-06,
+      "loss": 1.1057,
+      "step": 17573
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3256258360405124e-06,
+      "loss": 0.833,
+      "step": 17574
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3255302885534115e-06,
+      "loss": 1.0032,
+      "step": 17575
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.32543474106631e-06,
+      "loss": 0.949,
+      "step": 17576
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325339193579209e-06,
+      "loss": 0.6229,
+      "step": 17577
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3252436460921077e-06,
+      "loss": 0.9568,
+      "step": 17578
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.325148098605007e-06,
+      "loss": 0.5934,
+      "step": 17579
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3250525511179062e-06,
+      "loss": 1.2681,
+      "step": 17580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.324957003630805e-06,
+      "loss": 1.357,
+      "step": 17581
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.324861456143704e-06,
+      "loss": 1.3088,
+      "step": 17582
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3247659086566025e-06,
+      "loss": 0.8471,
+      "step": 17583
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3246703611695015e-06,
+      "loss": 0.9959,
+      "step": 17584
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3245748136824e-06,
+      "loss": 1.2004,
+      "step": 17585
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3244792661952996e-06,
+      "loss": 1.1001,
+      "step": 17586
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3243837187081982e-06,
+      "loss": 1.0413,
+      "step": 17587
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3242881712210973e-06,
+      "loss": 1.0629,
+      "step": 17588
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3241926237339963e-06,
+      "loss": 0.9389,
+      "step": 17589
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.324097076246895e-06,
+      "loss": 1.1168,
+      "step": 17590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.324001528759794e-06,
+      "loss": 0.6847,
+      "step": 17591
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3239059812726926e-06,
+      "loss": 1.4248,
+      "step": 17592
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.323810433785592e-06,
+      "loss": 0.9793,
+      "step": 17593
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3237148862984906e-06,
+      "loss": 1.6145,
+      "step": 17594
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3236193388113897e-06,
+      "loss": 1.1223,
+      "step": 17595
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3235237913242883e-06,
+      "loss": 0.5514,
+      "step": 17596
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3234282438371873e-06,
+      "loss": 0.553,
+      "step": 17597
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.323332696350086e-06,
+      "loss": 1.0081,
+      "step": 17598
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.323237148862985e-06,
+      "loss": 0.6709,
+      "step": 17599
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.323141601375884e-06,
+      "loss": 0.8875,
+      "step": 17600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.323046053888783e-06,
+      "loss": 0.988,
+      "step": 17601
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322950506401682e-06,
+      "loss": 0.7138,
+      "step": 17602
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3228549589145807e-06,
+      "loss": 1.3287,
+      "step": 17603
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3227594114274798e-06,
+      "loss": 0.7744,
+      "step": 17604
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3226638639403784e-06,
+      "loss": 0.9098,
+      "step": 17605
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3225683164532774e-06,
+      "loss": 1.1187,
+      "step": 17606
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322472768966176e-06,
+      "loss": 0.5126,
+      "step": 17607
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3223772214790755e-06,
+      "loss": 0.7138,
+      "step": 17608
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3222816739919745e-06,
+      "loss": 0.6666,
+      "step": 17609
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322186126504873e-06,
+      "loss": 0.7222,
+      "step": 17610
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322090579017772e-06,
+      "loss": 0.9022,
+      "step": 17611
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321995031530671e-06,
+      "loss": 0.8795,
+      "step": 17612
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.32189948404357e-06,
+      "loss": 0.6379,
+      "step": 17613
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3218039365564684e-06,
+      "loss": 0.9551,
+      "step": 17614
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321708389069368e-06,
+      "loss": 1.0621,
+      "step": 17615
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321612841582267e-06,
+      "loss": 0.6897,
+      "step": 17616
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3215172940951656e-06,
+      "loss": 0.877,
+      "step": 17617
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3214217466080646e-06,
+      "loss": 0.8533,
+      "step": 17618
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3213261991209632e-06,
+      "loss": 0.7809,
+      "step": 17619
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3212306516338623e-06,
+      "loss": 0.6996,
+      "step": 17620
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321135104146761e-06,
+      "loss": 1.2337,
+      "step": 17621
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3210395566596603e-06,
+      "loss": 0.7228,
+      "step": 17622
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3209440091725594e-06,
+      "loss": 0.646,
+      "step": 17623
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.320848461685458e-06,
+      "loss": 1.1503,
+      "step": 17624
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.320752914198357e-06,
+      "loss": 0.8846,
+      "step": 17625
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3206573667112556e-06,
+      "loss": 0.9393,
+      "step": 17626
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3205618192241547e-06,
+      "loss": 0.981,
+      "step": 17627
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3204662717370533e-06,
+      "loss": 0.8021,
+      "step": 17628
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3203707242499528e-06,
+      "loss": 1.0879,
+      "step": 17629
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.320275176762852e-06,
+      "loss": 0.8604,
+      "step": 17630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3201796292757504e-06,
+      "loss": 1.1253,
+      "step": 17631
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3200840817886494e-06,
+      "loss": 1.1811,
+      "step": 17632
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319988534301548e-06,
+      "loss": 1.1175,
+      "step": 17633
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319892986814447e-06,
+      "loss": 0.9404,
+      "step": 17634
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3197974393273457e-06,
+      "loss": 0.7175,
+      "step": 17635
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3197018918402448e-06,
+      "loss": 0.7195,
+      "step": 17636
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319606344353144e-06,
+      "loss": 0.6099,
+      "step": 17637
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319510796866043e-06,
+      "loss": 0.5057,
+      "step": 17638
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319415249378942e-06,
+      "loss": 1.0062,
+      "step": 17639
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3193197018918405e-06,
+      "loss": 0.7777,
+      "step": 17640
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3192241544047395e-06,
+      "loss": 0.715,
+      "step": 17641
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319128606917638e-06,
+      "loss": 0.5531,
+      "step": 17642
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.319033059430537e-06,
+      "loss": 0.9725,
+      "step": 17643
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3189375119434362e-06,
+      "loss": 0.5984,
+      "step": 17644
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3188419644563353e-06,
+      "loss": 0.8398,
+      "step": 17645
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.318746416969234e-06,
+      "loss": 1.0817,
+      "step": 17646
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.318650869482133e-06,
+      "loss": 1.0283,
+      "step": 17647
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3185553219950315e-06,
+      "loss": 0.7337,
+      "step": 17648
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3184597745079306e-06,
+      "loss": 1.032,
+      "step": 17649
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.318364227020829e-06,
+      "loss": 0.9658,
+      "step": 17650
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3182686795337286e-06,
+      "loss": 0.8469,
+      "step": 17651
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3181731320466277e-06,
+      "loss": 0.8533,
+      "step": 17652
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3180775845595263e-06,
+      "loss": 0.9913,
+      "step": 17653
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3179820370724253e-06,
+      "loss": 0.9202,
+      "step": 17654
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317886489585324e-06,
+      "loss": 1.1385,
+      "step": 17655
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317790942098223e-06,
+      "loss": 0.4965,
+      "step": 17656
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3176953946111216e-06,
+      "loss": 1.205,
+      "step": 17657
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317599847124021e-06,
+      "loss": 1.0889,
+      "step": 17658
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.31750429963692e-06,
+      "loss": 1.2168,
+      "step": 17659
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3174087521498187e-06,
+      "loss": 0.7536,
+      "step": 17660
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3173132046627178e-06,
+      "loss": 0.9861,
+      "step": 17661
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3172176571756164e-06,
+      "loss": 1.3299,
+      "step": 17662
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3171221096885154e-06,
+      "loss": 0.9133,
+      "step": 17663
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317026562201414e-06,
+      "loss": 1.343,
+      "step": 17664
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3169310147143135e-06,
+      "loss": 0.9755,
+      "step": 17665
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3168354672272125e-06,
+      "loss": 0.9512,
+      "step": 17666
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316739919740111e-06,
+      "loss": 0.6579,
+      "step": 17667
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.31664437225301e-06,
+      "loss": 0.5657,
+      "step": 17668
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316548824765909e-06,
+      "loss": 0.5317,
+      "step": 17669
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316453277278808e-06,
+      "loss": 0.9409,
+      "step": 17670
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3163577297917064e-06,
+      "loss": 1.3988,
+      "step": 17671
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3162621823046055e-06,
+      "loss": 0.7459,
+      "step": 17672
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316166634817505e-06,
+      "loss": 0.8338,
+      "step": 17673
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3160710873304036e-06,
+      "loss": 0.9767,
+      "step": 17674
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3159755398433026e-06,
+      "loss": 0.6507,
+      "step": 17675
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3158799923562012e-06,
+      "loss": 1.0478,
+      "step": 17676
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3157844448691003e-06,
+      "loss": 0.6145,
+      "step": 17677
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.315688897381999e-06,
+      "loss": 0.9863,
+      "step": 17678
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315593349894898e-06,
+      "loss": 1.3088,
+      "step": 17679
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3154978024077974e-06,
+      "loss": 1.4193,
+      "step": 17680
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315402254920696e-06,
+      "loss": 0.7298,
+      "step": 17681
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.315306707433595e-06,
+      "loss": 1.646,
+      "step": 17682
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3152111599464936e-06,
+      "loss": 0.7479,
+      "step": 17683
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3151156124593927e-06,
+      "loss": 0.6672,
+      "step": 17684
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3150200649722913e-06,
+      "loss": 1.1689,
+      "step": 17685
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3149245174851903e-06,
+      "loss": 0.8865,
+      "step": 17686
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3148289699980894e-06,
+      "loss": 1.6203,
+      "step": 17687
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3147334225109884e-06,
+      "loss": 1.2249,
+      "step": 17688
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.314637875023887e-06,
+      "loss": 1.653,
+      "step": 17689
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.314542327536786e-06,
+      "loss": 1.1404,
+      "step": 17690
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.314446780049685e-06,
+      "loss": 1.0399,
+      "step": 17691
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3143512325625837e-06,
+      "loss": 0.6725,
+      "step": 17692
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3142556850754828e-06,
+      "loss": 0.5192,
+      "step": 17693
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.314160137588382e-06,
+      "loss": 0.8237,
+      "step": 17694
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.314064590101281e-06,
+      "loss": 0.4451,
+      "step": 17695
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3139690426141794e-06,
+      "loss": 0.5531,
+      "step": 17696
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3138734951270785e-06,
+      "loss": 0.9422,
+      "step": 17697
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.313777947639977e-06,
+      "loss": 0.94,
+      "step": 17698
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.313682400152876e-06,
+      "loss": 0.8465,
+      "step": 17699
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3135868526657748e-06,
+      "loss": 0.7482,
+      "step": 17700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3134913051786742e-06,
+      "loss": 0.5637,
+      "step": 17701
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3133957576915733e-06,
+      "loss": 1.0057,
+      "step": 17702
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.313300210204472e-06,
+      "loss": 0.9773,
+      "step": 17703
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.313204662717371e-06,
+      "loss": 0.8387,
+      "step": 17704
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3131091152302695e-06,
+      "loss": 0.462,
+      "step": 17705
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3130135677431686e-06,
+      "loss": 1.0036,
+      "step": 17706
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312918020256067e-06,
+      "loss": 0.6281,
+      "step": 17707
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3128224727689662e-06,
+      "loss": 1.3078,
+      "step": 17708
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3127269252818657e-06,
+      "loss": 0.7649,
+      "step": 17709
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3126313777947643e-06,
+      "loss": 0.9539,
+      "step": 17710
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3125358303076633e-06,
+      "loss": 0.7864,
+      "step": 17711
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312440282820562e-06,
+      "loss": 0.7265,
+      "step": 17712
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312344735333461e-06,
+      "loss": 0.7278,
+      "step": 17713
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3122491878463596e-06,
+      "loss": 0.7162,
+      "step": 17714
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3121536403592586e-06,
+      "loss": 1.3928,
+      "step": 17715
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312058092872158e-06,
+      "loss": 0.9588,
+      "step": 17716
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3119625453850567e-06,
+      "loss": 0.8008,
+      "step": 17717
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3118669978979558e-06,
+      "loss": 0.8027,
+      "step": 17718
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3117714504108544e-06,
+      "loss": 0.7365,
+      "step": 17719
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3116759029237534e-06,
+      "loss": 0.7139,
+      "step": 17720
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311580355436652e-06,
+      "loss": 1.1579,
+      "step": 17721
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311484807949551e-06,
+      "loss": 0.8586,
+      "step": 17722
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3113892604624505e-06,
+      "loss": 0.9564,
+      "step": 17723
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311293712975349e-06,
+      "loss": 1.1146,
+      "step": 17724
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311198165488248e-06,
+      "loss": 0.965,
+      "step": 17725
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311102618001147e-06,
+      "loss": 0.7619,
+      "step": 17726
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311007070514046e-06,
+      "loss": 1.0367,
+      "step": 17727
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3109115230269444e-06,
+      "loss": 0.7303,
+      "step": 17728
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3108159755398435e-06,
+      "loss": 1.1764,
+      "step": 17729
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.310720428052743e-06,
+      "loss": 1.0374,
+      "step": 17730
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3106248805656416e-06,
+      "loss": 0.5946,
+      "step": 17731
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3105293330785406e-06,
+      "loss": 0.5997,
+      "step": 17732
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3104337855914392e-06,
+      "loss": 0.9281,
+      "step": 17733
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3103382381043383e-06,
+      "loss": 0.8607,
+      "step": 17734
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310242690617237e-06,
+      "loss": 1.2098,
+      "step": 17735
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310147143130136e-06,
+      "loss": 0.697,
+      "step": 17736
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.310051595643035e-06,
+      "loss": 1.2317,
+      "step": 17737
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.309956048155934e-06,
+      "loss": 1.0225,
+      "step": 17738
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3098605006688326e-06,
+      "loss": 0.9741,
+      "step": 17739
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3097649531817316e-06,
+      "loss": 0.8687,
+      "step": 17740
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3096694056946303e-06,
+      "loss": 1.0488,
+      "step": 17741
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3095738582075293e-06,
+      "loss": 0.8419,
+      "step": 17742
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3094783107204283e-06,
+      "loss": 0.8941,
+      "step": 17743
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.309382763233327e-06,
+      "loss": 0.6531,
+      "step": 17744
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3092872157462264e-06,
+      "loss": 1.2046,
+      "step": 17745
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.309191668259125e-06,
+      "loss": 1.1317,
+      "step": 17746
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.309096120772024e-06,
+      "loss": 0.9174,
+      "step": 17747
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3090005732849227e-06,
+      "loss": 0.9588,
+      "step": 17748
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3089050257978217e-06,
+      "loss": 0.9578,
+      "step": 17749
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3088094783107203e-06,
+      "loss": 0.9825,
+      "step": 17750
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3087139308236194e-06,
+      "loss": 1.1293,
+      "step": 17751
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.308618383336519e-06,
+      "loss": 0.7852,
+      "step": 17752
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3085228358494174e-06,
+      "loss": 0.8873,
+      "step": 17753
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3084272883623165e-06,
+      "loss": 0.6208,
+      "step": 17754
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.308331740875215e-06,
+      "loss": 0.8254,
+      "step": 17755
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.308236193388114e-06,
+      "loss": 0.6934,
+      "step": 17756
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3081406459010128e-06,
+      "loss": 0.6468,
+      "step": 17757
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.308045098413912e-06,
+      "loss": 1.0302,
+      "step": 17758
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3079495509268113e-06,
+      "loss": 1.1291,
+      "step": 17759
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.30785400343971e-06,
+      "loss": 1.1751,
+      "step": 17760
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.307758455952609e-06,
+      "loss": 1.3623,
+      "step": 17761
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3076629084655075e-06,
+      "loss": 0.7412,
+      "step": 17762
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3075673609784066e-06,
+      "loss": 0.8766,
+      "step": 17763
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.307471813491305e-06,
+      "loss": 0.8541,
+      "step": 17764
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3073762660042042e-06,
+      "loss": 0.7245,
+      "step": 17765
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3072807185171037e-06,
+      "loss": 0.7952,
+      "step": 17766
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3071851710300023e-06,
+      "loss": 1.3011,
+      "step": 17767
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3070896235429013e-06,
+      "loss": 1.1008,
+      "step": 17768
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3069940760558e-06,
+      "loss": 0.9631,
+      "step": 17769
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306898528568699e-06,
+      "loss": 0.8029,
+      "step": 17770
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3068029810815976e-06,
+      "loss": 0.857,
+      "step": 17771
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3067074335944966e-06,
+      "loss": 1.0196,
+      "step": 17772
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306611886107396e-06,
+      "loss": 1.0892,
+      "step": 17773
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3065163386202947e-06,
+      "loss": 0.9328,
+      "step": 17774
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3064207911331938e-06,
+      "loss": 1.1489,
+      "step": 17775
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3063252436460924e-06,
+      "loss": 1.0933,
+      "step": 17776
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3062296961589914e-06,
+      "loss": 0.5377,
+      "step": 17777
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.30613414867189e-06,
+      "loss": 0.8804,
+      "step": 17778
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306038601184789e-06,
+      "loss": 1.1853,
+      "step": 17779
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3059430536976877e-06,
+      "loss": 0.6249,
+      "step": 17780
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305847506210587e-06,
+      "loss": 1.0773,
+      "step": 17781
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305751958723486e-06,
+      "loss": 1.2243,
+      "step": 17782
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305656411236385e-06,
+      "loss": 0.5712,
+      "step": 17783
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.305560863749284e-06,
+      "loss": 0.7306,
+      "step": 17784
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3054653162621824e-06,
+      "loss": 0.7711,
+      "step": 17785
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3053697687750815e-06,
+      "loss": 1.0489,
+      "step": 17786
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30527422128798e-06,
+      "loss": 0.7784,
+      "step": 17787
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3051786738008796e-06,
+      "loss": 0.816,
+      "step": 17788
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.305083126313778e-06,
+      "loss": 1.0863,
+      "step": 17789
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3049875788266772e-06,
+      "loss": 0.5619,
+      "step": 17790
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.304892031339576e-06,
+      "loss": 0.8766,
+      "step": 17791
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.304796483852475e-06,
+      "loss": 0.6145,
+      "step": 17792
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.304700936365374e-06,
+      "loss": 0.8366,
+      "step": 17793
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3046053888782725e-06,
+      "loss": 0.8836,
+      "step": 17794
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.304509841391172e-06,
+      "loss": 0.8125,
+      "step": 17795
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3044142939040706e-06,
+      "loss": 0.882,
+      "step": 17796
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3043187464169696e-06,
+      "loss": 1.1182,
+      "step": 17797
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3042231989298683e-06,
+      "loss": 0.78,
+      "step": 17798
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3041276514427673e-06,
+      "loss": 0.7307,
+      "step": 17799
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.304032103955666e-06,
+      "loss": 0.9468,
+      "step": 17800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.303936556468565e-06,
+      "loss": 0.9118,
+      "step": 17801
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3038410089814644e-06,
+      "loss": 0.7767,
+      "step": 17802
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.303745461494363e-06,
+      "loss": 1.0328,
+      "step": 17803
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.303649914007262e-06,
+      "loss": 0.9039,
+      "step": 17804
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3035543665201607e-06,
+      "loss": 0.9191,
+      "step": 17805
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3034588190330597e-06,
+      "loss": 0.7769,
+      "step": 17806
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3033632715459583e-06,
+      "loss": 0.7624,
+      "step": 17807
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3032677240588574e-06,
+      "loss": 0.9413,
+      "step": 17808
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.303172176571757e-06,
+      "loss": 1.1871,
+      "step": 17809
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3030766290846554e-06,
+      "loss": 0.5556,
+      "step": 17810
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3029810815975545e-06,
+      "loss": 0.9965,
+      "step": 17811
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302885534110453e-06,
+      "loss": 1.0878,
+      "step": 17812
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302789986623352e-06,
+      "loss": 0.8294,
+      "step": 17813
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3026944391362508e-06,
+      "loss": 0.7394,
+      "step": 17814
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.30259889164915e-06,
+      "loss": 0.8109,
+      "step": 17815
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3025033441620484e-06,
+      "loss": 1.1628,
+      "step": 17816
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302407796674948e-06,
+      "loss": 0.5425,
+      "step": 17817
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302312249187847e-06,
+      "loss": 0.8017,
+      "step": 17818
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3022167017007455e-06,
+      "loss": 0.7522,
+      "step": 17819
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3021211542136446e-06,
+      "loss": 0.7297,
+      "step": 17820
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302025606726543e-06,
+      "loss": 0.8803,
+      "step": 17821
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3019300592394422e-06,
+      "loss": 0.8076,
+      "step": 17822
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.301834511752341e-06,
+      "loss": 0.9548,
+      "step": 17823
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3017389642652403e-06,
+      "loss": 0.6474,
+      "step": 17824
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3016434167781393e-06,
+      "loss": 0.8698,
+      "step": 17825
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.301547869291038e-06,
+      "loss": 1.0293,
+      "step": 17826
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.301452321803937e-06,
+      "loss": 0.7042,
+      "step": 17827
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3013567743168356e-06,
+      "loss": 0.8769,
+      "step": 17828
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3012612268297346e-06,
+      "loss": 0.9225,
+      "step": 17829
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3011656793426333e-06,
+      "loss": 0.7226,
+      "step": 17830
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3010701318555327e-06,
+      "loss": 0.97,
+      "step": 17831
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3009745843684318e-06,
+      "loss": 0.7107,
+      "step": 17832
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3008790368813304e-06,
+      "loss": 0.7728,
+      "step": 17833
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.3007834893942294e-06,
+      "loss": 0.9968,
+      "step": 17834
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.300687941907128e-06,
+      "loss": 1.1908,
+      "step": 17835
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300592394420027e-06,
+      "loss": 0.8355,
+      "step": 17836
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3004968469329257e-06,
+      "loss": 0.708,
+      "step": 17837
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300401299445825e-06,
+      "loss": 0.3458,
+      "step": 17838
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3003057519587238e-06,
+      "loss": 0.8902,
+      "step": 17839
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.300210204471623e-06,
+      "loss": 0.5463,
+      "step": 17840
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3001146569845214e-06,
+      "loss": 1.0335,
+      "step": 17841
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.3000191094974204e-06,
+      "loss": 1.1199,
+      "step": 17842
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299923562010319e-06,
+      "loss": 0.9935,
+      "step": 17843
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299828014523218e-06,
+      "loss": 0.8568,
+      "step": 17844
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2997324670361176e-06,
+      "loss": 1.1076,
+      "step": 17845
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299636919549016e-06,
+      "loss": 0.7709,
+      "step": 17846
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2995413720619152e-06,
+      "loss": 1.219,
+      "step": 17847
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299445824574814e-06,
+      "loss": 0.7277,
+      "step": 17848
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299350277087713e-06,
+      "loss": 0.9238,
+      "step": 17849
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2992547296006115e-06,
+      "loss": 0.8512,
+      "step": 17850
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2991591821135105e-06,
+      "loss": 1.2499,
+      "step": 17851
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.299063634626409e-06,
+      "loss": 0.8649,
+      "step": 17852
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2989680871393086e-06,
+      "loss": 1.1339,
+      "step": 17853
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2988725396522076e-06,
+      "loss": 1.7467,
+      "step": 17854
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2987769921651063e-06,
+      "loss": 0.557,
+      "step": 17855
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2986814446780053e-06,
+      "loss": 0.8878,
+      "step": 17856
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.298585897190904e-06,
+      "loss": 0.8972,
+      "step": 17857
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.298490349703803e-06,
+      "loss": 0.7167,
+      "step": 17858
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2983948022167016e-06,
+      "loss": 0.9863,
+      "step": 17859
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.298299254729601e-06,
+      "loss": 0.8024,
+      "step": 17860
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2982037072425e-06,
+      "loss": 0.7874,
+      "step": 17861
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2981081597553987e-06,
+      "loss": 1.0294,
+      "step": 17862
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2980126122682977e-06,
+      "loss": 0.7043,
+      "step": 17863
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2979170647811963e-06,
+      "loss": 1.4279,
+      "step": 17864
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2978215172940954e-06,
+      "loss": 0.9079,
+      "step": 17865
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297725969806994e-06,
+      "loss": 0.6413,
+      "step": 17866
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2976304223198934e-06,
+      "loss": 0.8069,
+      "step": 17867
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2975348748327925e-06,
+      "loss": 1.3793,
+      "step": 17868
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297439327345691e-06,
+      "loss": 0.7327,
+      "step": 17869
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.29734377985859e-06,
+      "loss": 0.9109,
+      "step": 17870
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2972482323714888e-06,
+      "loss": 1.0083,
+      "step": 17871
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297152684884388e-06,
+      "loss": 0.9936,
+      "step": 17872
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2970571373972864e-06,
+      "loss": 0.6768,
+      "step": 17873
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296961589910186e-06,
+      "loss": 1.6203,
+      "step": 17874
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296866042423085e-06,
+      "loss": 0.7349,
+      "step": 17875
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2967704949359835e-06,
+      "loss": 1.0938,
+      "step": 17876
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2966749474488826e-06,
+      "loss": 0.9138,
+      "step": 17877
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296579399961781e-06,
+      "loss": 0.9315,
+      "step": 17878
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2964838524746802e-06,
+      "loss": 0.7476,
+      "step": 17879
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296388304987579e-06,
+      "loss": 0.7655,
+      "step": 17880
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2962927575004783e-06,
+      "loss": 0.6478,
+      "step": 17881
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296197210013377e-06,
+      "loss": 0.8139,
+      "step": 17882
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296101662526276e-06,
+      "loss": 1.0335,
+      "step": 17883
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296006115039175e-06,
+      "loss": 1.1101,
+      "step": 17884
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2959105675520736e-06,
+      "loss": 1.2174,
+      "step": 17885
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2958150200649726e-06,
+      "loss": 0.8877,
+      "step": 17886
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2957194725778713e-06,
+      "loss": 0.985,
+      "step": 17887
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2956239250907703e-06,
+      "loss": 1.305,
+      "step": 17888
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2955283776036693e-06,
+      "loss": 1.6192,
+      "step": 17889
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2954328301165684e-06,
+      "loss": 0.6566,
+      "step": 17890
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295337282629467e-06,
+      "loss": 1.151,
+      "step": 17891
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.295241735142366e-06,
+      "loss": 0.8792,
+      "step": 17892
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2951461876552646e-06,
+      "loss": 0.6354,
+      "step": 17893
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2950506401681637e-06,
+      "loss": 0.7804,
+      "step": 17894
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2949550926810627e-06,
+      "loss": 1.0616,
+      "step": 17895
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2948595451939618e-06,
+      "loss": 0.9191,
+      "step": 17896
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.294763997706861e-06,
+      "loss": 0.7622,
+      "step": 17897
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2946684502197594e-06,
+      "loss": 1.0204,
+      "step": 17898
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2945729027326584e-06,
+      "loss": 0.9246,
+      "step": 17899
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.294477355245557e-06,
+      "loss": 0.7133,
+      "step": 17900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.294381807758456e-06,
+      "loss": 0.6171,
+      "step": 17901
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2942862602713547e-06,
+      "loss": 0.9844,
+      "step": 17902
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.294190712784254e-06,
+      "loss": 1.0724,
+      "step": 17903
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2940951652971532e-06,
+      "loss": 0.5389,
+      "step": 17904
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.293999617810052e-06,
+      "loss": 0.9503,
+      "step": 17905
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.293904070322951e-06,
+      "loss": 0.6597,
+      "step": 17906
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2938085228358495e-06,
+      "loss": 0.7719,
+      "step": 17907
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2937129753487485e-06,
+      "loss": 0.7068,
+      "step": 17908
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.293617427861647e-06,
+      "loss": 1.1811,
+      "step": 17909
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2935218803745466e-06,
+      "loss": 0.7305,
+      "step": 17910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2934263328874456e-06,
+      "loss": 1.2842,
+      "step": 17911
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2933307854003443e-06,
+      "loss": 0.9547,
+      "step": 17912
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2932352379132433e-06,
+      "loss": 0.6765,
+      "step": 17913
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.293139690426142e-06,
+      "loss": 0.7598,
+      "step": 17914
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.293044142939041e-06,
+      "loss": 0.8763,
+      "step": 17915
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2929485954519396e-06,
+      "loss": 0.9729,
+      "step": 17916
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292853047964839e-06,
+      "loss": 1.0809,
+      "step": 17917
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292757500477738e-06,
+      "loss": 1.1145,
+      "step": 17918
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2926619529906367e-06,
+      "loss": 0.902,
+      "step": 17919
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2925664055035357e-06,
+      "loss": 0.732,
+      "step": 17920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2924708580164343e-06,
+      "loss": 0.6729,
+      "step": 17921
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2923753105293334e-06,
+      "loss": 1.3196,
+      "step": 17922
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292279763042232e-06,
+      "loss": 0.6838,
+      "step": 17923
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292184215555131e-06,
+      "loss": 0.8147,
+      "step": 17924
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2920886680680305e-06,
+      "loss": 1.0705,
+      "step": 17925
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291993120580929e-06,
+      "loss": 1.3638,
+      "step": 17926
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291897573093828e-06,
+      "loss": 0.88,
+      "step": 17927
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2918020256067268e-06,
+      "loss": 1.0644,
+      "step": 17928
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291706478119626e-06,
+      "loss": 1.0101,
+      "step": 17929
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2916109306325244e-06,
+      "loss": 1.0987,
+      "step": 17930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2915153831454234e-06,
+      "loss": 0.7608,
+      "step": 17931
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2914198356583225e-06,
+      "loss": 1.1268,
+      "step": 17932
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2913242881712215e-06,
+      "loss": 0.808,
+      "step": 17933
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2912287406841206e-06,
+      "loss": 1.4505,
+      "step": 17934
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291133193197019e-06,
+      "loss": 1.159,
+      "step": 17935
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2910376457099182e-06,
+      "loss": 0.957,
+      "step": 17936
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290942098222817e-06,
+      "loss": 0.8108,
+      "step": 17937
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290846550735716e-06,
+      "loss": 0.8873,
+      "step": 17938
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290751003248615e-06,
+      "loss": 1.1273,
+      "step": 17939
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290655455761514e-06,
+      "loss": 0.8943,
+      "step": 17940
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2905599082744126e-06,
+      "loss": 0.8685,
+      "step": 17941
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2904643607873116e-06,
+      "loss": 0.6715,
+      "step": 17942
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2903688133002102e-06,
+      "loss": 0.601,
+      "step": 17943
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2902732658131093e-06,
+      "loss": 0.6162,
+      "step": 17944
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.290177718326008e-06,
+      "loss": 0.884,
+      "step": 17945
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2900821708389073e-06,
+      "loss": 0.9057,
+      "step": 17946
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2899866233518064e-06,
+      "loss": 1.0323,
+      "step": 17947
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.289891075864705e-06,
+      "loss": 0.9413,
+      "step": 17948
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.289795528377604e-06,
+      "loss": 1.1008,
+      "step": 17949
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2896999808905026e-06,
+      "loss": 0.5547,
+      "step": 17950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2896044334034017e-06,
+      "loss": 0.6609,
+      "step": 17951
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2895088859163003e-06,
+      "loss": 0.6725,
+      "step": 17952
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2894133384291998e-06,
+      "loss": 1.0745,
+      "step": 17953
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.289317790942099e-06,
+      "loss": 0.4465,
+      "step": 17954
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2892222434549974e-06,
+      "loss": 0.553,
+      "step": 17955
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2891266959678964e-06,
+      "loss": 0.9701,
+      "step": 17956
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.289031148480795e-06,
+      "loss": 1.1401,
+      "step": 17957
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.288935600993694e-06,
+      "loss": 1.413,
+      "step": 17958
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2888400535065927e-06,
+      "loss": 1.1485,
+      "step": 17959
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2887445060194918e-06,
+      "loss": 0.8091,
+      "step": 17960
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2886489585323912e-06,
+      "loss": 1.1786,
+      "step": 17961
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.28855341104529e-06,
+      "loss": 0.4444,
+      "step": 17962
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.288457863558189e-06,
+      "loss": 0.9944,
+      "step": 17963
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2883623160710875e-06,
+      "loss": 0.6848,
+      "step": 17964
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2882667685839865e-06,
+      "loss": 0.8197,
+      "step": 17965
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.288171221096885e-06,
+      "loss": 0.7931,
+      "step": 17966
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.288075673609784e-06,
+      "loss": 0.8307,
+      "step": 17967
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2879801261226836e-06,
+      "loss": 0.7516,
+      "step": 17968
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2878845786355823e-06,
+      "loss": 0.9239,
+      "step": 17969
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2877890311484813e-06,
+      "loss": 0.8945,
+      "step": 17970
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.28769348366138e-06,
+      "loss": 0.832,
+      "step": 17971
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.287597936174279e-06,
+      "loss": 0.8214,
+      "step": 17972
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2875023886871776e-06,
+      "loss": 1.1564,
+      "step": 17973
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2874068412000766e-06,
+      "loss": 0.6775,
+      "step": 17974
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.287311293712976e-06,
+      "loss": 0.5248,
+      "step": 17975
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2872157462258747e-06,
+      "loss": 0.8136,
+      "step": 17976
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2871201987387737e-06,
+      "loss": 0.963,
+      "step": 17977
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2870246512516723e-06,
+      "loss": 0.6656,
+      "step": 17978
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2869291037645714e-06,
+      "loss": 0.8515,
+      "step": 17979
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.28683355627747e-06,
+      "loss": 1.1071,
+      "step": 17980
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286738008790369e-06,
+      "loss": 0.8231,
+      "step": 17981
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286642461303268e-06,
+      "loss": 0.8129,
+      "step": 17982
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286546913816167e-06,
+      "loss": 1.0142,
+      "step": 17983
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2864513663290657e-06,
+      "loss": 1.0279,
+      "step": 17984
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2863558188419648e-06,
+      "loss": 0.7046,
+      "step": 17985
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286260271354864e-06,
+      "loss": 1.5768,
+      "step": 17986
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2861647238677624e-06,
+      "loss": 0.9847,
+      "step": 17987
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2860691763806614e-06,
+      "loss": 0.9405,
+      "step": 17988
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2859736288935605e-06,
+      "loss": 0.6545,
+      "step": 17989
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2858780814064595e-06,
+      "loss": 1.3658,
+      "step": 17990
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285782533919358e-06,
+      "loss": 1.0841,
+      "step": 17991
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285686986432257e-06,
+      "loss": 0.8823,
+      "step": 17992
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285591438945156e-06,
+      "loss": 0.9301,
+      "step": 17993
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285495891458055e-06,
+      "loss": 0.8825,
+      "step": 17994
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2854003439709534e-06,
+      "loss": 1.0316,
+      "step": 17995
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2853047964838525e-06,
+      "loss": 1.0847,
+      "step": 17996
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.285209248996752e-06,
+      "loss": 1.1046,
+      "step": 17997
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2851137015096506e-06,
+      "loss": 1.0152,
+      "step": 17998
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2850181540225496e-06,
+      "loss": 0.654,
+      "step": 17999
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.284922606535448e-06,
+      "loss": 1.1547,
+      "step": 18000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2848270590483472e-06,
+      "loss": 0.967,
+      "step": 18001
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.284731511561246e-06,
+      "loss": 0.7907,
+      "step": 18002
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.284635964074145e-06,
+      "loss": 0.8233,
+      "step": 18003
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2845404165870444e-06,
+      "loss": 0.5737,
+      "step": 18004
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.284444869099943e-06,
+      "loss": 0.7952,
+      "step": 18005
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.284349321612842e-06,
+      "loss": 1.1289,
+      "step": 18006
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2842537741257406e-06,
+      "loss": 0.9615,
+      "step": 18007
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2841582266386397e-06,
+      "loss": 0.9403,
+      "step": 18008
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2840626791515383e-06,
+      "loss": 0.5048,
+      "step": 18009
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2839671316644373e-06,
+      "loss": 1.1775,
+      "step": 18010
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283871584177337e-06,
+      "loss": 0.6458,
+      "step": 18011
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2837760366902354e-06,
+      "loss": 0.9855,
+      "step": 18012
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2836804892031344e-06,
+      "loss": 0.8575,
+      "step": 18013
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283584941716033e-06,
+      "loss": 1.0001,
+      "step": 18014
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283489394228932e-06,
+      "loss": 0.6443,
+      "step": 18015
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2833938467418307e-06,
+      "loss": 0.9826,
+      "step": 18016
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2832982992547297e-06,
+      "loss": 0.8406,
+      "step": 18017
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283202751767629e-06,
+      "loss": 1.2277,
+      "step": 18018
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283107204280528e-06,
+      "loss": 0.5817,
+      "step": 18019
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.283011656793427e-06,
+      "loss": 1.1141,
+      "step": 18020
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2829161093063255e-06,
+      "loss": 0.845,
+      "step": 18021
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2828205618192245e-06,
+      "loss": 1.1631,
+      "step": 18022
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.282725014332123e-06,
+      "loss": 1.2461,
+      "step": 18023
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.282629466845022e-06,
+      "loss": 0.7179,
+      "step": 18024
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2825339193579216e-06,
+      "loss": 0.9566,
+      "step": 18025
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2824383718708202e-06,
+      "loss": 0.6354,
+      "step": 18026
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2823428243837193e-06,
+      "loss": 0.901,
+      "step": 18027
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.282247276896618e-06,
+      "loss": 0.6848,
+      "step": 18028
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.282151729409517e-06,
+      "loss": 1.1989,
+      "step": 18029
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2820561819224156e-06,
+      "loss": 0.7132,
+      "step": 18030
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2819606344353146e-06,
+      "loss": 0.8354,
+      "step": 18031
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281865086948213e-06,
+      "loss": 0.5992,
+      "step": 18032
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2817695394611127e-06,
+      "loss": 1.2458,
+      "step": 18033
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2816739919740113e-06,
+      "loss": 1.433,
+      "step": 18034
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2815784444869103e-06,
+      "loss": 1.0167,
+      "step": 18035
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2814828969998094e-06,
+      "loss": 1.033,
+      "step": 18036
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281387349512708e-06,
+      "loss": 0.7306,
+      "step": 18037
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281291802025607e-06,
+      "loss": 0.9862,
+      "step": 18038
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2811962545385056e-06,
+      "loss": 1.0194,
+      "step": 18039
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281100707051405e-06,
+      "loss": 0.6755,
+      "step": 18040
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2810051595643037e-06,
+      "loss": 0.8226,
+      "step": 18041
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2809096120772027e-06,
+      "loss": 0.6807,
+      "step": 18042
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2808140645901014e-06,
+      "loss": 0.5751,
+      "step": 18043
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2807185171030004e-06,
+      "loss": 0.824,
+      "step": 18044
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280622969615899e-06,
+      "loss": 1.2142,
+      "step": 18045
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280527422128798e-06,
+      "loss": 0.8418,
+      "step": 18046
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2804318746416975e-06,
+      "loss": 0.6632,
+      "step": 18047
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280336327154596e-06,
+      "loss": 0.8864,
+      "step": 18048
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280240779667495e-06,
+      "loss": 0.9598,
+      "step": 18049
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2801452321803938e-06,
+      "loss": 0.8283,
+      "step": 18050
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.280049684693293e-06,
+      "loss": 0.9258,
+      "step": 18051
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2799541372061914e-06,
+      "loss": 0.8282,
+      "step": 18052
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2798585897190905e-06,
+      "loss": 1.7902,
+      "step": 18053
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27976304223199e-06,
+      "loss": 0.9499,
+      "step": 18054
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2796674947448886e-06,
+      "loss": 0.6403,
+      "step": 18055
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2795719472577876e-06,
+      "loss": 0.9572,
+      "step": 18056
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.279476399770686e-06,
+      "loss": 0.72,
+      "step": 18057
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2793808522835852e-06,
+      "loss": 1.5224,
+      "step": 18058
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.279285304796484e-06,
+      "loss": 1.0255,
+      "step": 18059
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.279189757309383e-06,
+      "loss": 1.1709,
+      "step": 18060
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2790942098222824e-06,
+      "loss": 0.5567,
+      "step": 18061
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.278998662335181e-06,
+      "loss": 0.5882,
+      "step": 18062
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27890311484808e-06,
+      "loss": 0.738,
+      "step": 18063
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2788075673609786e-06,
+      "loss": 0.751,
+      "step": 18064
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2787120198738777e-06,
+      "loss": 0.8346,
+      "step": 18065
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2786164723867763e-06,
+      "loss": 0.6355,
+      "step": 18066
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2785209248996753e-06,
+      "loss": 0.8581,
+      "step": 18067
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.278425377412574e-06,
+      "loss": 1.1267,
+      "step": 18068
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2783298299254734e-06,
+      "loss": 0.908,
+      "step": 18069
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2782342824383724e-06,
+      "loss": 0.712,
+      "step": 18070
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.278138734951271e-06,
+      "loss": 0.5029,
+      "step": 18071
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27804318746417e-06,
+      "loss": 0.9571,
+      "step": 18072
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2779476399770687e-06,
+      "loss": 1.2376,
+      "step": 18073
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2778520924899677e-06,
+      "loss": 0.9777,
+      "step": 18074
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2777565450028664e-06,
+      "loss": 0.7592,
+      "step": 18075
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.277660997515766e-06,
+      "loss": 0.7938,
+      "step": 18076
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.277565450028665e-06,
+      "loss": 0.9182,
+      "step": 18077
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2774699025415635e-06,
+      "loss": 0.9126,
+      "step": 18078
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2773743550544625e-06,
+      "loss": 1.2777,
+      "step": 18079
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.277278807567361e-06,
+      "loss": 0.9499,
+      "step": 18080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.27718326008026e-06,
+      "loss": 0.5899,
+      "step": 18081
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2770877125931588e-06,
+      "loss": 1.3427,
+      "step": 18082
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2769921651060582e-06,
+      "loss": 0.8582,
+      "step": 18083
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276896617618957e-06,
+      "loss": 0.922,
+      "step": 18084
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276801070131856e-06,
+      "loss": 0.8494,
+      "step": 18085
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2767055226447545e-06,
+      "loss": 0.9739,
+      "step": 18086
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2766099751576536e-06,
+      "loss": 0.756,
+      "step": 18087
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2765144276705526e-06,
+      "loss": 0.8888,
+      "step": 18088
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276418880183451e-06,
+      "loss": 0.4717,
+      "step": 18089
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2763233326963507e-06,
+      "loss": 1.1072,
+      "step": 18090
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2762277852092493e-06,
+      "loss": 1.3545,
+      "step": 18091
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2761322377221483e-06,
+      "loss": 1.14,
+      "step": 18092
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276036690235047e-06,
+      "loss": 0.8823,
+      "step": 18093
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.275941142747946e-06,
+      "loss": 0.5913,
+      "step": 18094
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2758455952608446e-06,
+      "loss": 1.1964,
+      "step": 18095
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2757500477737436e-06,
+      "loss": 0.5384,
+      "step": 18096
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.275654500286643e-06,
+      "loss": 0.9284,
+      "step": 18097
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2755589527995417e-06,
+      "loss": 0.8313,
+      "step": 18098
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2754634053124407e-06,
+      "loss": 0.5968,
+      "step": 18099
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2753678578253394e-06,
+      "loss": 1.0957,
+      "step": 18100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2752723103382384e-06,
+      "loss": 0.9808,
+      "step": 18101
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275176762851137e-06,
+      "loss": 1.321,
+      "step": 18102
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.275081215364036e-06,
+      "loss": 1.0123,
+      "step": 18103
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2749856678769347e-06,
+      "loss": 1.2601,
+      "step": 18104
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274890120389834e-06,
+      "loss": 1.1199,
+      "step": 18105
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274794572902733e-06,
+      "loss": 1.0761,
+      "step": 18106
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2746990254156318e-06,
+      "loss": 0.9695,
+      "step": 18107
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274603477928531e-06,
+      "loss": 0.7472,
+      "step": 18108
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2745079304414294e-06,
+      "loss": 0.5289,
+      "step": 18109
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2744123829543285e-06,
+      "loss": 0.9057,
+      "step": 18110
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274316835467227e-06,
+      "loss": 1.5247,
+      "step": 18111
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2742212879801266e-06,
+      "loss": 0.6838,
+      "step": 18112
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2741257404930256e-06,
+      "loss": 0.7976,
+      "step": 18113
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.274030193005924e-06,
+      "loss": 1.1862,
+      "step": 18114
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2739346455188232e-06,
+      "loss": 0.5851,
+      "step": 18115
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.273839098031722e-06,
+      "loss": 0.6316,
+      "step": 18116
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.273743550544621e-06,
+      "loss": 0.7541,
+      "step": 18117
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2736480030575195e-06,
+      "loss": 0.8372,
+      "step": 18118
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.273552455570419e-06,
+      "loss": 0.998,
+      "step": 18119
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.273456908083318e-06,
+      "loss": 0.7838,
+      "step": 18120
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2733613605962166e-06,
+      "loss": 0.5004,
+      "step": 18121
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2732658131091157e-06,
+      "loss": 0.4249,
+      "step": 18122
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2731702656220143e-06,
+      "loss": 0.7213,
+      "step": 18123
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2730747181349133e-06,
+      "loss": 1.1202,
+      "step": 18124
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272979170647812e-06,
+      "loss": 1.1612,
+      "step": 18125
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2728836231607114e-06,
+      "loss": 0.7853,
+      "step": 18126
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2727880756736104e-06,
+      "loss": 1.2536,
+      "step": 18127
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272692528186509e-06,
+      "loss": 0.7253,
+      "step": 18128
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272596980699408e-06,
+      "loss": 0.9601,
+      "step": 18129
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2725014332123067e-06,
+      "loss": 1.1522,
+      "step": 18130
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2724058857252057e-06,
+      "loss": 1.1464,
+      "step": 18131
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2723103382381044e-06,
+      "loss": 1.0449,
+      "step": 18132
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272214790751004e-06,
+      "loss": 0.9889,
+      "step": 18133
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2721192432639024e-06,
+      "loss": 1.3127,
+      "step": 18134
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2720236957768015e-06,
+      "loss": 0.898,
+      "step": 18135
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2719281482897e-06,
+      "loss": 0.7078,
+      "step": 18136
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271832600802599e-06,
+      "loss": 0.7675,
+      "step": 18137
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2717370533154977e-06,
+      "loss": 0.739,
+      "step": 18138
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2716415058283968e-06,
+      "loss": 0.713,
+      "step": 18139
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271545958341296e-06,
+      "loss": 0.778,
+      "step": 18140
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271450410854195e-06,
+      "loss": 0.9589,
+      "step": 18141
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271354863367094e-06,
+      "loss": 1.0386,
+      "step": 18142
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2712593158799925e-06,
+      "loss": 1.4671,
+      "step": 18143
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2711637683928916e-06,
+      "loss": 0.4967,
+      "step": 18144
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.27106822090579e-06,
+      "loss": 1.2726,
+      "step": 18145
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.270972673418689e-06,
+      "loss": 1.003,
+      "step": 18146
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.270877125931588e-06,
+      "loss": 1.0458,
+      "step": 18147
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2707815784444873e-06,
+      "loss": 0.6477,
+      "step": 18148
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2706860309573863e-06,
+      "loss": 1.1315,
+      "step": 18149
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270590483470285e-06,
+      "loss": 1.1692,
+      "step": 18150
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.270494935983184e-06,
+      "loss": 0.7162,
+      "step": 18151
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2703993884960826e-06,
+      "loss": 0.8075,
+      "step": 18152
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2703038410089816e-06,
+      "loss": 0.6022,
+      "step": 18153
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2702082935218802e-06,
+      "loss": 1.0764,
+      "step": 18154
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2701127460347797e-06,
+      "loss": 0.8036,
+      "step": 18155
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2700171985476787e-06,
+      "loss": 0.7812,
+      "step": 18156
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2699216510605774e-06,
+      "loss": 0.9158,
+      "step": 18157
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2698261035734764e-06,
+      "loss": 0.8504,
+      "step": 18158
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269730556086375e-06,
+      "loss": 1.0836,
+      "step": 18159
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269635008599274e-06,
+      "loss": 1.2344,
+      "step": 18160
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2695394611121727e-06,
+      "loss": 0.9854,
+      "step": 18161
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269443913625072e-06,
+      "loss": 1.1784,
+      "step": 18162
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269348366137971e-06,
+      "loss": 0.81,
+      "step": 18163
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2692528186508698e-06,
+      "loss": 1.1218,
+      "step": 18164
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.269157271163769e-06,
+      "loss": 1.0797,
+      "step": 18165
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2690617236766674e-06,
+      "loss": 0.9381,
+      "step": 18166
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2689661761895665e-06,
+      "loss": 0.8318,
+      "step": 18167
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.268870628702465e-06,
+      "loss": 1.0187,
+      "step": 18168
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2687750812153646e-06,
+      "loss": 0.584,
+      "step": 18169
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2686795337282636e-06,
+      "loss": 0.7998,
+      "step": 18170
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.268583986241162e-06,
+      "loss": 1.0261,
+      "step": 18171
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2684884387540612e-06,
+      "loss": 0.8756,
+      "step": 18172
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.26839289126696e-06,
+      "loss": 0.7471,
+      "step": 18173
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.268297343779859e-06,
+      "loss": 0.4436,
+      "step": 18174
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2682017962927575e-06,
+      "loss": 1.1115,
+      "step": 18175
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2681062488056566e-06,
+      "loss": 0.667,
+      "step": 18176
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2680107013185556e-06,
+      "loss": 0.8822,
+      "step": 18177
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2679151538314546e-06,
+      "loss": 1.3687,
+      "step": 18178
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2678196063443537e-06,
+      "loss": 0.6712,
+      "step": 18179
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2677240588572523e-06,
+      "loss": 0.9624,
+      "step": 18180
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2676285113701513e-06,
+      "loss": 0.6494,
+      "step": 18181
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.26753296388305e-06,
+      "loss": 0.6929,
+      "step": 18182
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.267437416395949e-06,
+      "loss": 0.8907,
+      "step": 18183
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.267341868908848e-06,
+      "loss": 0.9055,
+      "step": 18184
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.267246321421747e-06,
+      "loss": 0.9176,
+      "step": 18185
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2671507739346457e-06,
+      "loss": 0.7592,
+      "step": 18186
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2670552264475447e-06,
+      "loss": 1.082,
+      "step": 18187
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2669596789604433e-06,
+      "loss": 1.1175,
+      "step": 18188
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2668641314733424e-06,
+      "loss": 0.9076,
+      "step": 18189
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2667685839862414e-06,
+      "loss": 0.4504,
+      "step": 18190
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2666730364991404e-06,
+      "loss": 0.6228,
+      "step": 18191
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2665774890120395e-06,
+      "loss": 0.8573,
+      "step": 18192
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266481941524938e-06,
+      "loss": 0.7391,
+      "step": 18193
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266386394037837e-06,
+      "loss": 0.9238,
+      "step": 18194
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2662908465507357e-06,
+      "loss": 1.1505,
+      "step": 18195
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2661952990636348e-06,
+      "loss": 0.6082,
+      "step": 18196
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2660997515765334e-06,
+      "loss": 0.5512,
+      "step": 18197
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.266004204089433e-06,
+      "loss": 0.6709,
+      "step": 18198
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.265908656602332e-06,
+      "loss": 0.8813,
+      "step": 18199
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2658131091152305e-06,
+      "loss": 0.7125,
+      "step": 18200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2657175616281296e-06,
+      "loss": 1.2101,
+      "step": 18201
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.265622014141028e-06,
+      "loss": 0.8988,
+      "step": 18202
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265526466653927e-06,
+      "loss": 0.9108,
+      "step": 18203
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265430919166826e-06,
+      "loss": 1.1555,
+      "step": 18204
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2653353716797253e-06,
+      "loss": 0.888,
+      "step": 18205
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2652398241926243e-06,
+      "loss": 0.7964,
+      "step": 18206
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265144276705523e-06,
+      "loss": 0.5657,
+      "step": 18207
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.265048729218422e-06,
+      "loss": 1.1799,
+      "step": 18208
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2649531817313206e-06,
+      "loss": 0.9871,
+      "step": 18209
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2648576342442196e-06,
+      "loss": 0.5582,
+      "step": 18210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2647620867571182e-06,
+      "loss": 0.7733,
+      "step": 18211
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2646665392700173e-06,
+      "loss": 0.8466,
+      "step": 18212
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2645709917829167e-06,
+      "loss": 0.5873,
+      "step": 18213
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2644754442958154e-06,
+      "loss": 1.2779,
+      "step": 18214
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2643798968087144e-06,
+      "loss": 1.3271,
+      "step": 18215
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.264284349321613e-06,
+      "loss": 0.8349,
+      "step": 18216
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.264188801834512e-06,
+      "loss": 0.7684,
+      "step": 18217
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2640932543474107e-06,
+      "loss": 0.7517,
+      "step": 18218
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2639977068603097e-06,
+      "loss": 0.4115,
+      "step": 18219
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.263902159373209e-06,
+      "loss": 0.9583,
+      "step": 18220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2638066118861078e-06,
+      "loss": 1.3594,
+      "step": 18221
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.263711064399007e-06,
+      "loss": 1.003,
+      "step": 18222
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2636155169119054e-06,
+      "loss": 1.0256,
+      "step": 18223
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2635199694248045e-06,
+      "loss": 0.6505,
+      "step": 18224
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.263424421937703e-06,
+      "loss": 0.9563,
+      "step": 18225
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.263328874450602e-06,
+      "loss": 0.7647,
+      "step": 18226
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.263233326963501e-06,
+      "loss": 0.9732,
+      "step": 18227
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2631377794764e-06,
+      "loss": 0.7611,
+      "step": 18228
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2630422319892992e-06,
+      "loss": 0.9314,
+      "step": 18229
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262946684502198e-06,
+      "loss": 0.9537,
+      "step": 18230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262851137015097e-06,
+      "loss": 0.8802,
+      "step": 18231
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2627555895279955e-06,
+      "loss": 0.7038,
+      "step": 18232
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2626600420408946e-06,
+      "loss": 0.9808,
+      "step": 18233
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2625644945537936e-06,
+      "loss": 0.7575,
+      "step": 18234
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2624689470666926e-06,
+      "loss": 1.068,
+      "step": 18235
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2623733995795912e-06,
+      "loss": 0.6486,
+      "step": 18236
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2622778520924903e-06,
+      "loss": 0.7297,
+      "step": 18237
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262182304605389e-06,
+      "loss": 0.5633,
+      "step": 18238
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262086757118288e-06,
+      "loss": 0.5929,
+      "step": 18239
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2619912096311866e-06,
+      "loss": 0.9199,
+      "step": 18240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261895662144086e-06,
+      "loss": 0.4722,
+      "step": 18241
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261800114656985e-06,
+      "loss": 1.0773,
+      "step": 18242
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2617045671698837e-06,
+      "loss": 0.5226,
+      "step": 18243
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2616090196827827e-06,
+      "loss": 1.0067,
+      "step": 18244
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2615134721956813e-06,
+      "loss": 0.9461,
+      "step": 18245
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2614179247085804e-06,
+      "loss": 0.6712,
+      "step": 18246
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261322377221479e-06,
+      "loss": 1.3906,
+      "step": 18247
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261226829734378e-06,
+      "loss": 0.8244,
+      "step": 18248
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2611312822472775e-06,
+      "loss": 0.7207,
+      "step": 18249
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.261035734760176e-06,
+      "loss": 0.8143,
+      "step": 18250
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.260940187273075e-06,
+      "loss": 0.8594,
+      "step": 18251
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2608446397859737e-06,
+      "loss": 0.6364,
+      "step": 18252
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2607490922988728e-06,
+      "loss": 0.9288,
+      "step": 18253
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2606535448117714e-06,
+      "loss": 1.2227,
+      "step": 18254
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2605579973246704e-06,
+      "loss": 0.922,
+      "step": 18255
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.26046244983757e-06,
+      "loss": 0.8077,
+      "step": 18256
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2603669023504685e-06,
+      "loss": 0.7641,
+      "step": 18257
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2602713548633676e-06,
+      "loss": 1.1614,
+      "step": 18258
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260175807376266e-06,
+      "loss": 0.4341,
+      "step": 18259
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.260080259889165e-06,
+      "loss": 1.663,
+      "step": 18260
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.259984712402064e-06,
+      "loss": 0.8187,
+      "step": 18261
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.259889164914963e-06,
+      "loss": 0.8233,
+      "step": 18262
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2597936174278623e-06,
+      "loss": 0.6017,
+      "step": 18263
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.259698069940761e-06,
+      "loss": 1.1371,
+      "step": 18264
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25960252245366e-06,
+      "loss": 0.6182,
+      "step": 18265
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2595069749665586e-06,
+      "loss": 1.3185,
+      "step": 18266
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2594114274794576e-06,
+      "loss": 0.6518,
+      "step": 18267
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2593158799923562e-06,
+      "loss": 1.1321,
+      "step": 18268
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2592203325052553e-06,
+      "loss": 0.7342,
+      "step": 18269
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2591247850181547e-06,
+      "loss": 1.0631,
+      "step": 18270
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2590292375310534e-06,
+      "loss": 1.1596,
+      "step": 18271
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2589336900439524e-06,
+      "loss": 1.1936,
+      "step": 18272
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.258838142556851e-06,
+      "loss": 0.9944,
+      "step": 18273
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25874259506975e-06,
+      "loss": 1.0222,
+      "step": 18274
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2586470475826487e-06,
+      "loss": 1.0814,
+      "step": 18275
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2585515000955477e-06,
+      "loss": 0.7649,
+      "step": 18276
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2584559526084467e-06,
+      "loss": 0.7802,
+      "step": 18277
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2583604051213458e-06,
+      "loss": 0.8062,
+      "step": 18278
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2582648576342444e-06,
+      "loss": 0.8393,
+      "step": 18279
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2581693101471434e-06,
+      "loss": 0.5608,
+      "step": 18280
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2580737626600425e-06,
+      "loss": 0.7669,
+      "step": 18281
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257978215172941e-06,
+      "loss": 1.1211,
+      "step": 18282
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.25788266768584e-06,
+      "loss": 0.809,
+      "step": 18283
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2577871201987387e-06,
+      "loss": 0.9309,
+      "step": 18284
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257691572711638e-06,
+      "loss": 1.0061,
+      "step": 18285
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257596025224537e-06,
+      "loss": 0.8334,
+      "step": 18286
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257500477737436e-06,
+      "loss": 0.8992,
+      "step": 18287
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2574049302503345e-06,
+      "loss": 0.7686,
+      "step": 18288
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2573093827632335e-06,
+      "loss": 0.6424,
+      "step": 18289
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257213835276132e-06,
+      "loss": 0.9198,
+      "step": 18290
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257118287789031e-06,
+      "loss": 0.9125,
+      "step": 18291
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2570227403019306e-06,
+      "loss": 1.0153,
+      "step": 18292
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2569271928148292e-06,
+      "loss": 1.1788,
+      "step": 18293
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2568316453277283e-06,
+      "loss": 1.0881,
+      "step": 18294
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256736097840627e-06,
+      "loss": 0.8005,
+      "step": 18295
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256640550353526e-06,
+      "loss": 0.8653,
+      "step": 18296
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2565450028664246e-06,
+      "loss": 0.7261,
+      "step": 18297
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2564494553793236e-06,
+      "loss": 0.8705,
+      "step": 18298
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256353907892223e-06,
+      "loss": 1.3845,
+      "step": 18299
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2562583604051217e-06,
+      "loss": 0.6371,
+      "step": 18300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2561628129180207e-06,
+      "loss": 0.8501,
+      "step": 18301
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2560672654309193e-06,
+      "loss": 0.9063,
+      "step": 18302
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2559717179438184e-06,
+      "loss": 0.6002,
+      "step": 18303
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.255876170456717e-06,
+      "loss": 0.6199,
+      "step": 18304
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.255780622969616e-06,
+      "loss": 0.8771,
+      "step": 18305
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2556850754825155e-06,
+      "loss": 1.0335,
+      "step": 18306
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255589527995414e-06,
+      "loss": 0.6113,
+      "step": 18307
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255493980508313e-06,
+      "loss": 0.8204,
+      "step": 18308
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2553984330212117e-06,
+      "loss": 1.5545,
+      "step": 18309
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2553028855341108e-06,
+      "loss": 0.7044,
+      "step": 18310
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2552073380470094e-06,
+      "loss": 0.8381,
+      "step": 18311
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2551117905599084e-06,
+      "loss": 1.1037,
+      "step": 18312
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.255016243072808e-06,
+      "loss": 1.296,
+      "step": 18313
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2549206955857065e-06,
+      "loss": 1.1394,
+      "step": 18314
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2548251480986056e-06,
+      "loss": 1.1418,
+      "step": 18315
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254729600611504e-06,
+      "loss": 0.6584,
+      "step": 18316
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254634053124403e-06,
+      "loss": 0.7016,
+      "step": 18317
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254538505637302e-06,
+      "loss": 0.6235,
+      "step": 18318
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254442958150201e-06,
+      "loss": 0.734,
+      "step": 18319
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2543474106630995e-06,
+      "loss": 1.1359,
+      "step": 18320
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254251863175999e-06,
+      "loss": 0.6267,
+      "step": 18321
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.254156315688898e-06,
+      "loss": 0.7515,
+      "step": 18322
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2540607682017966e-06,
+      "loss": 1.0455,
+      "step": 18323
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2539652207146956e-06,
+      "loss": 0.6486,
+      "step": 18324
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2538696732275942e-06,
+      "loss": 0.7899,
+      "step": 18325
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2537741257404933e-06,
+      "loss": 1.2375,
+      "step": 18326
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.253678578253392e-06,
+      "loss": 0.9691,
+      "step": 18327
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2535830307662914e-06,
+      "loss": 0.8289,
+      "step": 18328
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25348748327919e-06,
+      "loss": 0.7205,
+      "step": 18329
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.253391935792089e-06,
+      "loss": 0.8058,
+      "step": 18330
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.253296388304988e-06,
+      "loss": 0.4988,
+      "step": 18331
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2532008408178867e-06,
+      "loss": 1.0634,
+      "step": 18332
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2531052933307857e-06,
+      "loss": 0.9672,
+      "step": 18333
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2530097458436843e-06,
+      "loss": 0.9905,
+      "step": 18334
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2529141983565838e-06,
+      "loss": 0.7885,
+      "step": 18335
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2528186508694824e-06,
+      "loss": 0.8379,
+      "step": 18336
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2527231033823814e-06,
+      "loss": 1.097,
+      "step": 18337
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25262755589528e-06,
+      "loss": 0.9468,
+      "step": 18338
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.252532008408179e-06,
+      "loss": 0.7828,
+      "step": 18339
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2524364609210777e-06,
+      "loss": 1.0203,
+      "step": 18340
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2523409134339767e-06,
+      "loss": 0.9193,
+      "step": 18341
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.252245365946876e-06,
+      "loss": 0.7961,
+      "step": 18342
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.252149818459775e-06,
+      "loss": 0.9952,
+      "step": 18343
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.252054270972674e-06,
+      "loss": 0.8108,
+      "step": 18344
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2519587234855725e-06,
+      "loss": 0.7782,
+      "step": 18345
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2518631759984715e-06,
+      "loss": 1.0593,
+      "step": 18346
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25176762851137e-06,
+      "loss": 0.8361,
+      "step": 18347
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.251672081024269e-06,
+      "loss": 0.8339,
+      "step": 18348
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2515765335371686e-06,
+      "loss": 0.8545,
+      "step": 18349
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2514809860500672e-06,
+      "loss": 1.1231,
+      "step": 18350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2513854385629663e-06,
+      "loss": 0.8991,
+      "step": 18351
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.251289891075865e-06,
+      "loss": 0.8254,
+      "step": 18352
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.251194343588764e-06,
+      "loss": 0.6916,
+      "step": 18353
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2510987961016626e-06,
+      "loss": 1.3222,
+      "step": 18354
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2510032486145616e-06,
+      "loss": 0.5783,
+      "step": 18355
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.25090770112746e-06,
+      "loss": 1.0602,
+      "step": 18356
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2508121536403597e-06,
+      "loss": 0.7094,
+      "step": 18357
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2507166061532587e-06,
+      "loss": 0.6665,
+      "step": 18358
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2506210586661573e-06,
+      "loss": 1.1962,
+      "step": 18359
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2505255111790564e-06,
+      "loss": 0.935,
+      "step": 18360
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250429963691955e-06,
+      "loss": 1.0978,
+      "step": 18361
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250334416204854e-06,
+      "loss": 0.7351,
+      "step": 18362
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2502388687177526e-06,
+      "loss": 0.6431,
+      "step": 18363
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250143321230652e-06,
+      "loss": 0.9393,
+      "step": 18364
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.250047773743551e-06,
+      "loss": 0.8954,
+      "step": 18365
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2499522262564497e-06,
+      "loss": 1.3709,
+      "step": 18366
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2498566787693488e-06,
+      "loss": 0.9589,
+      "step": 18367
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2497611312822474e-06,
+      "loss": 0.8639,
+      "step": 18368
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2496655837951464e-06,
+      "loss": 1.3446,
+      "step": 18369
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.249570036308045e-06,
+      "loss": 1.278,
+      "step": 18370
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2494744888209445e-06,
+      "loss": 0.5322,
+      "step": 18371
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2493789413338436e-06,
+      "loss": 0.5208,
+      "step": 18372
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.249283393846742e-06,
+      "loss": 0.7867,
+      "step": 18373
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.249187846359641e-06,
+      "loss": 0.8607,
+      "step": 18374
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.24909229887254e-06,
+      "loss": 1.1673,
+      "step": 18375
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.248996751385439e-06,
+      "loss": 0.6098,
+      "step": 18376
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2489012038983375e-06,
+      "loss": 0.8805,
+      "step": 18377
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.248805656411237e-06,
+      "loss": 0.8549,
+      "step": 18378
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2487101089241356e-06,
+      "loss": 0.9365,
+      "step": 18379
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2486145614370346e-06,
+      "loss": 0.737,
+      "step": 18380
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.248519013949933e-06,
+      "loss": 0.8167,
+      "step": 18381
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2484234664628322e-06,
+      "loss": 1.2712,
+      "step": 18382
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2483279189757313e-06,
+      "loss": 1.1821,
+      "step": 18383
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.24823237148863e-06,
+      "loss": 0.7161,
+      "step": 18384
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2481368240015294e-06,
+      "loss": 1.051,
+      "step": 18385
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.248041276514428e-06,
+      "loss": 1.0025,
+      "step": 18386
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.247945729027327e-06,
+      "loss": 0.6529,
+      "step": 18387
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2478501815402256e-06,
+      "loss": 0.8683,
+      "step": 18388
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2477546340531247e-06,
+      "loss": 0.7233,
+      "step": 18389
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2476590865660233e-06,
+      "loss": 0.5667,
+      "step": 18390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2475635390789223e-06,
+      "loss": 0.9092,
+      "step": 18391
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.247467991591821e-06,
+      "loss": 0.6674,
+      "step": 18392
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2473724441047204e-06,
+      "loss": 0.9123,
+      "step": 18393
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2472768966176194e-06,
+      "loss": 0.8115,
+      "step": 18394
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.247181349130518e-06,
+      "loss": 0.7666,
+      "step": 18395
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.247085801643417e-06,
+      "loss": 1.0326,
+      "step": 18396
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2469902541563157e-06,
+      "loss": 1.059,
+      "step": 18397
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2468947066692147e-06,
+      "loss": 0.6986,
+      "step": 18398
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2467991591821134e-06,
+      "loss": 0.8074,
+      "step": 18399
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.246703611695013e-06,
+      "loss": 0.5541,
+      "step": 18400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.246608064207912e-06,
+      "loss": 0.9687,
+      "step": 18401
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2465125167208105e-06,
+      "loss": 1.0609,
+      "step": 18402
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2464169692337095e-06,
+      "loss": 1.1862,
+      "step": 18403
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.246321421746608e-06,
+      "loss": 0.8912,
+      "step": 18404
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.246225874259507e-06,
+      "loss": 1.4444,
+      "step": 18405
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2461303267724058e-06,
+      "loss": 0.8176,
+      "step": 18406
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2460347792853052e-06,
+      "loss": 0.7795,
+      "step": 18407
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2459392317982043e-06,
+      "loss": 1.343,
+      "step": 18408
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.245843684311103e-06,
+      "loss": 0.9406,
+      "step": 18409
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.245748136824002e-06,
+      "loss": 0.6054,
+      "step": 18410
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2456525893369005e-06,
+      "loss": 0.8024,
+      "step": 18411
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2455570418497996e-06,
+      "loss": 1.0139,
+      "step": 18412
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.245461494362698e-06,
+      "loss": 0.6971,
+      "step": 18413
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2453659468755977e-06,
+      "loss": 1.1264,
+      "step": 18414
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2452703993884967e-06,
+      "loss": 0.6111,
+      "step": 18415
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2451748519013953e-06,
+      "loss": 1.1064,
+      "step": 18416
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2450793044142944e-06,
+      "loss": 0.565,
+      "step": 18417
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244983756927193e-06,
+      "loss": 0.7362,
+      "step": 18418
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244888209440092e-06,
+      "loss": 0.7491,
+      "step": 18419
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2447926619529906e-06,
+      "loss": 0.7291,
+      "step": 18420
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24469711446589e-06,
+      "loss": 0.7618,
+      "step": 18421
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244601566978789e-06,
+      "loss": 0.6077,
+      "step": 18422
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2445060194916877e-06,
+      "loss": 0.9042,
+      "step": 18423
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2444104720045868e-06,
+      "loss": 1.0473,
+      "step": 18424
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2443149245174854e-06,
+      "loss": 0.9069,
+      "step": 18425
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2442193770303844e-06,
+      "loss": 0.9475,
+      "step": 18426
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244123829543283e-06,
+      "loss": 0.6299,
+      "step": 18427
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.244028282056182e-06,
+      "loss": 0.8114,
+      "step": 18428
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.243932734569081e-06,
+      "loss": 0.7414,
+      "step": 18429
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24383718708198e-06,
+      "loss": 0.8263,
+      "step": 18430
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2437416395948788e-06,
+      "loss": 0.9398,
+      "step": 18431
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.243646092107778e-06,
+      "loss": 0.8323,
+      "step": 18432
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2435505446206764e-06,
+      "loss": 1.2381,
+      "step": 18433
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2434549971335755e-06,
+      "loss": 1.0753,
+      "step": 18434
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2433594496464745e-06,
+      "loss": 0.5034,
+      "step": 18435
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2432639021593735e-06,
+      "loss": 1.2152,
+      "step": 18436
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2431683546722726e-06,
+      "loss": 1.1044,
+      "step": 18437
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.243072807185171e-06,
+      "loss": 0.7211,
+      "step": 18438
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2429772596980702e-06,
+      "loss": 1.2948,
+      "step": 18439
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242881712210969e-06,
+      "loss": 1.0854,
+      "step": 18440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242786164723868e-06,
+      "loss": 0.9123,
+      "step": 18441
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2426906172367665e-06,
+      "loss": 0.5972,
+      "step": 18442
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242595069749666e-06,
+      "loss": 0.817,
+      "step": 18443
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242499522262565e-06,
+      "loss": 1.2534,
+      "step": 18444
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2424039747754636e-06,
+      "loss": 0.8286,
+      "step": 18445
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2423084272883627e-06,
+      "loss": 1.0405,
+      "step": 18446
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2422128798012613e-06,
+      "loss": 1.0218,
+      "step": 18447
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2421173323141603e-06,
+      "loss": 1.0897,
+      "step": 18448
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242021784827059e-06,
+      "loss": 0.5394,
+      "step": 18449
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2419262373399584e-06,
+      "loss": 1.0662,
+      "step": 18450
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2418306898528574e-06,
+      "loss": 0.9495,
+      "step": 18451
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.241735142365756e-06,
+      "loss": 0.9407,
+      "step": 18452
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.241639594878655e-06,
+      "loss": 0.871,
+      "step": 18453
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2415440473915537e-06,
+      "loss": 1.113,
+      "step": 18454
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2414484999044527e-06,
+      "loss": 0.6019,
+      "step": 18455
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2413529524173514e-06,
+      "loss": 0.9541,
+      "step": 18456
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.241257404930251e-06,
+      "loss": 0.6123,
+      "step": 18457
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.24116185744315e-06,
+      "loss": 1.0137,
+      "step": 18458
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2410663099560485e-06,
+      "loss": 1.1503,
+      "step": 18459
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2409707624689475e-06,
+      "loss": 0.7474,
+      "step": 18460
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.240875214981846e-06,
+      "loss": 0.6956,
+      "step": 18461
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.240779667494745e-06,
+      "loss": 1.3654,
+      "step": 18462
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2406841200076438e-06,
+      "loss": 1.2279,
+      "step": 18463
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240588572520543e-06,
+      "loss": 0.6172,
+      "step": 18464
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2404930250334423e-06,
+      "loss": 0.5826,
+      "step": 18465
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240397477546341e-06,
+      "loss": 0.8283,
+      "step": 18466
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.24030193005924e-06,
+      "loss": 0.6857,
+      "step": 18467
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2402063825721385e-06,
+      "loss": 0.9117,
+      "step": 18468
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2401108350850376e-06,
+      "loss": 0.9236,
+      "step": 18469
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.240015287597936e-06,
+      "loss": 1.0126,
+      "step": 18470
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2399197401108352e-06,
+      "loss": 0.7138,
+      "step": 18471
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2398241926237343e-06,
+      "loss": 1.0977,
+      "step": 18472
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2397286451366333e-06,
+      "loss": 1.0437,
+      "step": 18473
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2396330976495324e-06,
+      "loss": 0.7402,
+      "step": 18474
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.239537550162431e-06,
+      "loss": 0.7315,
+      "step": 18475
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.23944200267533e-06,
+      "loss": 0.8372,
+      "step": 18476
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2393464551882286e-06,
+      "loss": 1.1925,
+      "step": 18477
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2392509077011277e-06,
+      "loss": 0.6983,
+      "step": 18478
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2391553602140267e-06,
+      "loss": 0.8977,
+      "step": 18479
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2390598127269257e-06,
+      "loss": 0.8424,
+      "step": 18480
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2389642652398244e-06,
+      "loss": 0.4828,
+      "step": 18481
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2388687177527234e-06,
+      "loss": 0.8299,
+      "step": 18482
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.238773170265622e-06,
+      "loss": 0.7632,
+      "step": 18483
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.238677622778521e-06,
+      "loss": 0.7562,
+      "step": 18484
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.23858207529142e-06,
+      "loss": 0.9439,
+      "step": 18485
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.238486527804319e-06,
+      "loss": 0.9748,
+      "step": 18486
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.238390980317218e-06,
+      "loss": 0.8667,
+      "step": 18487
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2382954328301168e-06,
+      "loss": 0.8484,
+      "step": 18488
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.238199885343016e-06,
+      "loss": 0.9502,
+      "step": 18489
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2381043378559144e-06,
+      "loss": 0.9512,
+      "step": 18490
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2380087903688135e-06,
+      "loss": 0.8175,
+      "step": 18491
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237913242881712e-06,
+      "loss": 0.98,
+      "step": 18492
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2378176953946115e-06,
+      "loss": 0.9852,
+      "step": 18493
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2377221479075106e-06,
+      "loss": 0.921,
+      "step": 18494
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237626600420409e-06,
+      "loss": 0.7019,
+      "step": 18495
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2375310529333082e-06,
+      "loss": 0.961,
+      "step": 18496
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237435505446207e-06,
+      "loss": 0.8549,
+      "step": 18497
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237339957959106e-06,
+      "loss": 0.4627,
+      "step": 18498
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2372444104720045e-06,
+      "loss": 0.9884,
+      "step": 18499
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2371488629849035e-06,
+      "loss": 1.1538,
+      "step": 18500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237053315497803e-06,
+      "loss": 1.4381,
+      "step": 18501
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2369577680107016e-06,
+      "loss": 0.8497,
+      "step": 18502
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2368622205236007e-06,
+      "loss": 0.534,
+      "step": 18503
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2367666730364993e-06,
+      "loss": 0.8664,
+      "step": 18504
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2366711255493983e-06,
+      "loss": 0.7547,
+      "step": 18505
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.236575578062297e-06,
+      "loss": 1.0561,
+      "step": 18506
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.236480030575196e-06,
+      "loss": 0.7916,
+      "step": 18507
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2363844830880954e-06,
+      "loss": 1.2378,
+      "step": 18508
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.236288935600994e-06,
+      "loss": 0.8978,
+      "step": 18509
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.236193388113893e-06,
+      "loss": 0.8542,
+      "step": 18510
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2360978406267917e-06,
+      "loss": 1.1064,
+      "step": 18511
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2360022931396907e-06,
+      "loss": 0.8614,
+      "step": 18512
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2359067456525894e-06,
+      "loss": 1.1526,
+      "step": 18513
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2358111981654884e-06,
+      "loss": 1.1372,
+      "step": 18514
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.235715650678388e-06,
+      "loss": 1.0362,
+      "step": 18515
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2356201031912865e-06,
+      "loss": 0.6071,
+      "step": 18516
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2355245557041855e-06,
+      "loss": 0.9504,
+      "step": 18517
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.235429008217084e-06,
+      "loss": 1.2067,
+      "step": 18518
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.235333460729983e-06,
+      "loss": 0.8609,
+      "step": 18519
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2352379132428818e-06,
+      "loss": 1.0855,
+      "step": 18520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.235142365755781e-06,
+      "loss": 0.9127,
+      "step": 18521
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23504681826868e-06,
+      "loss": 0.3891,
+      "step": 18522
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.234951270781579e-06,
+      "loss": 0.7029,
+      "step": 18523
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.234855723294478e-06,
+      "loss": 0.8377,
+      "step": 18524
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2347601758073765e-06,
+      "loss": 0.947,
+      "step": 18525
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2346646283202756e-06,
+      "loss": 1.1126,
+      "step": 18526
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.234569080833174e-06,
+      "loss": 1.1723,
+      "step": 18527
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2344735333460732e-06,
+      "loss": 0.779,
+      "step": 18528
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2343779858589723e-06,
+      "loss": 0.9215,
+      "step": 18529
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2342824383718713e-06,
+      "loss": 0.9356,
+      "step": 18530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23418689088477e-06,
+      "loss": 1.0415,
+      "step": 18531
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.234091343397669e-06,
+      "loss": 1.1373,
+      "step": 18532
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2339957959105676e-06,
+      "loss": 0.7831,
+      "step": 18533
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2339002484234666e-06,
+      "loss": 0.7162,
+      "step": 18534
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2338047009363652e-06,
+      "loss": 0.6112,
+      "step": 18535
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2337091534492643e-06,
+      "loss": 0.7267,
+      "step": 18536
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2336136059621637e-06,
+      "loss": 0.5651,
+      "step": 18537
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2335180584750624e-06,
+      "loss": 0.7226,
+      "step": 18538
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2334225109879614e-06,
+      "loss": 0.5644,
+      "step": 18539
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23332696350086e-06,
+      "loss": 0.8829,
+      "step": 18540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.233231416013759e-06,
+      "loss": 0.6817,
+      "step": 18541
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2331358685266577e-06,
+      "loss": 0.4669,
+      "step": 18542
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2330403210395567e-06,
+      "loss": 0.6167,
+      "step": 18543
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.232944773552456e-06,
+      "loss": 0.8188,
+      "step": 18544
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2328492260653548e-06,
+      "loss": 0.5339,
+      "step": 18545
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.232753678578254e-06,
+      "loss": 0.7356,
+      "step": 18546
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2326581310911524e-06,
+      "loss": 0.8994,
+      "step": 18547
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2325625836040515e-06,
+      "loss": 0.7043,
+      "step": 18548
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.23246703611695e-06,
+      "loss": 1.1104,
+      "step": 18549
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.232371488629849e-06,
+      "loss": 1.2269,
+      "step": 18550
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2322759411427486e-06,
+      "loss": 1.0827,
+      "step": 18551
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.232180393655647e-06,
+      "loss": 0.7979,
+      "step": 18552
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2320848461685462e-06,
+      "loss": 0.7109,
+      "step": 18553
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231989298681445e-06,
+      "loss": 0.5434,
+      "step": 18554
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231893751194344e-06,
+      "loss": 1.0225,
+      "step": 18555
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2317982037072425e-06,
+      "loss": 0.9845,
+      "step": 18556
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2317026562201415e-06,
+      "loss": 0.5124,
+      "step": 18557
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231607108733041e-06,
+      "loss": 0.8144,
+      "step": 18558
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2315115612459396e-06,
+      "loss": 0.8455,
+      "step": 18559
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2314160137588387e-06,
+      "loss": 0.7906,
+      "step": 18560
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2313204662717373e-06,
+      "loss": 0.7088,
+      "step": 18561
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2312249187846363e-06,
+      "loss": 0.8448,
+      "step": 18562
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231129371297535e-06,
+      "loss": 0.4871,
+      "step": 18563
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231033823810434e-06,
+      "loss": 1.1181,
+      "step": 18564
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2309382763233334e-06,
+      "loss": 0.8843,
+      "step": 18565
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.230842728836232e-06,
+      "loss": 1.0713,
+      "step": 18566
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.230747181349131e-06,
+      "loss": 0.9934,
+      "step": 18567
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2306516338620297e-06,
+      "loss": 0.8638,
+      "step": 18568
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2305560863749287e-06,
+      "loss": 0.8298,
+      "step": 18569
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2304605388878274e-06,
+      "loss": 0.4484,
+      "step": 18570
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2303649914007264e-06,
+      "loss": 0.8814,
+      "step": 18571
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230269443913625e-06,
+      "loss": 0.369,
+      "step": 18572
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2301738964265245e-06,
+      "loss": 0.99,
+      "step": 18573
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.230078348939423e-06,
+      "loss": 1.0144,
+      "step": 18574
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229982801452322e-06,
+      "loss": 0.8915,
+      "step": 18575
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229887253965221e-06,
+      "loss": 0.8959,
+      "step": 18576
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2297917064781198e-06,
+      "loss": 0.8917,
+      "step": 18577
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229696158991019e-06,
+      "loss": 1.1432,
+      "step": 18578
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2296006115039174e-06,
+      "loss": 0.7924,
+      "step": 18579
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229505064016817e-06,
+      "loss": 0.8064,
+      "step": 18580
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2294095165297155e-06,
+      "loss": 0.5128,
+      "step": 18581
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2293139690426145e-06,
+      "loss": 0.8185,
+      "step": 18582
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229218421555513e-06,
+      "loss": 0.8068,
+      "step": 18583
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229122874068412e-06,
+      "loss": 0.6341,
+      "step": 18584
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.229027326581311e-06,
+      "loss": 0.5648,
+      "step": 18585
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.22893177909421e-06,
+      "loss": 1.1314,
+      "step": 18586
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2288362316071093e-06,
+      "loss": 0.8795,
+      "step": 18587
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.228740684120008e-06,
+      "loss": 0.7556,
+      "step": 18588
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.228645136632907e-06,
+      "loss": 0.5971,
+      "step": 18589
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2285495891458056e-06,
+      "loss": 0.4356,
+      "step": 18590
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2284540416587046e-06,
+      "loss": 0.7798,
+      "step": 18591
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2283584941716032e-06,
+      "loss": 0.7041,
+      "step": 18592
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2282629466845023e-06,
+      "loss": 0.9174,
+      "step": 18593
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2281673991974017e-06,
+      "loss": 0.8872,
+      "step": 18594
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2280718517103004e-06,
+      "loss": 0.6498,
+      "step": 18595
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2279763042231994e-06,
+      "loss": 1.2226,
+      "step": 18596
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.227880756736098e-06,
+      "loss": 0.893,
+      "step": 18597
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.227785209248997e-06,
+      "loss": 0.9547,
+      "step": 18598
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2276896617618957e-06,
+      "loss": 0.8156,
+      "step": 18599
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2275941142747947e-06,
+      "loss": 1.109,
+      "step": 18600
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.227498566787694e-06,
+      "loss": 1.0688,
+      "step": 18601
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2274030193005928e-06,
+      "loss": 0.687,
+      "step": 18602
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.227307471813492e-06,
+      "loss": 1.3977,
+      "step": 18603
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2272119243263904e-06,
+      "loss": 1.077,
+      "step": 18604
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2271163768392895e-06,
+      "loss": 0.7095,
+      "step": 18605
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.227020829352188e-06,
+      "loss": 0.654,
+      "step": 18606
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226925281865087e-06,
+      "loss": 0.7376,
+      "step": 18607
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2268297343779857e-06,
+      "loss": 0.6541,
+      "step": 18608
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226734186890885e-06,
+      "loss": 1.0944,
+      "step": 18609
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2266386394037842e-06,
+      "loss": 0.9021,
+      "step": 18610
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226543091916683e-06,
+      "loss": 0.9001,
+      "step": 18611
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226447544429582e-06,
+      "loss": 0.8639,
+      "step": 18612
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2263519969424805e-06,
+      "loss": 0.7673,
+      "step": 18613
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2262564494553795e-06,
+      "loss": 0.6531,
+      "step": 18614
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226160901968278e-06,
+      "loss": 0.9093,
+      "step": 18615
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2260653544811776e-06,
+      "loss": 0.916,
+      "step": 18616
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2259698069940767e-06,
+      "loss": 0.5354,
+      "step": 18617
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2258742595069753e-06,
+      "loss": 0.9016,
+      "step": 18618
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2257787120198743e-06,
+      "loss": 0.9695,
+      "step": 18619
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.225683164532773e-06,
+      "loss": 0.4388,
+      "step": 18620
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.225587617045672e-06,
+      "loss": 0.892,
+      "step": 18621
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2254920695585706e-06,
+      "loss": 0.7569,
+      "step": 18622
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22539652207147e-06,
+      "loss": 0.4054,
+      "step": 18623
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2253009745843687e-06,
+      "loss": 1.0711,
+      "step": 18624
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2252054270972677e-06,
+      "loss": 0.9775,
+      "step": 18625
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2251098796101667e-06,
+      "loss": 0.8952,
+      "step": 18626
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2250143321230654e-06,
+      "loss": 0.8722,
+      "step": 18627
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2249187846359644e-06,
+      "loss": 0.7871,
+      "step": 18628
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.224823237148863e-06,
+      "loss": 1.119,
+      "step": 18629
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2247276896617625e-06,
+      "loss": 0.4102,
+      "step": 18630
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.224632142174661e-06,
+      "loss": 0.6331,
+      "step": 18631
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22453659468756e-06,
+      "loss": 0.761,
+      "step": 18632
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2244410472004587e-06,
+      "loss": 1.1096,
+      "step": 18633
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2243454997133578e-06,
+      "loss": 0.8076,
+      "step": 18634
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2242499522262564e-06,
+      "loss": 0.981,
+      "step": 18635
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2241544047391554e-06,
+      "loss": 1.3218,
+      "step": 18636
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.224058857252055e-06,
+      "loss": 0.5631,
+      "step": 18637
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2239633097649535e-06,
+      "loss": 0.6624,
+      "step": 18638
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2238677622778525e-06,
+      "loss": 0.7334,
+      "step": 18639
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223772214790751e-06,
+      "loss": 0.9413,
+      "step": 18640
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22367666730365e-06,
+      "loss": 0.8159,
+      "step": 18641
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223581119816549e-06,
+      "loss": 0.8,
+      "step": 18642
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223485572329448e-06,
+      "loss": 0.9162,
+      "step": 18643
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2233900248423465e-06,
+      "loss": 0.6286,
+      "step": 18644
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223294477355246e-06,
+      "loss": 1.0333,
+      "step": 18645
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223198929868145e-06,
+      "loss": 0.9382,
+      "step": 18646
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2231033823810436e-06,
+      "loss": 0.6668,
+      "step": 18647
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2230078348939426e-06,
+      "loss": 0.6983,
+      "step": 18648
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2229122874068412e-06,
+      "loss": 0.7742,
+      "step": 18649
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2228167399197403e-06,
+      "loss": 0.5383,
+      "step": 18650
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222721192432639e-06,
+      "loss": 0.8892,
+      "step": 18651
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2226256449455384e-06,
+      "loss": 0.4939,
+      "step": 18652
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2225300974584374e-06,
+      "loss": 1.0588,
+      "step": 18653
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222434549971336e-06,
+      "loss": 0.9828,
+      "step": 18654
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222339002484235e-06,
+      "loss": 0.8374,
+      "step": 18655
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2222434549971337e-06,
+      "loss": 0.8067,
+      "step": 18656
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2221479075100327e-06,
+      "loss": 0.5243,
+      "step": 18657
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2220523600229313e-06,
+      "loss": 0.8978,
+      "step": 18658
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2219568125358308e-06,
+      "loss": 0.8423,
+      "step": 18659
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22186126504873e-06,
+      "loss": 1.1305,
+      "step": 18660
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2217657175616284e-06,
+      "loss": 1.1027,
+      "step": 18661
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2216701700745275e-06,
+      "loss": 0.6052,
+      "step": 18662
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.221574622587426e-06,
+      "loss": 0.8778,
+      "step": 18663
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.221479075100325e-06,
+      "loss": 0.9325,
+      "step": 18664
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2213835276132237e-06,
+      "loss": 0.934,
+      "step": 18665
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.221287980126123e-06,
+      "loss": 0.789,
+      "step": 18666
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2211924326390222e-06,
+      "loss": 0.6318,
+      "step": 18667
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.221096885151921e-06,
+      "loss": 1.2792,
+      "step": 18668
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22100133766482e-06,
+      "loss": 0.8938,
+      "step": 18669
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2209057901777185e-06,
+      "loss": 0.9726,
+      "step": 18670
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2208102426906175e-06,
+      "loss": 0.5091,
+      "step": 18671
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.220714695203516e-06,
+      "loss": 0.9792,
+      "step": 18672
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2206191477164156e-06,
+      "loss": 0.6995,
+      "step": 18673
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2205236002293142e-06,
+      "loss": 0.7562,
+      "step": 18674
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2204280527422133e-06,
+      "loss": 0.7626,
+      "step": 18675
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220332505255112e-06,
+      "loss": 1.2789,
+      "step": 18676
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.220236957768011e-06,
+      "loss": 0.5965,
+      "step": 18677
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.22014141028091e-06,
+      "loss": 0.9727,
+      "step": 18678
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2200458627938086e-06,
+      "loss": 0.754,
+      "step": 18679
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2199503153067076e-06,
+      "loss": 0.6946,
+      "step": 18680
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2198547678196067e-06,
+      "loss": 0.8221,
+      "step": 18681
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2197592203325057e-06,
+      "loss": 1.0358,
+      "step": 18682
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2196636728454043e-06,
+      "loss": 0.8553,
+      "step": 18683
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2195681253583034e-06,
+      "loss": 0.9807,
+      "step": 18684
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.219472577871202e-06,
+      "loss": 1.1165,
+      "step": 18685
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.219377030384101e-06,
+      "loss": 0.8716,
+      "step": 18686
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2192814828969996e-06,
+      "loss": 1.0434,
+      "step": 18687
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.219185935409899e-06,
+      "loss": 0.7485,
+      "step": 18688
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.219090387922798e-06,
+      "loss": 0.6899,
+      "step": 18689
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2189948404356967e-06,
+      "loss": 1.3114,
+      "step": 18690
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2188992929485958e-06,
+      "loss": 1.0454,
+      "step": 18691
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2188037454614944e-06,
+      "loss": 0.7846,
+      "step": 18692
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2187081979743934e-06,
+      "loss": 0.6759,
+      "step": 18693
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.218612650487292e-06,
+      "loss": 0.8742,
+      "step": 18694
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2185171030001915e-06,
+      "loss": 0.6471,
+      "step": 18695
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2184215555130905e-06,
+      "loss": 0.8789,
+      "step": 18696
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.218326008025989e-06,
+      "loss": 0.7842,
+      "step": 18697
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.218230460538888e-06,
+      "loss": 0.5177,
+      "step": 18698
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.218134913051787e-06,
+      "loss": 0.6518,
+      "step": 18699
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.218039365564686e-06,
+      "loss": 0.9403,
+      "step": 18700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2179438180775845e-06,
+      "loss": 0.8239,
+      "step": 18701
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217848270590484e-06,
+      "loss": 0.9823,
+      "step": 18702
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217752723103383e-06,
+      "loss": 1.0475,
+      "step": 18703
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2176571756162816e-06,
+      "loss": 0.52,
+      "step": 18704
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2175616281291806e-06,
+      "loss": 0.7838,
+      "step": 18705
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2174660806420792e-06,
+      "loss": 0.9076,
+      "step": 18706
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2173705331549783e-06,
+      "loss": 1.0666,
+      "step": 18707
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217274985667877e-06,
+      "loss": 0.9451,
+      "step": 18708
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2171794381807764e-06,
+      "loss": 0.8703,
+      "step": 18709
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2170838906936754e-06,
+      "loss": 1.0214,
+      "step": 18710
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.216988343206574e-06,
+      "loss": 1.1478,
+      "step": 18711
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.216892795719473e-06,
+      "loss": 0.7873,
+      "step": 18712
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2167972482323717e-06,
+      "loss": 0.9029,
+      "step": 18713
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2167017007452707e-06,
+      "loss": 0.895,
+      "step": 18714
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2166061532581693e-06,
+      "loss": 0.9133,
+      "step": 18715
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2165106057710684e-06,
+      "loss": 0.7309,
+      "step": 18716
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.216415058283968e-06,
+      "loss": 1.1335,
+      "step": 18717
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2163195107968664e-06,
+      "loss": 1.0518,
+      "step": 18718
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2162239633097655e-06,
+      "loss": 1.1157,
+      "step": 18719
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.216128415822664e-06,
+      "loss": 0.8986,
+      "step": 18720
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.216032868335563e-06,
+      "loss": 0.9769,
+      "step": 18721
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2159373208484617e-06,
+      "loss": 0.6353,
+      "step": 18722
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2158417733613608e-06,
+      "loss": 1.0433,
+      "step": 18723
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.21574622587426e-06,
+      "loss": 0.7508,
+      "step": 18724
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.215650678387159e-06,
+      "loss": 0.6687,
+      "step": 18725
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2155551309000575e-06,
+      "loss": 0.8547,
+      "step": 18726
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2154595834129565e-06,
+      "loss": 0.6639,
+      "step": 18727
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215364035925855e-06,
+      "loss": 0.9255,
+      "step": 18728
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215268488438754e-06,
+      "loss": 1.174,
+      "step": 18729
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.215172940951653e-06,
+      "loss": 1.1941,
+      "step": 18730
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2150773934645522e-06,
+      "loss": 0.6182,
+      "step": 18731
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2149818459774513e-06,
+      "loss": 0.8679,
+      "step": 18732
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21488629849035e-06,
+      "loss": 0.7452,
+      "step": 18733
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.214790751003249e-06,
+      "loss": 0.847,
+      "step": 18734
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2146952035161475e-06,
+      "loss": 1.4566,
+      "step": 18735
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2145996560290466e-06,
+      "loss": 1.0674,
+      "step": 18736
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.214504108541945e-06,
+      "loss": 1.0368,
+      "step": 18737
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2144085610548447e-06,
+      "loss": 0.945,
+      "step": 18738
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2143130135677437e-06,
+      "loss": 1.2648,
+      "step": 18739
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2142174660806423e-06,
+      "loss": 0.3606,
+      "step": 18740
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2141219185935414e-06,
+      "loss": 0.8621,
+      "step": 18741
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21402637110644e-06,
+      "loss": 1.1498,
+      "step": 18742
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.213930823619339e-06,
+      "loss": 1.1299,
+      "step": 18743
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2138352761322376e-06,
+      "loss": 0.7151,
+      "step": 18744
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.213739728645137e-06,
+      "loss": 0.8306,
+      "step": 18745
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.213644181158036e-06,
+      "loss": 1.009,
+      "step": 18746
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2135486336709347e-06,
+      "loss": 1.1608,
+      "step": 18747
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2134530861838338e-06,
+      "loss": 1.0151,
+      "step": 18748
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2133575386967324e-06,
+      "loss": 0.8016,
+      "step": 18749
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2132619912096314e-06,
+      "loss": 0.7848,
+      "step": 18750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.21316644372253e-06,
+      "loss": 0.8797,
+      "step": 18751
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.213070896235429e-06,
+      "loss": 1.0009,
+      "step": 18752
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2129753487483285e-06,
+      "loss": 0.6834,
+      "step": 18753
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212879801261227e-06,
+      "loss": 0.7931,
+      "step": 18754
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212784253774126e-06,
+      "loss": 0.9417,
+      "step": 18755
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212688706287025e-06,
+      "loss": 0.595,
+      "step": 18756
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212593158799924e-06,
+      "loss": 1.3654,
+      "step": 18757
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2124976113128225e-06,
+      "loss": 0.8293,
+      "step": 18758
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2124020638257215e-06,
+      "loss": 0.7532,
+      "step": 18759
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212306516338621e-06,
+      "loss": 1.4495,
+      "step": 18760
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2122109688515196e-06,
+      "loss": 0.8842,
+      "step": 18761
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2121154213644186e-06,
+      "loss": 0.7318,
+      "step": 18762
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2120198738773172e-06,
+      "loss": 0.7442,
+      "step": 18763
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2119243263902163e-06,
+      "loss": 0.9175,
+      "step": 18764
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211828778903115e-06,
+      "loss": 0.8527,
+      "step": 18765
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211733231416014e-06,
+      "loss": 0.865,
+      "step": 18766
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211637683928913e-06,
+      "loss": 1.3197,
+      "step": 18767
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211542136441812e-06,
+      "loss": 0.5098,
+      "step": 18768
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211446588954711e-06,
+      "loss": 0.9465,
+      "step": 18769
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2113510414676097e-06,
+      "loss": 0.3391,
+      "step": 18770
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2112554939805087e-06,
+      "loss": 0.9238,
+      "step": 18771
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2111599464934073e-06,
+      "loss": 0.7034,
+      "step": 18772
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2110643990063064e-06,
+      "loss": 1.2837,
+      "step": 18773
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2109688515192054e-06,
+      "loss": 1.0943,
+      "step": 18774
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2108733040321044e-06,
+      "loss": 0.9313,
+      "step": 18775
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.210777756545003e-06,
+      "loss": 0.8419,
+      "step": 18776
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.210682209057902e-06,
+      "loss": 0.6748,
+      "step": 18777
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2105866615708007e-06,
+      "loss": 0.6094,
+      "step": 18778
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2104911140836997e-06,
+      "loss": 0.8102,
+      "step": 18779
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2103955665965988e-06,
+      "loss": 0.793,
+      "step": 18780
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210300019109498e-06,
+      "loss": 0.8448,
+      "step": 18781
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.210204471622397e-06,
+      "loss": 0.8655,
+      "step": 18782
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2101089241352955e-06,
+      "loss": 1.1092,
+      "step": 18783
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2100133766481945e-06,
+      "loss": 0.5955,
+      "step": 18784
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209917829161093e-06,
+      "loss": 0.7735,
+      "step": 18785
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209822281673992e-06,
+      "loss": 0.788,
+      "step": 18786
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2097267341868908e-06,
+      "loss": 1.0432,
+      "step": 18787
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.20963118669979e-06,
+      "loss": 0.9093,
+      "step": 18788
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2095356392126893e-06,
+      "loss": 1.1331,
+      "step": 18789
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209440091725588e-06,
+      "loss": 1.4734,
+      "step": 18790
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209344544238487e-06,
+      "loss": 1.2278,
+      "step": 18791
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2092489967513855e-06,
+      "loss": 0.5614,
+      "step": 18792
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2091534492642846e-06,
+      "loss": 0.9193,
+      "step": 18793
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.209057901777183e-06,
+      "loss": 1.2128,
+      "step": 18794
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2089623542900822e-06,
+      "loss": 0.7388,
+      "step": 18795
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2088668068029817e-06,
+      "loss": 1.3721,
+      "step": 18796
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2087712593158803e-06,
+      "loss": 0.5639,
+      "step": 18797
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2086757118287793e-06,
+      "loss": 0.7125,
+      "step": 18798
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.208580164341678e-06,
+      "loss": 1.0227,
+      "step": 18799
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.208484616854577e-06,
+      "loss": 0.5401,
+      "step": 18800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2083890693674756e-06,
+      "loss": 0.7994,
+      "step": 18801
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2082935218803747e-06,
+      "loss": 1.0542,
+      "step": 18802
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.208197974393274e-06,
+      "loss": 1.034,
+      "step": 18803
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2081024269061727e-06,
+      "loss": 0.5708,
+      "step": 18804
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2080068794190718e-06,
+      "loss": 0.8043,
+      "step": 18805
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2079113319319704e-06,
+      "loss": 0.9733,
+      "step": 18806
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2078157844448694e-06,
+      "loss": 0.4934,
+      "step": 18807
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207720236957768e-06,
+      "loss": 0.5711,
+      "step": 18808
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207624689470667e-06,
+      "loss": 0.817,
+      "step": 18809
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2075291419835665e-06,
+      "loss": 0.8246,
+      "step": 18810
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207433594496465e-06,
+      "loss": 0.866,
+      "step": 18811
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207338047009364e-06,
+      "loss": 1.0829,
+      "step": 18812
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207242499522263e-06,
+      "loss": 0.5417,
+      "step": 18813
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.207146952035162e-06,
+      "loss": 0.673,
+      "step": 18814
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2070514045480605e-06,
+      "loss": 0.982,
+      "step": 18815
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2069558570609595e-06,
+      "loss": 1.0163,
+      "step": 18816
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2068603095738585e-06,
+      "loss": 0.9351,
+      "step": 18817
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2067647620867576e-06,
+      "loss": 1.2534,
+      "step": 18818
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2066692145996566e-06,
+      "loss": 0.6827,
+      "step": 18819
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2065736671125552e-06,
+      "loss": 0.6614,
+      "step": 18820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2064781196254543e-06,
+      "loss": 1.0434,
+      "step": 18821
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.206382572138353e-06,
+      "loss": 0.945,
+      "step": 18822
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.206287024651252e-06,
+      "loss": 0.9046,
+      "step": 18823
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2061914771641505e-06,
+      "loss": 1.1027,
+      "step": 18824
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.20609592967705e-06,
+      "loss": 0.9378,
+      "step": 18825
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2060003821899486e-06,
+      "loss": 1.0021,
+      "step": 18826
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2059048347028477e-06,
+      "loss": 0.8725,
+      "step": 18827
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2058092872157463e-06,
+      "loss": 0.8572,
+      "step": 18828
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2057137397286453e-06,
+      "loss": 0.6283,
+      "step": 18829
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.205618192241544e-06,
+      "loss": 1.3257,
+      "step": 18830
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205522644754443e-06,
+      "loss": 0.8258,
+      "step": 18831
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2054270972673424e-06,
+      "loss": 0.7919,
+      "step": 18832
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.205331549780241e-06,
+      "loss": 0.9385,
+      "step": 18833
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.20523600229314e-06,
+      "loss": 1.1997,
+      "step": 18834
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2051404548060387e-06,
+      "loss": 0.8003,
+      "step": 18835
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2050449073189377e-06,
+      "loss": 0.8339,
+      "step": 18836
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2049493598318363e-06,
+      "loss": 1.3924,
+      "step": 18837
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2048538123447354e-06,
+      "loss": 0.9108,
+      "step": 18838
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.204758264857635e-06,
+      "loss": 0.4885,
+      "step": 18839
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2046627173705335e-06,
+      "loss": 0.7059,
+      "step": 18840
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2045671698834325e-06,
+      "loss": 0.683,
+      "step": 18841
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.204471622396331e-06,
+      "loss": 1.1373,
+      "step": 18842
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.20437607490923e-06,
+      "loss": 0.5813,
+      "step": 18843
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2042805274221288e-06,
+      "loss": 0.9905,
+      "step": 18844
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.204184979935028e-06,
+      "loss": 0.9043,
+      "step": 18845
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2040894324479273e-06,
+      "loss": 1.026,
+      "step": 18846
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203993884960826e-06,
+      "loss": 1.0407,
+      "step": 18847
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203898337473725e-06,
+      "loss": 0.8424,
+      "step": 18848
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2038027899866235e-06,
+      "loss": 1.1278,
+      "step": 18849
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2037072424995226e-06,
+      "loss": 0.9502,
+      "step": 18850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203611695012421e-06,
+      "loss": 0.7677,
+      "step": 18851
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2035161475253202e-06,
+      "loss": 1.0966,
+      "step": 18852
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2034206000382197e-06,
+      "loss": 0.5396,
+      "step": 18853
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2033250525511183e-06,
+      "loss": 0.6881,
+      "step": 18854
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2032295050640173e-06,
+      "loss": 0.8908,
+      "step": 18855
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203133957576916e-06,
+      "loss": 0.9798,
+      "step": 18856
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203038410089815e-06,
+      "loss": 0.6976,
+      "step": 18857
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2029428626027136e-06,
+      "loss": 0.9456,
+      "step": 18858
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2028473151156127e-06,
+      "loss": 0.9694,
+      "step": 18859
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2027517676285113e-06,
+      "loss": 1.3365,
+      "step": 18860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2026562201414107e-06,
+      "loss": 1.1306,
+      "step": 18861
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2025606726543098e-06,
+      "loss": 0.8167,
+      "step": 18862
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2024651251672084e-06,
+      "loss": 0.7054,
+      "step": 18863
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2023695776801074e-06,
+      "loss": 0.5755,
+      "step": 18864
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.202274030193006e-06,
+      "loss": 0.6928,
+      "step": 18865
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.202178482705905e-06,
+      "loss": 0.8231,
+      "step": 18866
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2020829352188037e-06,
+      "loss": 0.9402,
+      "step": 18867
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201987387731703e-06,
+      "loss": 1.0663,
+      "step": 18868
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2018918402446018e-06,
+      "loss": 1.1675,
+      "step": 18869
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201796292757501e-06,
+      "loss": 0.8133,
+      "step": 18870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2017007452704e-06,
+      "loss": 1.1424,
+      "step": 18871
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2016051977832985e-06,
+      "loss": 1.2749,
+      "step": 18872
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2015096502961975e-06,
+      "loss": 1.1997,
+      "step": 18873
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201414102809096e-06,
+      "loss": 0.5922,
+      "step": 18874
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2013185553219956e-06,
+      "loss": 1.1037,
+      "step": 18875
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201223007834894e-06,
+      "loss": 1.284,
+      "step": 18876
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2011274603477932e-06,
+      "loss": 0.9735,
+      "step": 18877
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201031912860692e-06,
+      "loss": 1.0276,
+      "step": 18878
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.200936365373591e-06,
+      "loss": 1.2099,
+      "step": 18879
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2008408178864895e-06,
+      "loss": 0.9605,
+      "step": 18880
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2007452703993885e-06,
+      "loss": 0.7062,
+      "step": 18881
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.200649722912288e-06,
+      "loss": 1.0645,
+      "step": 18882
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2005541754251866e-06,
+      "loss": 0.8693,
+      "step": 18883
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2004586279380857e-06,
+      "loss": 1.0296,
+      "step": 18884
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2003630804509843e-06,
+      "loss": 1.2712,
+      "step": 18885
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.2002675329638833e-06,
+      "loss": 0.6821,
+      "step": 18886
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200171985476782e-06,
+      "loss": 0.5825,
+      "step": 18887
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.200076437989681e-06,
+      "loss": 0.6171,
+      "step": 18888
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1999808905025804e-06,
+      "loss": 0.7988,
+      "step": 18889
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.199885343015479e-06,
+      "loss": 0.7771,
+      "step": 18890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.199789795528378e-06,
+      "loss": 0.8768,
+      "step": 18891
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1996942480412767e-06,
+      "loss": 0.992,
+      "step": 18892
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1995987005541757e-06,
+      "loss": 1.0641,
+      "step": 18893
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1995031530670743e-06,
+      "loss": 0.5407,
+      "step": 18894
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1994076055799734e-06,
+      "loss": 0.9288,
+      "step": 18895
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.199312058092872e-06,
+      "loss": 0.9797,
+      "step": 18896
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1992165106057715e-06,
+      "loss": 0.5435,
+      "step": 18897
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1991209631186705e-06,
+      "loss": 0.7786,
+      "step": 18898
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.199025415631569e-06,
+      "loss": 0.7991,
+      "step": 18899
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.198929868144468e-06,
+      "loss": 0.6855,
+      "step": 18900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1988343206573668e-06,
+      "loss": 0.7195,
+      "step": 18901
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.198738773170266e-06,
+      "loss": 0.9137,
+      "step": 18902
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1986432256831644e-06,
+      "loss": 0.5956,
+      "step": 18903
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.198547678196064e-06,
+      "loss": 0.8836,
+      "step": 18904
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.198452130708963e-06,
+      "loss": 0.8779,
+      "step": 18905
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1983565832218615e-06,
+      "loss": 0.9151,
+      "step": 18906
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1982610357347606e-06,
+      "loss": 1.1029,
+      "step": 18907
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.198165488247659e-06,
+      "loss": 1.2223,
+      "step": 18908
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1980699407605582e-06,
+      "loss": 1.1023,
+      "step": 18909
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197974393273457e-06,
+      "loss": 0.4929,
+      "step": 18910
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1978788457863563e-06,
+      "loss": 0.9936,
+      "step": 18911
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1977832982992553e-06,
+      "loss": 0.8978,
+      "step": 18912
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197687750812154e-06,
+      "loss": 1.1423,
+      "step": 18913
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197592203325053e-06,
+      "loss": 0.5941,
+      "step": 18914
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1974966558379516e-06,
+      "loss": 0.678,
+      "step": 18915
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1974011083508507e-06,
+      "loss": 0.7503,
+      "step": 18916
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1973055608637493e-06,
+      "loss": 0.5848,
+      "step": 18917
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1972100133766487e-06,
+      "loss": 0.7791,
+      "step": 18918
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1971144658895473e-06,
+      "loss": 0.5634,
+      "step": 18919
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1970189184024464e-06,
+      "loss": 1.3792,
+      "step": 18920
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1969233709153454e-06,
+      "loss": 0.8781,
+      "step": 18921
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196827823428244e-06,
+      "loss": 0.7983,
+      "step": 18922
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196732275941143e-06,
+      "loss": 0.8106,
+      "step": 18923
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1966367284540417e-06,
+      "loss": 0.7398,
+      "step": 18924
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196541180966941e-06,
+      "loss": 0.6644,
+      "step": 18925
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1964456334798398e-06,
+      "loss": 0.6386,
+      "step": 18926
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196350085992739e-06,
+      "loss": 0.629,
+      "step": 18927
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1962545385056374e-06,
+      "loss": 1.0599,
+      "step": 18928
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1961589910185365e-06,
+      "loss": 0.8877,
+      "step": 18929
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196063443531435e-06,
+      "loss": 1.3455,
+      "step": 18930
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.195967896044334e-06,
+      "loss": 0.5517,
+      "step": 18931
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1958723485572327e-06,
+      "loss": 0.7206,
+      "step": 18932
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.195776801070132e-06,
+      "loss": 0.7472,
+      "step": 18933
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1956812535830312e-06,
+      "loss": 1.2692,
+      "step": 18934
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.19558570609593e-06,
+      "loss": 0.8041,
+      "step": 18935
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195490158608829e-06,
+      "loss": 0.7427,
+      "step": 18936
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1953946111217275e-06,
+      "loss": 0.5491,
+      "step": 18937
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1952990636346265e-06,
+      "loss": 0.9829,
+      "step": 18938
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.195203516147525e-06,
+      "loss": 0.7611,
+      "step": 18939
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1951079686604246e-06,
+      "loss": 1.1765,
+      "step": 18940
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1950124211733237e-06,
+      "loss": 0.6715,
+      "step": 18941
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1949168736862223e-06,
+      "loss": 0.8072,
+      "step": 18942
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1948213261991213e-06,
+      "loss": 1.2436,
+      "step": 18943
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.19472577871202e-06,
+      "loss": 0.8819,
+      "step": 18944
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.194630231224919e-06,
+      "loss": 1.6033,
+      "step": 18945
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1945346837378176e-06,
+      "loss": 1.0005,
+      "step": 18946
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.194439136250717e-06,
+      "loss": 0.8377,
+      "step": 18947
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.194343588763616e-06,
+      "loss": 1.0149,
+      "step": 18948
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1942480412765147e-06,
+      "loss": 1.1605,
+      "step": 18949
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1941524937894137e-06,
+      "loss": 0.6303,
+      "step": 18950
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1940569463023123e-06,
+      "loss": 0.8338,
+      "step": 18951
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1939613988152114e-06,
+      "loss": 0.7465,
+      "step": 18952
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.19386585132811e-06,
+      "loss": 0.9728,
+      "step": 18953
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1937703038410095e-06,
+      "loss": 0.7536,
+      "step": 18954
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1936747563539085e-06,
+      "loss": 0.6428,
+      "step": 18955
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.193579208866807e-06,
+      "loss": 0.7021,
+      "step": 18956
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.193483661379706e-06,
+      "loss": 0.7839,
+      "step": 18957
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1933881138926048e-06,
+      "loss": 0.9324,
+      "step": 18958
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.193292566405504e-06,
+      "loss": 0.743,
+      "step": 18959
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1931970189184024e-06,
+      "loss": 1.1496,
+      "step": 18960
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.193101471431302e-06,
+      "loss": 0.9634,
+      "step": 18961
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.193005923944201e-06,
+      "loss": 1.0739,
+      "step": 18962
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1929103764570995e-06,
+      "loss": 1.0236,
+      "step": 18963
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1928148289699986e-06,
+      "loss": 0.7694,
+      "step": 18964
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.192719281482897e-06,
+      "loss": 0.7007,
+      "step": 18965
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1926237339957962e-06,
+      "loss": 0.8853,
+      "step": 18966
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.192528186508695e-06,
+      "loss": 0.8318,
+      "step": 18967
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.192432639021594e-06,
+      "loss": 0.9002,
+      "step": 18968
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.192337091534493e-06,
+      "loss": 0.9977,
+      "step": 18969
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.192241544047392e-06,
+      "loss": 0.8721,
+      "step": 18970
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1921459965602906e-06,
+      "loss": 1.066,
+      "step": 18971
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1920504490731896e-06,
+      "loss": 0.6764,
+      "step": 18972
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1919549015860887e-06,
+      "loss": 1.1975,
+      "step": 18973
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1918593540989873e-06,
+      "loss": 0.8364,
+      "step": 18974
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1917638066118863e-06,
+      "loss": 0.9153,
+      "step": 18975
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1916682591247853e-06,
+      "loss": 0.8025,
+      "step": 18976
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1915727116376844e-06,
+      "loss": 0.4561,
+      "step": 18977
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.191477164150583e-06,
+      "loss": 0.7202,
+      "step": 18978
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.191381616663482e-06,
+      "loss": 1.2712,
+      "step": 18979
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1912860691763807e-06,
+      "loss": 0.3928,
+      "step": 18980
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1911905216892797e-06,
+      "loss": 0.9144,
+      "step": 18981
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1910949742021783e-06,
+      "loss": 1.014,
+      "step": 18982
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1909994267150778e-06,
+      "loss": 0.9897,
+      "step": 18983
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.190903879227977e-06,
+      "loss": 0.6578,
+      "step": 18984
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1908083317408754e-06,
+      "loss": 1.1726,
+      "step": 18985
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1907127842537745e-06,
+      "loss": 0.5066,
+      "step": 18986
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.190617236766673e-06,
+      "loss": 1.1233,
+      "step": 18987
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190521689279572e-06,
+      "loss": 0.6448,
+      "step": 18988
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1904261417924707e-06,
+      "loss": 0.9254,
+      "step": 18989
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.19033059430537e-06,
+      "loss": 0.9639,
+      "step": 18990
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1902350468182692e-06,
+      "loss": 1.1728,
+      "step": 18991
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190139499331168e-06,
+      "loss": 0.7129,
+      "step": 18992
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.190043951844067e-06,
+      "loss": 0.6713,
+      "step": 18993
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1899484043569655e-06,
+      "loss": 0.7504,
+      "step": 18994
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1898528568698645e-06,
+      "loss": 0.5055,
+      "step": 18995
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.189757309382763e-06,
+      "loss": 0.86,
+      "step": 18996
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1896617618956626e-06,
+      "loss": 0.9218,
+      "step": 18997
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1895662144085617e-06,
+      "loss": 0.9547,
+      "step": 18998
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1894706669214603e-06,
+      "loss": 0.921,
+      "step": 18999
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1893751194343593e-06,
+      "loss": 0.7122,
+      "step": 19000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.189279571947258e-06,
+      "loss": 1.1087,
+      "step": 19001
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.189184024460157e-06,
+      "loss": 1.0002,
+      "step": 19002
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1890884769730556e-06,
+      "loss": 0.9763,
+      "step": 19003
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1889929294859546e-06,
+      "loss": 1.3549,
+      "step": 19004
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.188897381998854e-06,
+      "loss": 0.6717,
+      "step": 19005
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1888018345117527e-06,
+      "loss": 0.6062,
+      "step": 19006
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1887062870246517e-06,
+      "loss": 0.5503,
+      "step": 19007
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1886107395375503e-06,
+      "loss": 0.5684,
+      "step": 19008
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1885151920504494e-06,
+      "loss": 1.1374,
+      "step": 19009
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.188419644563348e-06,
+      "loss": 0.8447,
+      "step": 19010
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.188324097076247e-06,
+      "loss": 0.7527,
+      "step": 19011
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1882285495891465e-06,
+      "loss": 0.9067,
+      "step": 19012
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.188133002102045e-06,
+      "loss": 0.8935,
+      "step": 19013
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.188037454614944e-06,
+      "loss": 0.4084,
+      "step": 19014
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1879419071278428e-06,
+      "loss": 0.8273,
+      "step": 19015
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.187846359640742e-06,
+      "loss": 0.7847,
+      "step": 19016
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1877508121536404e-06,
+      "loss": 0.6829,
+      "step": 19017
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1876552646665395e-06,
+      "loss": 1.0112,
+      "step": 19018
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1875597171794385e-06,
+      "loss": 1.1651,
+      "step": 19019
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1874641696923375e-06,
+      "loss": 0.8437,
+      "step": 19020
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.187368622205236e-06,
+      "loss": 1.0188,
+      "step": 19021
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.187273074718135e-06,
+      "loss": 0.998,
+      "step": 19022
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1871775272310342e-06,
+      "loss": 0.8502,
+      "step": 19023
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.187081979743933e-06,
+      "loss": 0.802,
+      "step": 19024
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186986432256832e-06,
+      "loss": 0.7114,
+      "step": 19025
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186890884769731e-06,
+      "loss": 1.0988,
+      "step": 19026
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.18679533728263e-06,
+      "loss": 1.1991,
+      "step": 19027
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1866997897955286e-06,
+      "loss": 0.6407,
+      "step": 19028
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1866042423084276e-06,
+      "loss": 0.9552,
+      "step": 19029
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1865086948213262e-06,
+      "loss": 1.1785,
+      "step": 19030
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1864131473342253e-06,
+      "loss": 0.9815,
+      "step": 19031
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186317599847124e-06,
+      "loss": 0.857,
+      "step": 19032
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1862220523600233e-06,
+      "loss": 0.9679,
+      "step": 19033
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1861265048729224e-06,
+      "loss": 0.7694,
+      "step": 19034
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186030957385821e-06,
+      "loss": 0.87,
+      "step": 19035
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.18593540989872e-06,
+      "loss": 0.8691,
+      "step": 19036
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1858398624116187e-06,
+      "loss": 0.6655,
+      "step": 19037
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1857443149245177e-06,
+      "loss": 0.5986,
+      "step": 19038
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1856487674374163e-06,
+      "loss": 0.8941,
+      "step": 19039
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1855532199503153e-06,
+      "loss": 0.5637,
+      "step": 19040
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185457672463215e-06,
+      "loss": 0.7889,
+      "step": 19041
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1853621249761134e-06,
+      "loss": 0.6167,
+      "step": 19042
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1852665774890125e-06,
+      "loss": 1.1541,
+      "step": 19043
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.185171030001911e-06,
+      "loss": 0.9985,
+      "step": 19044
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.18507548251481e-06,
+      "loss": 1.0301,
+      "step": 19045
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1849799350277087e-06,
+      "loss": 1.1163,
+      "step": 19046
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1848843875406078e-06,
+      "loss": 0.8955,
+      "step": 19047
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1847888400535072e-06,
+      "loss": 0.7946,
+      "step": 19048
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.184693292566406e-06,
+      "loss": 0.785,
+      "step": 19049
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.184597745079305e-06,
+      "loss": 1.0376,
+      "step": 19050
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1845021975922035e-06,
+      "loss": 0.7228,
+      "step": 19051
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1844066501051025e-06,
+      "loss": 0.6503,
+      "step": 19052
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.184311102618001e-06,
+      "loss": 1.0607,
+      "step": 19053
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1842155551309e-06,
+      "loss": 0.6525,
+      "step": 19054
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1841200076437997e-06,
+      "loss": 0.9543,
+      "step": 19055
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1840244601566983e-06,
+      "loss": 0.7669,
+      "step": 19056
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1839289126695973e-06,
+      "loss": 1.2098,
+      "step": 19057
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.183833365182496e-06,
+      "loss": 1.1349,
+      "step": 19058
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.183737817695395e-06,
+      "loss": 0.85,
+      "step": 19059
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1836422702082936e-06,
+      "loss": 0.7651,
+      "step": 19060
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1835467227211926e-06,
+      "loss": 0.5675,
+      "step": 19061
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1834511752340917e-06,
+      "loss": 0.9599,
+      "step": 19062
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1833556277469907e-06,
+      "loss": 0.475,
+      "step": 19063
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1832600802598897e-06,
+      "loss": 1.4157,
+      "step": 19064
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1831645327727883e-06,
+      "loss": 1.1603,
+      "step": 19065
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1830689852856874e-06,
+      "loss": 0.6243,
+      "step": 19066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182973437798586e-06,
+      "loss": 0.6597,
+      "step": 19067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182877890311485e-06,
+      "loss": 1.0132,
+      "step": 19068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182782342824384e-06,
+      "loss": 0.995,
+      "step": 19069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182686795337283e-06,
+      "loss": 1.3254,
+      "step": 19070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1825912478501817e-06,
+      "loss": 0.8945,
+      "step": 19071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1824957003630808e-06,
+      "loss": 1.4387,
+      "step": 19072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1824001528759794e-06,
+      "loss": 0.9026,
+      "step": 19073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1823046053888784e-06,
+      "loss": 1.3479,
+      "step": 19074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1822090579017775e-06,
+      "loss": 0.6298,
+      "step": 19075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182113510414676e-06,
+      "loss": 1.1658,
+      "step": 19076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1820179629275755e-06,
+      "loss": 0.6477,
+      "step": 19077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181922415440474e-06,
+      "loss": 0.8929,
+      "step": 19078
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181826867953373e-06,
+      "loss": 0.7843,
+      "step": 19079
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181731320466272e-06,
+      "loss": 1.0856,
+      "step": 19080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181635772979171e-06,
+      "loss": 1.24,
+      "step": 19081
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1815402254920695e-06,
+      "loss": 0.6241,
+      "step": 19082
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1814446780049685e-06,
+      "loss": 0.9783,
+      "step": 19083
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.181349130517868e-06,
+      "loss": 0.6393,
+      "step": 19084
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1812535830307666e-06,
+      "loss": 0.5323,
+      "step": 19085
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1811580355436656e-06,
+      "loss": 1.2158,
+      "step": 19086
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1810624880565642e-06,
+      "loss": 0.779,
+      "step": 19087
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1809669405694633e-06,
+      "loss": 0.8835,
+      "step": 19088
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180871393082362e-06,
+      "loss": 0.8781,
+      "step": 19089
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180775845595261e-06,
+      "loss": 0.9273,
+      "step": 19090
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1806802981081604e-06,
+      "loss": 0.8749,
+      "step": 19091
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180584750621059e-06,
+      "loss": 1.0714,
+      "step": 19092
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180489203133958e-06,
+      "loss": 1.3622,
+      "step": 19093
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1803936556468567e-06,
+      "loss": 0.6945,
+      "step": 19094
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1802981081597557e-06,
+      "loss": 1.2427,
+      "step": 19095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1802025606726543e-06,
+      "loss": 0.7902,
+      "step": 19096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1801070131855533e-06,
+      "loss": 0.9177,
+      "step": 19097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.180011465698453e-06,
+      "loss": 0.8053,
+      "step": 19098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1799159182113514e-06,
+      "loss": 1.1398,
+      "step": 19099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1798203707242505e-06,
+      "loss": 0.9972,
+      "step": 19100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.179724823237149e-06,
+      "loss": 0.5743,
+      "step": 19101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.179629275750048e-06,
+      "loss": 0.9422,
+      "step": 19102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1795337282629467e-06,
+      "loss": 0.8559,
+      "step": 19103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1794381807758458e-06,
+      "loss": 1.1976,
+      "step": 19104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1793426332887452e-06,
+      "loss": 0.8316,
+      "step": 19105
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.179247085801644e-06,
+      "loss": 0.6451,
+      "step": 19106
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.179151538314543e-06,
+      "loss": 0.985,
+      "step": 19107
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1790559908274415e-06,
+      "loss": 0.5439,
+      "step": 19108
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1789604433403405e-06,
+      "loss": 0.7348,
+      "step": 19109
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.178864895853239e-06,
+      "loss": 0.9123,
+      "step": 19110
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.178769348366138e-06,
+      "loss": 0.7563,
+      "step": 19111
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.178673800879037e-06,
+      "loss": 0.6838,
+      "step": 19112
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1785782533919363e-06,
+      "loss": 0.7408,
+      "step": 19113
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1784827059048353e-06,
+      "loss": 1.2392,
+      "step": 19114
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.178387158417734e-06,
+      "loss": 0.8973,
+      "step": 19115
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.178291610930633e-06,
+      "loss": 0.673,
+      "step": 19116
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1781960634435316e-06,
+      "loss": 0.7974,
+      "step": 19117
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1781005159564306e-06,
+      "loss": 0.6736,
+      "step": 19118
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1780049684693292e-06,
+      "loss": 0.6356,
+      "step": 19119
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1779094209822287e-06,
+      "loss": 1.4171,
+      "step": 19120
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1778138734951273e-06,
+      "loss": 1.0907,
+      "step": 19121
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1777183260080263e-06,
+      "loss": 1.0297,
+      "step": 19122
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.177622778520925e-06,
+      "loss": 1.3702,
+      "step": 19123
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.177527231033824e-06,
+      "loss": 0.995,
+      "step": 19124
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1774316835467226e-06,
+      "loss": 1.2262,
+      "step": 19125
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1773361360596217e-06,
+      "loss": 1.0105,
+      "step": 19126
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.177240588572521e-06,
+      "loss": 0.8278,
+      "step": 19127
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1771450410854197e-06,
+      "loss": 0.762,
+      "step": 19128
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1770494935983188e-06,
+      "loss": 1.4081,
+      "step": 19129
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1769539461112174e-06,
+      "loss": 0.7093,
+      "step": 19130
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1768583986241164e-06,
+      "loss": 1.0433,
+      "step": 19131
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176762851137015e-06,
+      "loss": 0.5292,
+      "step": 19132
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176667303649914e-06,
+      "loss": 0.9208,
+      "step": 19133
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1765717561628135e-06,
+      "loss": 0.9286,
+      "step": 19134
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176476208675712e-06,
+      "loss": 0.749,
+      "step": 19135
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176380661188611e-06,
+      "loss": 0.6059,
+      "step": 19136
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.17628511370151e-06,
+      "loss": 1.3657,
+      "step": 19137
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176189566214409e-06,
+      "loss": 0.6088,
+      "step": 19138
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1760940187273075e-06,
+      "loss": 1.3195,
+      "step": 19139
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1759984712402065e-06,
+      "loss": 1.1312,
+      "step": 19140
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.175902923753106e-06,
+      "loss": 0.6083,
+      "step": 19141
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1758073762660046e-06,
+      "loss": 1.132,
+      "step": 19142
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1757118287789036e-06,
+      "loss": 0.8113,
+      "step": 19143
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1756162812918022e-06,
+      "loss": 0.8351,
+      "step": 19144
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1755207338047013e-06,
+      "loss": 1.1497,
+      "step": 19145
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1754251863176e-06,
+      "loss": 1.1541,
+      "step": 19146
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175329638830499e-06,
+      "loss": 0.8067,
+      "step": 19147
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1752340913433975e-06,
+      "loss": 0.933,
+      "step": 19148
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175138543856297e-06,
+      "loss": 0.736,
+      "step": 19149
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.175042996369196e-06,
+      "loss": 1.522,
+      "step": 19150
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1749474488820947e-06,
+      "loss": 0.7808,
+      "step": 19151
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1748519013949937e-06,
+      "loss": 0.6783,
+      "step": 19152
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1747563539078923e-06,
+      "loss": 0.6652,
+      "step": 19153
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1746608064207913e-06,
+      "loss": 1.1216,
+      "step": 19154
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.17456525893369e-06,
+      "loss": 1.0216,
+      "step": 19155
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1744697114465894e-06,
+      "loss": 1.336,
+      "step": 19156
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1743741639594885e-06,
+      "loss": 1.0514,
+      "step": 19157
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.174278616472387e-06,
+      "loss": 0.7053,
+      "step": 19158
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.174183068985286e-06,
+      "loss": 0.6916,
+      "step": 19159
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1740875214981847e-06,
+      "loss": 0.3407,
+      "step": 19160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1739919740110838e-06,
+      "loss": 1.0336,
+      "step": 19161
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1738964265239824e-06,
+      "loss": 0.9419,
+      "step": 19162
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.173800879036882e-06,
+      "loss": 0.9124,
+      "step": 19163
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1737053315497805e-06,
+      "loss": 0.7124,
+      "step": 19164
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1736097840626795e-06,
+      "loss": 1.0639,
+      "step": 19165
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1735142365755785e-06,
+      "loss": 1.021,
+      "step": 19166
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.173418689088477e-06,
+      "loss": 0.8555,
+      "step": 19167
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.173323141601376e-06,
+      "loss": 1.1414,
+      "step": 19168
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.173227594114275e-06,
+      "loss": 1.2798,
+      "step": 19169
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1731320466271743e-06,
+      "loss": 1.2468,
+      "step": 19170
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.173036499140073e-06,
+      "loss": 0.7777,
+      "step": 19171
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.172940951652972e-06,
+      "loss": 0.9362,
+      "step": 19172
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1728454041658705e-06,
+      "loss": 1.3748,
+      "step": 19173
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1727498566787696e-06,
+      "loss": 0.6983,
+      "step": 19174
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.172654309191668e-06,
+      "loss": 1.331,
+      "step": 19175
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1725587617045672e-06,
+      "loss": 1.4596,
+      "step": 19176
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1724632142174667e-06,
+      "loss": 0.9245,
+      "step": 19177
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1723676667303653e-06,
+      "loss": 0.7927,
+      "step": 19178
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1722721192432643e-06,
+      "loss": 0.7936,
+      "step": 19179
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.172176571756163e-06,
+      "loss": 1.0038,
+      "step": 19180
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.172081024269062e-06,
+      "loss": 1.1089,
+      "step": 19181
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1719854767819606e-06,
+      "loss": 1.3922,
+      "step": 19182
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1718899292948596e-06,
+      "loss": 1.1243,
+      "step": 19183
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1717943818077583e-06,
+      "loss": 1.2707,
+      "step": 19184
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1716988343206577e-06,
+      "loss": 0.9252,
+      "step": 19185
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1716032868335568e-06,
+      "loss": 0.8745,
+      "step": 19186
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1715077393464554e-06,
+      "loss": 0.831,
+      "step": 19187
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1714121918593544e-06,
+      "loss": 1.0709,
+      "step": 19188
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.171316644372253e-06,
+      "loss": 0.8156,
+      "step": 19189
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.171221096885152e-06,
+      "loss": 0.9521,
+      "step": 19190
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1711255493980507e-06,
+      "loss": 0.9094,
+      "step": 19191
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.17103000191095e-06,
+      "loss": 0.9163,
+      "step": 19192
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.170934454423849e-06,
+      "loss": 0.6148,
+      "step": 19193
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.170838906936748e-06,
+      "loss": 0.7722,
+      "step": 19194
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.170743359449647e-06,
+      "loss": 1.0158,
+      "step": 19195
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1706478119625455e-06,
+      "loss": 1.1652,
+      "step": 19196
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1705522644754445e-06,
+      "loss": 0.8169,
+      "step": 19197
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.170456716988343e-06,
+      "loss": 1.0813,
+      "step": 19198
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1703611695012426e-06,
+      "loss": 0.9941,
+      "step": 19199
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1702656220141416e-06,
+      "loss": 0.6214,
+      "step": 19200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1701700745270402e-06,
+      "loss": 0.8717,
+      "step": 19201
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1700745270399393e-06,
+      "loss": 0.9413,
+      "step": 19202
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169978979552838e-06,
+      "loss": 1.1739,
+      "step": 19203
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169883432065737e-06,
+      "loss": 0.8284,
+      "step": 19204
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1697878845786355e-06,
+      "loss": 1.051,
+      "step": 19205
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169692337091535e-06,
+      "loss": 0.621,
+      "step": 19206
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169596789604434e-06,
+      "loss": 1.1803,
+      "step": 19207
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1695012421173326e-06,
+      "loss": 0.8537,
+      "step": 19208
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1694056946302317e-06,
+      "loss": 0.9579,
+      "step": 19209
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1693101471431303e-06,
+      "loss": 0.5799,
+      "step": 19210
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1692145996560293e-06,
+      "loss": 0.8259,
+      "step": 19211
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.169119052168928e-06,
+      "loss": 0.9402,
+      "step": 19212
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1690235046818274e-06,
+      "loss": 1.0963,
+      "step": 19213
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.168927957194726e-06,
+      "loss": 0.5582,
+      "step": 19214
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.168832409707625e-06,
+      "loss": 0.6742,
+      "step": 19215
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.168736862220524e-06,
+      "loss": 0.3877,
+      "step": 19216
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1686413147334227e-06,
+      "loss": 0.8789,
+      "step": 19217
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1685457672463218e-06,
+      "loss": 0.6117,
+      "step": 19218
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1684502197592204e-06,
+      "loss": 0.9319,
+      "step": 19219
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1683546722721194e-06,
+      "loss": 0.883,
+      "step": 19220
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1682591247850185e-06,
+      "loss": 0.6956,
+      "step": 19221
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1681635772979175e-06,
+      "loss": 0.6239,
+      "step": 19222
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.168068029810816e-06,
+      "loss": 1.0886,
+      "step": 19223
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.167972482323715e-06,
+      "loss": 0.9375,
+      "step": 19224
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1678769348366138e-06,
+      "loss": 0.8869,
+      "step": 19225
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.167781387349513e-06,
+      "loss": 0.7596,
+      "step": 19226
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1676858398624114e-06,
+      "loss": 0.6138,
+      "step": 19227
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.167590292375311e-06,
+      "loss": 1.1835,
+      "step": 19228
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.16749474488821e-06,
+      "loss": 0.6956,
+      "step": 19229
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1673991974011085e-06,
+      "loss": 0.9641,
+      "step": 19230
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1673036499140076e-06,
+      "loss": 0.7366,
+      "step": 19231
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.167208102426906e-06,
+      "loss": 0.777,
+      "step": 19232
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1671125549398052e-06,
+      "loss": 1.3266,
+      "step": 19233
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.167017007452704e-06,
+      "loss": 0.7268,
+      "step": 19234
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1669214599656033e-06,
+      "loss": 0.5131,
+      "step": 19235
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1668259124785023e-06,
+      "loss": 0.978,
+      "step": 19236
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.166730364991401e-06,
+      "loss": 1.1629,
+      "step": 19237
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1666348175043e-06,
+      "loss": 0.45,
+      "step": 19238
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1665392700171986e-06,
+      "loss": 1.2131,
+      "step": 19239
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1664437225300976e-06,
+      "loss": 0.9164,
+      "step": 19240
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1663481750429963e-06,
+      "loss": 0.7807,
+      "step": 19241
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1662526275558957e-06,
+      "loss": 0.9523,
+      "step": 19242
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1661570800687948e-06,
+      "loss": 1.216,
+      "step": 19243
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1660615325816934e-06,
+      "loss": 1.1626,
+      "step": 19244
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1659659850945924e-06,
+      "loss": 0.8439,
+      "step": 19245
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.165870437607491e-06,
+      "loss": 1.0239,
+      "step": 19246
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.16577489012039e-06,
+      "loss": 0.9019,
+      "step": 19247
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1656793426332887e-06,
+      "loss": 0.6918,
+      "step": 19248
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.165583795146188e-06,
+      "loss": 0.7207,
+      "step": 19249
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165488247659087e-06,
+      "loss": 1.1222,
+      "step": 19250
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165392700171986e-06,
+      "loss": 0.7994,
+      "step": 19251
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165297152684885e-06,
+      "loss": 0.654,
+      "step": 19252
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1652016051977835e-06,
+      "loss": 1.0356,
+      "step": 19253
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1651060577106825e-06,
+      "loss": 0.7936,
+      "step": 19254
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.165010510223581e-06,
+      "loss": 1.2434,
+      "step": 19255
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.16491496273648e-06,
+      "loss": 0.7229,
+      "step": 19256
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1648194152493796e-06,
+      "loss": 1.128,
+      "step": 19257
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1647238677622782e-06,
+      "loss": 0.9067,
+      "step": 19258
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1646283202751773e-06,
+      "loss": 0.734,
+      "step": 19259
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.164532772788076e-06,
+      "loss": 0.4875,
+      "step": 19260
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.164437225300975e-06,
+      "loss": 0.7837,
+      "step": 19261
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1643416778138735e-06,
+      "loss": 0.6707,
+      "step": 19262
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1642461303267726e-06,
+      "loss": 0.5603,
+      "step": 19263
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1641505828396716e-06,
+      "loss": 0.5322,
+      "step": 19264
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1640550353525706e-06,
+      "loss": 0.9394,
+      "step": 19265
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1639594878654693e-06,
+      "loss": 1.1262,
+      "step": 19266
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1638639403783683e-06,
+      "loss": 1.0122,
+      "step": 19267
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1637683928912673e-06,
+      "loss": 1.1025,
+      "step": 19268
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.163672845404166e-06,
+      "loss": 0.7886,
+      "step": 19269
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.163577297917065e-06,
+      "loss": 0.7328,
+      "step": 19270
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.163481750429964e-06,
+      "loss": 0.6912,
+      "step": 19271
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.163386202942863e-06,
+      "loss": 0.5778,
+      "step": 19272
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1632906554557617e-06,
+      "loss": 1.1352,
+      "step": 19273
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1631951079686607e-06,
+      "loss": 0.6297,
+      "step": 19274
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1630995604815593e-06,
+      "loss": 0.9748,
+      "step": 19275
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1630040129944584e-06,
+      "loss": 0.9475,
+      "step": 19276
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162908465507357e-06,
+      "loss": 1.0299,
+      "step": 19277
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1628129180202565e-06,
+      "loss": 0.8122,
+      "step": 19278
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1627173705331555e-06,
+      "loss": 0.855,
+      "step": 19279
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162621823046054e-06,
+      "loss": 0.5515,
+      "step": 19280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162526275558953e-06,
+      "loss": 0.9294,
+      "step": 19281
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1624307280718518e-06,
+      "loss": 1.1079,
+      "step": 19282
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162335180584751e-06,
+      "loss": 0.7149,
+      "step": 19283
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1622396330976494e-06,
+      "loss": 0.7428,
+      "step": 19284
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162144085610549e-06,
+      "loss": 0.9298,
+      "step": 19285
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162048538123448e-06,
+      "loss": 0.6404,
+      "step": 19286
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1619529906363465e-06,
+      "loss": 1.1772,
+      "step": 19287
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1618574431492456e-06,
+      "loss": 0.5563,
+      "step": 19288
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161761895662144e-06,
+      "loss": 0.6814,
+      "step": 19289
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1616663481750432e-06,
+      "loss": 1.1058,
+      "step": 19290
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161570800687942e-06,
+      "loss": 0.8891,
+      "step": 19291
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161475253200841e-06,
+      "loss": 0.9992,
+      "step": 19292
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1613797057137403e-06,
+      "loss": 0.8591,
+      "step": 19293
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161284158226639e-06,
+      "loss": 0.7974,
+      "step": 19294
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.161188610739538e-06,
+      "loss": 1.0964,
+      "step": 19295
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1610930632524366e-06,
+      "loss": 0.9633,
+      "step": 19296
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1609975157653356e-06,
+      "loss": 1.2175,
+      "step": 19297
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1609019682782343e-06,
+      "loss": 0.622,
+      "step": 19298
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1608064207911333e-06,
+      "loss": 0.7796,
+      "step": 19299
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1607108733040328e-06,
+      "loss": 0.7668,
+      "step": 19300
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1606153258169314e-06,
+      "loss": 0.8803,
+      "step": 19301
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1605197783298304e-06,
+      "loss": 1.0037,
+      "step": 19302
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160424230842729e-06,
+      "loss": 1.1682,
+      "step": 19303
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160328683355628e-06,
+      "loss": 0.5936,
+      "step": 19304
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1602331358685267e-06,
+      "loss": 0.993,
+      "step": 19305
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1601375883814257e-06,
+      "loss": 0.6227,
+      "step": 19306
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.160042040894325e-06,
+      "loss": 0.5875,
+      "step": 19307
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159946493407224e-06,
+      "loss": 0.6359,
+      "step": 19308
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159850945920123e-06,
+      "loss": 1.2066,
+      "step": 19309
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1597553984330215e-06,
+      "loss": 0.9837,
+      "step": 19310
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1596598509459205e-06,
+      "loss": 1.2221,
+      "step": 19311
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159564303458819e-06,
+      "loss": 0.7543,
+      "step": 19312
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159468755971718e-06,
+      "loss": 0.7254,
+      "step": 19313
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159373208484617e-06,
+      "loss": 0.7906,
+      "step": 19314
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1592776609975162e-06,
+      "loss": 0.8499,
+      "step": 19315
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159182113510415e-06,
+      "loss": 0.9851,
+      "step": 19316
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.159086566023314e-06,
+      "loss": 0.8825,
+      "step": 19317
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.158991018536213e-06,
+      "loss": 0.8162,
+      "step": 19318
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1588954710491115e-06,
+      "loss": 1.1736,
+      "step": 19319
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1587999235620106e-06,
+      "loss": 0.9287,
+      "step": 19320
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1587043760749096e-06,
+      "loss": 1.0533,
+      "step": 19321
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1586088285878086e-06,
+      "loss": 0.6644,
+      "step": 19322
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1585132811007073e-06,
+      "loss": 0.7399,
+      "step": 19323
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1584177336136063e-06,
+      "loss": 0.7115,
+      "step": 19324
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.158322186126505e-06,
+      "loss": 0.7628,
+      "step": 19325
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.158226638639404e-06,
+      "loss": 0.924,
+      "step": 19326
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1581310911523026e-06,
+      "loss": 0.9406,
+      "step": 19327
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1580355436652016e-06,
+      "loss": 0.7709,
+      "step": 19328
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157939996178101e-06,
+      "loss": 0.8967,
+      "step": 19329
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1578444486909997e-06,
+      "loss": 0.4506,
+      "step": 19330
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1577489012038987e-06,
+      "loss": 0.3864,
+      "step": 19331
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1576533537167973e-06,
+      "loss": 0.9588,
+      "step": 19332
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1575578062296964e-06,
+      "loss": 0.9445,
+      "step": 19333
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157462258742595e-06,
+      "loss": 0.9021,
+      "step": 19334
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157366711255494e-06,
+      "loss": 0.8343,
+      "step": 19335
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1572711637683935e-06,
+      "loss": 1.6024,
+      "step": 19336
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157175616281292e-06,
+      "loss": 0.9068,
+      "step": 19337
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157080068794191e-06,
+      "loss": 0.632,
+      "step": 19338
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1569845213070898e-06,
+      "loss": 0.6446,
+      "step": 19339
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.156888973819989e-06,
+      "loss": 0.8749,
+      "step": 19340
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1567934263328874e-06,
+      "loss": 1.1961,
+      "step": 19341
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1566978788457865e-06,
+      "loss": 0.874,
+      "step": 19342
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.156602331358686e-06,
+      "loss": 0.8053,
+      "step": 19343
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1565067838715845e-06,
+      "loss": 1.2917,
+      "step": 19344
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1564112363844836e-06,
+      "loss": 0.8196,
+      "step": 19345
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.156315688897382e-06,
+      "loss": 0.7595,
+      "step": 19346
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1562201414102812e-06,
+      "loss": 0.9293,
+      "step": 19347
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.15612459392318e-06,
+      "loss": 0.8598,
+      "step": 19348
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.156029046436079e-06,
+      "loss": 0.8689,
+      "step": 19349
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1559334989489783e-06,
+      "loss": 0.8162,
+      "step": 19350
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.155837951461877e-06,
+      "loss": 0.6936,
+      "step": 19351
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.155742403974776e-06,
+      "loss": 0.7751,
+      "step": 19352
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1556468564876746e-06,
+      "loss": 1.0961,
+      "step": 19353
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1555513090005736e-06,
+      "loss": 1.0575,
+      "step": 19354
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1554557615134723e-06,
+      "loss": 1.1333,
+      "step": 19355
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1553602140263713e-06,
+      "loss": 1.0134,
+      "step": 19356
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1552646665392703e-06,
+      "loss": 1.0467,
+      "step": 19357
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1551691190521694e-06,
+      "loss": 1.2595,
+      "step": 19358
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1550735715650684e-06,
+      "loss": 1.0734,
+      "step": 19359
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154978024077967e-06,
+      "loss": 0.7773,
+      "step": 19360
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154882476590866e-06,
+      "loss": 0.7372,
+      "step": 19361
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1547869291037647e-06,
+      "loss": 0.6768,
+      "step": 19362
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1546913816166637e-06,
+      "loss": 0.6981,
+      "step": 19363
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1545958341295623e-06,
+      "loss": 0.8791,
+      "step": 19364
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154500286642462e-06,
+      "loss": 0.9561,
+      "step": 19365
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1544047391553604e-06,
+      "loss": 0.8103,
+      "step": 19366
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1543091916682595e-06,
+      "loss": 0.6884,
+      "step": 19367
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154213644181158e-06,
+      "loss": 0.9778,
+      "step": 19368
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154118096694057e-06,
+      "loss": 1.0043,
+      "step": 19369
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.154022549206956e-06,
+      "loss": 0.8892,
+      "step": 19370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1539270017198548e-06,
+      "loss": 0.5594,
+      "step": 19371
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1538314542327542e-06,
+      "loss": 0.7442,
+      "step": 19372
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.153735906745653e-06,
+      "loss": 0.6453,
+      "step": 19373
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.153640359258552e-06,
+      "loss": 0.7001,
+      "step": 19374
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1535448117714505e-06,
+      "loss": 0.7141,
+      "step": 19375
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1534492642843495e-06,
+      "loss": 0.9141,
+      "step": 19376
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.153353716797248e-06,
+      "loss": 1.1692,
+      "step": 19377
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.153258169310147e-06,
+      "loss": 0.8719,
+      "step": 19378
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1531626218230466e-06,
+      "loss": 0.9128,
+      "step": 19379
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1530670743359453e-06,
+      "loss": 0.5127,
+      "step": 19380
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1529715268488443e-06,
+      "loss": 0.4966,
+      "step": 19381
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.152875979361743e-06,
+      "loss": 0.6955,
+      "step": 19382
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.152780431874642e-06,
+      "loss": 1.3197,
+      "step": 19383
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1526848843875406e-06,
+      "loss": 0.7173,
+      "step": 19384
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1525893369004396e-06,
+      "loss": 0.7244,
+      "step": 19385
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.152493789413339e-06,
+      "loss": 0.8974,
+      "step": 19386
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1523982419262377e-06,
+      "loss": 0.9671,
+      "step": 19387
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1523026944391367e-06,
+      "loss": 0.773,
+      "step": 19388
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1522071469520353e-06,
+      "loss": 1.0105,
+      "step": 19389
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1521115994649344e-06,
+      "loss": 0.9004,
+      "step": 19390
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.152016051977833e-06,
+      "loss": 0.6481,
+      "step": 19391
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.151920504490732e-06,
+      "loss": 0.8342,
+      "step": 19392
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1518249570036315e-06,
+      "loss": 1.1869,
+      "step": 19393
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.15172940951653e-06,
+      "loss": 0.5214,
+      "step": 19394
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.151633862029429e-06,
+      "loss": 0.7366,
+      "step": 19395
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1515383145423278e-06,
+      "loss": 0.9876,
+      "step": 19396
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.151442767055227e-06,
+      "loss": 0.9905,
+      "step": 19397
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1513472195681254e-06,
+      "loss": 0.9431,
+      "step": 19398
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1512516720810245e-06,
+      "loss": 1.088,
+      "step": 19399
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.151156124593923e-06,
+      "loss": 1.1513,
+      "step": 19400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1510605771068225e-06,
+      "loss": 0.4569,
+      "step": 19401
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1509650296197216e-06,
+      "loss": 1.1316,
+      "step": 19402
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.15086948213262e-06,
+      "loss": 0.5274,
+      "step": 19403
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1507739346455192e-06,
+      "loss": 0.7486,
+      "step": 19404
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.150678387158418e-06,
+      "loss": 1.0427,
+      "step": 19405
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.150582839671317e-06,
+      "loss": 1.0359,
+      "step": 19406
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1504872921842155e-06,
+      "loss": 0.7176,
+      "step": 19407
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150391744697115e-06,
+      "loss": 0.5772,
+      "step": 19408
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.150296197210014e-06,
+      "loss": 1.2121,
+      "step": 19409
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1502006497229126e-06,
+      "loss": 1.0373,
+      "step": 19410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1501051022358116e-06,
+      "loss": 1.1563,
+      "step": 19411
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1500095547487103e-06,
+      "loss": 0.6798,
+      "step": 19412
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1499140072616093e-06,
+      "loss": 0.7543,
+      "step": 19413
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.149818459774508e-06,
+      "loss": 0.5766,
+      "step": 19414
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1497229122874074e-06,
+      "loss": 1.0118,
+      "step": 19415
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.149627364800306e-06,
+      "loss": 0.7129,
+      "step": 19416
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.149531817313205e-06,
+      "loss": 0.8072,
+      "step": 19417
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1494362698261036e-06,
+      "loss": 1.0071,
+      "step": 19418
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1493407223390027e-06,
+      "loss": 1.2538,
+      "step": 19419
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1492451748519013e-06,
+      "loss": 1.4697,
+      "step": 19420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1491496273648003e-06,
+      "loss": 0.9401,
+      "step": 19421
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1490540798777e-06,
+      "loss": 0.9122,
+      "step": 19422
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1489585323905984e-06,
+      "loss": 0.7338,
+      "step": 19423
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1488629849034975e-06,
+      "loss": 1.1547,
+      "step": 19424
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.148767437416396e-06,
+      "loss": 1.0777,
+      "step": 19425
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.148671889929295e-06,
+      "loss": 0.4796,
+      "step": 19426
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1485763424421937e-06,
+      "loss": 0.5515,
+      "step": 19427
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1484807949550928e-06,
+      "loss": 1.266,
+      "step": 19428
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1483852474679922e-06,
+      "loss": 0.9236,
+      "step": 19429
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.148289699980891e-06,
+      "loss": 0.7876,
+      "step": 19430
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.14819415249379e-06,
+      "loss": 1.1728,
+      "step": 19431
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1480986050066885e-06,
+      "loss": 0.6267,
+      "step": 19432
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1480030575195875e-06,
+      "loss": 0.591,
+      "step": 19433
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.147907510032486e-06,
+      "loss": 1.0137,
+      "step": 19434
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.147811962545385e-06,
+      "loss": 0.7402,
+      "step": 19435
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.147716415058284e-06,
+      "loss": 1.0537,
+      "step": 19436
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1476208675711833e-06,
+      "loss": 0.7591,
+      "step": 19437
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1475253200840823e-06,
+      "loss": 0.6225,
+      "step": 19438
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.147429772596981e-06,
+      "loss": 0.9541,
+      "step": 19439
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.14733422510988e-06,
+      "loss": 0.5986,
+      "step": 19440
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1472386776227786e-06,
+      "loss": 0.7853,
+      "step": 19441
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1471431301356776e-06,
+      "loss": 1.2041,
+      "step": 19442
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1470475826485762e-06,
+      "loss": 0.8055,
+      "step": 19443
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1469520351614757e-06,
+      "loss": 0.9525,
+      "step": 19444
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1468564876743747e-06,
+      "loss": 1.0573,
+      "step": 19445
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1467609401872733e-06,
+      "loss": 0.9836,
+      "step": 19446
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1466653927001724e-06,
+      "loss": 0.5982,
+      "step": 19447
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146569845213071e-06,
+      "loss": 0.6802,
+      "step": 19448
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.14647429772597e-06,
+      "loss": 0.926,
+      "step": 19449
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1463787502388686e-06,
+      "loss": 1.0252,
+      "step": 19450
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146283202751768e-06,
+      "loss": 0.5612,
+      "step": 19451
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146187655264667e-06,
+      "loss": 0.8106,
+      "step": 19452
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1460921077775658e-06,
+      "loss": 1.3826,
+      "step": 19453
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.145996560290465e-06,
+      "loss": 0.7429,
+      "step": 19454
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1459010128033634e-06,
+      "loss": 0.7581,
+      "step": 19455
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1458054653162625e-06,
+      "loss": 0.758,
+      "step": 19456
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.145709917829161e-06,
+      "loss": 1.2244,
+      "step": 19457
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1456143703420605e-06,
+      "loss": 0.6827,
+      "step": 19458
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.145518822854959e-06,
+      "loss": 0.73,
+      "step": 19459
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145423275367858e-06,
+      "loss": 1.1449,
+      "step": 19460
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1453277278807572e-06,
+      "loss": 0.8305,
+      "step": 19461
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145232180393656e-06,
+      "loss": 1.0669,
+      "step": 19462
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.145136632906555e-06,
+      "loss": 0.959,
+      "step": 19463
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1450410854194535e-06,
+      "loss": 1.3228,
+      "step": 19464
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144945537932353e-06,
+      "loss": 1.1673,
+      "step": 19465
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1448499904452516e-06,
+      "loss": 0.8439,
+      "step": 19466
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1447544429581506e-06,
+      "loss": 0.6947,
+      "step": 19467
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1446588954710492e-06,
+      "loss": 0.6464,
+      "step": 19468
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1445633479839483e-06,
+      "loss": 1.1634,
+      "step": 19469
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144467800496847e-06,
+      "loss": 0.9149,
+      "step": 19470
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144372253009746e-06,
+      "loss": 1.015,
+      "step": 19471
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144276705522645e-06,
+      "loss": 0.7099,
+      "step": 19472
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144181158035544e-06,
+      "loss": 0.9169,
+      "step": 19473
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.144085610548443e-06,
+      "loss": 0.9158,
+      "step": 19474
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1439900630613416e-06,
+      "loss": 0.7829,
+      "step": 19475
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1438945155742407e-06,
+      "loss": 1.3393,
+      "step": 19476
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1437989680871393e-06,
+      "loss": 0.6809,
+      "step": 19477
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1437034206000383e-06,
+      "loss": 0.8418,
+      "step": 19478
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.143607873112937e-06,
+      "loss": 1.0094,
+      "step": 19479
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1435123256258364e-06,
+      "loss": 0.7527,
+      "step": 19480
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1434167781387355e-06,
+      "loss": 0.586,
+      "step": 19481
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.143321230651634e-06,
+      "loss": 0.7046,
+      "step": 19482
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.143225683164533e-06,
+      "loss": 0.9221,
+      "step": 19483
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1431301356774317e-06,
+      "loss": 0.9053,
+      "step": 19484
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1430345881903308e-06,
+      "loss": 0.8217,
+      "step": 19485
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1429390407032294e-06,
+      "loss": 0.7706,
+      "step": 19486
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142843493216129e-06,
+      "loss": 1.1203,
+      "step": 19487
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142747945729028e-06,
+      "loss": 0.9607,
+      "step": 19488
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1426523982419265e-06,
+      "loss": 0.9093,
+      "step": 19489
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1425568507548255e-06,
+      "loss": 1.0185,
+      "step": 19490
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142461303267724e-06,
+      "loss": 0.6049,
+      "step": 19491
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142365755780623e-06,
+      "loss": 1.0019,
+      "step": 19492
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142270208293522e-06,
+      "loss": 1.253,
+      "step": 19493
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1421746608064213e-06,
+      "loss": 1.1102,
+      "step": 19494
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1420791133193203e-06,
+      "loss": 0.9021,
+      "step": 19495
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.141983565832219e-06,
+      "loss": 0.6717,
+      "step": 19496
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.141888018345118e-06,
+      "loss": 1.1386,
+      "step": 19497
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1417924708580166e-06,
+      "loss": 0.4975,
+      "step": 19498
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1416969233709156e-06,
+      "loss": 0.9973,
+      "step": 19499
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1416013758838142e-06,
+      "loss": 0.9136,
+      "step": 19500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1415058283967137e-06,
+      "loss": 0.4444,
+      "step": 19501
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1414102809096127e-06,
+      "loss": 0.985,
+      "step": 19502
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1413147334225113e-06,
+      "loss": 0.9398,
+      "step": 19503
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1412191859354104e-06,
+      "loss": 1.0007,
+      "step": 19504
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.141123638448309e-06,
+      "loss": 0.6223,
+      "step": 19505
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.141028090961208e-06,
+      "loss": 0.9497,
+      "step": 19506
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1409325434741066e-06,
+      "loss": 1.0581,
+      "step": 19507
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1408369959870057e-06,
+      "loss": 1.196,
+      "step": 19508
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1407414484999047e-06,
+      "loss": 0.797,
+      "step": 19509
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1406459010128038e-06,
+      "loss": 0.9248,
+      "step": 19510
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.140550353525703e-06,
+      "loss": 0.8628,
+      "step": 19511
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1404548060386014e-06,
+      "loss": 0.489,
+      "step": 19512
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1403592585515005e-06,
+      "loss": 0.6739,
+      "step": 19513
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140263711064399e-06,
+      "loss": 0.6835,
+      "step": 19514
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140168163577298e-06,
+      "loss": 1.0937,
+      "step": 19515
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.140072616090197e-06,
+      "loss": 0.8063,
+      "step": 19516
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139977068603096e-06,
+      "loss": 0.6116,
+      "step": 19517
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139881521115995e-06,
+      "loss": 0.7928,
+      "step": 19518
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139785973628894e-06,
+      "loss": 0.6118,
+      "step": 19519
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1396904261417925e-06,
+      "loss": 1.0866,
+      "step": 19520
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1395948786546915e-06,
+      "loss": 1.1333,
+      "step": 19521
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.13949933116759e-06,
+      "loss": 0.9262,
+      "step": 19522
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1394037836804896e-06,
+      "loss": 0.65,
+      "step": 19523
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1393082361933886e-06,
+      "loss": 1.08,
+      "step": 19524
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1392126887062872e-06,
+      "loss": 0.8858,
+      "step": 19525
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1391171412191863e-06,
+      "loss": 0.8991,
+      "step": 19526
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.139021593732085e-06,
+      "loss": 1.5234,
+      "step": 19527
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.138926046244984e-06,
+      "loss": 1.2215,
+      "step": 19528
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1388304987578825e-06,
+      "loss": 0.7529,
+      "step": 19529
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.138734951270782e-06,
+      "loss": 1.0896,
+      "step": 19530
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.138639403783681e-06,
+      "loss": 0.6868,
+      "step": 19531
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1385438562965796e-06,
+      "loss": 0.864,
+      "step": 19532
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1384483088094787e-06,
+      "loss": 0.87,
+      "step": 19533
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1383527613223773e-06,
+      "loss": 1.7221,
+      "step": 19534
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1382572138352763e-06,
+      "loss": 0.8881,
+      "step": 19535
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.138161666348175e-06,
+      "loss": 1.4623,
+      "step": 19536
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1380661188610744e-06,
+      "loss": 1.0798,
+      "step": 19537
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1379705713739735e-06,
+      "loss": 0.9929,
+      "step": 19538
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.137875023886872e-06,
+      "loss": 0.5167,
+      "step": 19539
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.137779476399771e-06,
+      "loss": 0.3917,
+      "step": 19540
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1376839289126697e-06,
+      "loss": 1.0626,
+      "step": 19541
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1375883814255688e-06,
+      "loss": 0.7035,
+      "step": 19542
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1374928339384674e-06,
+      "loss": 0.9275,
+      "step": 19543
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1373972864513664e-06,
+      "loss": 1.1444,
+      "step": 19544
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.137301738964266e-06,
+      "loss": 0.6933,
+      "step": 19545
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1372061914771645e-06,
+      "loss": 0.8795,
+      "step": 19546
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1371106439900635e-06,
+      "loss": 1.0656,
+      "step": 19547
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.137015096502962e-06,
+      "loss": 0.8659,
+      "step": 19548
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136919549015861e-06,
+      "loss": 0.7816,
+      "step": 19549
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.13682400152876e-06,
+      "loss": 0.6207,
+      "step": 19550
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136728454041659e-06,
+      "loss": 1.0121,
+      "step": 19551
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1366329065545583e-06,
+      "loss": 0.6912,
+      "step": 19552
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136537359067457e-06,
+      "loss": 0.639,
+      "step": 19553
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136441811580356e-06,
+      "loss": 0.6248,
+      "step": 19554
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1363462640932546e-06,
+      "loss": 0.6997,
+      "step": 19555
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1362507166061536e-06,
+      "loss": 0.8117,
+      "step": 19556
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1361551691190522e-06,
+      "loss": 1.1161,
+      "step": 19557
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1360596216319513e-06,
+      "loss": 0.7829,
+      "step": 19558
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1359640741448503e-06,
+      "loss": 0.6375,
+      "step": 19559
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1358685266577493e-06,
+      "loss": 0.974,
+      "step": 19560
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.135772979170648e-06,
+      "loss": 0.64,
+      "step": 19561
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.135677431683547e-06,
+      "loss": 0.7655,
+      "step": 19562
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.135581884196446e-06,
+      "loss": 0.9899,
+      "step": 19563
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1354863367093446e-06,
+      "loss": 0.678,
+      "step": 19564
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1353907892222437e-06,
+      "loss": 0.9705,
+      "step": 19565
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1352952417351427e-06,
+      "loss": 1.0882,
+      "step": 19566
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1351996942480418e-06,
+      "loss": 0.8684,
+      "step": 19567
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1351041467609404e-06,
+      "loss": 0.6797,
+      "step": 19568
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1350085992738394e-06,
+      "loss": 0.7025,
+      "step": 19569
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134913051786738e-06,
+      "loss": 1.1552,
+      "step": 19570
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134817504299637e-06,
+      "loss": 0.8005,
+      "step": 19571
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1347219568125357e-06,
+      "loss": 0.7349,
+      "step": 19572
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134626409325435e-06,
+      "loss": 0.5205,
+      "step": 19573
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134530861838334e-06,
+      "loss": 0.7415,
+      "step": 19574
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134435314351233e-06,
+      "loss": 0.5863,
+      "step": 19575
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134339766864132e-06,
+      "loss": 0.893,
+      "step": 19576
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1342442193770304e-06,
+      "loss": 0.9396,
+      "step": 19577
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1341486718899295e-06,
+      "loss": 1.1336,
+      "step": 19578
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134053124402828e-06,
+      "loss": 0.8189,
+      "step": 19579
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133957576915727e-06,
+      "loss": 0.8863,
+      "step": 19580
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1338620294286266e-06,
+      "loss": 0.6999,
+      "step": 19581
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1337664819415252e-06,
+      "loss": 0.5334,
+      "step": 19582
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1336709344544243e-06,
+      "loss": 0.8061,
+      "step": 19583
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133575386967323e-06,
+      "loss": 1.2416,
+      "step": 19584
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133479839480222e-06,
+      "loss": 1.3727,
+      "step": 19585
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1333842919931205e-06,
+      "loss": 0.553,
+      "step": 19586
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1332887445060196e-06,
+      "loss": 1.113,
+      "step": 19587
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.133193197018919e-06,
+      "loss": 0.4351,
+      "step": 19588
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1330976495318176e-06,
+      "loss": 0.676,
+      "step": 19589
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1330021020447167e-06,
+      "loss": 0.7274,
+      "step": 19590
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1329065545576153e-06,
+      "loss": 1.072,
+      "step": 19591
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1328110070705143e-06,
+      "loss": 0.6186,
+      "step": 19592
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132715459583413e-06,
+      "loss": 1.0804,
+      "step": 19593
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132619912096312e-06,
+      "loss": 1.237,
+      "step": 19594
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1325243646092114e-06,
+      "loss": 1.1481,
+      "step": 19595
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.13242881712211e-06,
+      "loss": 0.6436,
+      "step": 19596
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132333269635009e-06,
+      "loss": 0.9276,
+      "step": 19597
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1322377221479077e-06,
+      "loss": 0.9705,
+      "step": 19598
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1321421746608068e-06,
+      "loss": 0.9087,
+      "step": 19599
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1320466271737054e-06,
+      "loss": 0.5396,
+      "step": 19600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1319510796866044e-06,
+      "loss": 0.7021,
+      "step": 19601
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131855532199504e-06,
+      "loss": 1.1941,
+      "step": 19602
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1317599847124025e-06,
+      "loss": 1.0426,
+      "step": 19603
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1316644372253015e-06,
+      "loss": 0.803,
+      "step": 19604
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1315688897382e-06,
+      "loss": 0.6394,
+      "step": 19605
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131473342251099e-06,
+      "loss": 0.786,
+      "step": 19606
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131377794763998e-06,
+      "loss": 0.7464,
+      "step": 19607
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131282247276897e-06,
+      "loss": 1.2306,
+      "step": 19608
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131186699789796e-06,
+      "loss": 1.0601,
+      "step": 19609
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131091152302695e-06,
+      "loss": 0.9292,
+      "step": 19610
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1309956048155935e-06,
+      "loss": 0.8724,
+      "step": 19611
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1309000573284926e-06,
+      "loss": 1.0971,
+      "step": 19612
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1308045098413916e-06,
+      "loss": 1.0653,
+      "step": 19613
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1307089623542902e-06,
+      "loss": 1.007,
+      "step": 19614
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1306134148671893e-06,
+      "loss": 0.5711,
+      "step": 19615
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130517867380088e-06,
+      "loss": 1.0435,
+      "step": 19616
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1304223198929873e-06,
+      "loss": 0.7435,
+      "step": 19617
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130326772405886e-06,
+      "loss": 0.5341,
+      "step": 19618
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.130231224918785e-06,
+      "loss": 0.6022,
+      "step": 19619
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1301356774316836e-06,
+      "loss": 0.6371,
+      "step": 19620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1300401299445826e-06,
+      "loss": 0.921,
+      "step": 19621
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1299445824574813e-06,
+      "loss": 1.2139,
+      "step": 19622
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1298490349703803e-06,
+      "loss": 0.5947,
+      "step": 19623
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1297534874832798e-06,
+      "loss": 0.7634,
+      "step": 19624
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1296579399961784e-06,
+      "loss": 0.8311,
+      "step": 19625
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1295623925090774e-06,
+      "loss": 1.1593,
+      "step": 19626
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129466845021976e-06,
+      "loss": 0.5993,
+      "step": 19627
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129371297534875e-06,
+      "loss": 0.7747,
+      "step": 19628
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1292757500477737e-06,
+      "loss": 0.8076,
+      "step": 19629
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1291802025606727e-06,
+      "loss": 0.8537,
+      "step": 19630
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.129084655073572e-06,
+      "loss": 1.4792,
+      "step": 19631
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.128989107586471e-06,
+      "loss": 0.5847,
+      "step": 19632
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.12889356009937e-06,
+      "loss": 0.7711,
+      "step": 19633
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1287980126122684e-06,
+      "loss": 1.2408,
+      "step": 19634
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1287024651251675e-06,
+      "loss": 1.0073,
+      "step": 19635
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.128606917638066e-06,
+      "loss": 0.6508,
+      "step": 19636
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.128511370150965e-06,
+      "loss": 0.5651,
+      "step": 19637
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1284158226638646e-06,
+      "loss": 0.9254,
+      "step": 19638
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1283202751767632e-06,
+      "loss": 0.7709,
+      "step": 19639
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1282247276896623e-06,
+      "loss": 1.0391,
+      "step": 19640
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.128129180202561e-06,
+      "loss": 0.928,
+      "step": 19641
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.12803363271546e-06,
+      "loss": 0.4914,
+      "step": 19642
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1279380852283585e-06,
+      "loss": 0.5579,
+      "step": 19643
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1278425377412576e-06,
+      "loss": 1.1053,
+      "step": 19644
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.127746990254157e-06,
+      "loss": 0.9065,
+      "step": 19645
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1276514427670556e-06,
+      "loss": 0.814,
+      "step": 19646
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1275558952799547e-06,
+      "loss": 0.9417,
+      "step": 19647
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1274603477928533e-06,
+      "loss": 0.6832,
+      "step": 19648
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1273648003057523e-06,
+      "loss": 0.9028,
+      "step": 19649
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.127269252818651e-06,
+      "loss": 0.8853,
+      "step": 19650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.12717370533155e-06,
+      "loss": 1.0757,
+      "step": 19651
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1270781578444486e-06,
+      "loss": 0.7827,
+      "step": 19652
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126982610357348e-06,
+      "loss": 1.4715,
+      "step": 19653
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126887062870247e-06,
+      "loss": 0.8285,
+      "step": 19654
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1267915153831457e-06,
+      "loss": 0.9383,
+      "step": 19655
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1266959678960448e-06,
+      "loss": 0.8921,
+      "step": 19656
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1266004204089434e-06,
+      "loss": 0.8189,
+      "step": 19657
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1265048729218424e-06,
+      "loss": 1.057,
+      "step": 19658
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126409325434741e-06,
+      "loss": 1.2441,
+      "step": 19659
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1263137779476405e-06,
+      "loss": 1.3111,
+      "step": 19660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126218230460539e-06,
+      "loss": 0.8271,
+      "step": 19661
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126122682973438e-06,
+      "loss": 0.8485,
+      "step": 19662
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1260271354863368e-06,
+      "loss": 0.6898,
+      "step": 19663
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125931587999236e-06,
+      "loss": 1.0406,
+      "step": 19664
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125836040512135e-06,
+      "loss": 0.5252,
+      "step": 19665
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1257404930250334e-06,
+      "loss": 0.6769,
+      "step": 19666
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125644945537933e-06,
+      "loss": 0.8611,
+      "step": 19667
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1255493980508315e-06,
+      "loss": 0.6734,
+      "step": 19668
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1254538505637306e-06,
+      "loss": 0.9716,
+      "step": 19669
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125358303076629e-06,
+      "loss": 0.7584,
+      "step": 19670
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1252627555895282e-06,
+      "loss": 0.8665,
+      "step": 19671
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125167208102427e-06,
+      "loss": 0.5029,
+      "step": 19672
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.125071660615326e-06,
+      "loss": 1.3493,
+      "step": 19673
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1249761131282253e-06,
+      "loss": 1.2108,
+      "step": 19674
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.124880565641124e-06,
+      "loss": 1.001,
+      "step": 19675
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.124785018154023e-06,
+      "loss": 1.1698,
+      "step": 19676
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1246894706669216e-06,
+      "loss": 0.8256,
+      "step": 19677
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1245939231798206e-06,
+      "loss": 0.8431,
+      "step": 19678
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1244983756927193e-06,
+      "loss": 0.8727,
+      "step": 19679
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1244028282056183e-06,
+      "loss": 0.9176,
+      "step": 19680
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1243072807185178e-06,
+      "loss": 1.143,
+      "step": 19681
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1242117332314164e-06,
+      "loss": 1.1798,
+      "step": 19682
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1241161857443154e-06,
+      "loss": 1.0334,
+      "step": 19683
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.124020638257214e-06,
+      "loss": 0.5635,
+      "step": 19684
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123925090770113e-06,
+      "loss": 0.7914,
+      "step": 19685
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1238295432830117e-06,
+      "loss": 1.1059,
+      "step": 19686
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1237339957959107e-06,
+      "loss": 1.097,
+      "step": 19687
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1236384483088093e-06,
+      "loss": 0.9482,
+      "step": 19688
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123542900821709e-06,
+      "loss": 0.7298,
+      "step": 19689
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123447353334608e-06,
+      "loss": 0.8955,
+      "step": 19690
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1233518058475064e-06,
+      "loss": 0.7425,
+      "step": 19691
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1232562583604055e-06,
+      "loss": 0.8341,
+      "step": 19692
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123160710873304e-06,
+      "loss": 0.808,
+      "step": 19693
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123065163386203e-06,
+      "loss": 0.5998,
+      "step": 19694
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1229696158991018e-06,
+      "loss": 0.7004,
+      "step": 19695
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1228740684120012e-06,
+      "loss": 0.924,
+      "step": 19696
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1227785209249003e-06,
+      "loss": 0.7998,
+      "step": 19697
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.122682973437799e-06,
+      "loss": 0.9478,
+      "step": 19698
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.122587425950698e-06,
+      "loss": 0.6713,
+      "step": 19699
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1224918784635965e-06,
+      "loss": 1.0298,
+      "step": 19700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1223963309764956e-06,
+      "loss": 0.7471,
+      "step": 19701
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.122300783489394e-06,
+      "loss": 0.3775,
+      "step": 19702
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1222052360022936e-06,
+      "loss": 0.972,
+      "step": 19703
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1221096885151927e-06,
+      "loss": 0.8579,
+      "step": 19704
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1220141410280913e-06,
+      "loss": 0.9068,
+      "step": 19705
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1219185935409903e-06,
+      "loss": 0.4925,
+      "step": 19706
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121823046053889e-06,
+      "loss": 0.8225,
+      "step": 19707
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121727498566788e-06,
+      "loss": 2.1311,
+      "step": 19708
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1216319510796866e-06,
+      "loss": 0.5325,
+      "step": 19709
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121536403592586e-06,
+      "loss": 1.0966,
+      "step": 19710
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1214408561054847e-06,
+      "loss": 1.3858,
+      "step": 19711
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1213453086183837e-06,
+      "loss": 0.8498,
+      "step": 19712
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1212497611312823e-06,
+      "loss": 0.4455,
+      "step": 19713
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1211542136441814e-06,
+      "loss": 0.6454,
+      "step": 19714
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.12105866615708e-06,
+      "loss": 0.7529,
+      "step": 19715
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120963118669979e-06,
+      "loss": 1.3043,
+      "step": 19716
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1208675711828785e-06,
+      "loss": 1.1776,
+      "step": 19717
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120772023695777e-06,
+      "loss": 0.7222,
+      "step": 19718
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120676476208676e-06,
+      "loss": 0.6689,
+      "step": 19719
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1205809287215748e-06,
+      "loss": 0.9415,
+      "step": 19720
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120485381234474e-06,
+      "loss": 0.9084,
+      "step": 19721
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1203898337473724e-06,
+      "loss": 0.7487,
+      "step": 19722
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1202942862602714e-06,
+      "loss": 0.8767,
+      "step": 19723
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.12019873877317e-06,
+      "loss": 0.67,
+      "step": 19724
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1201031912860695e-06,
+      "loss": 0.7593,
+      "step": 19725
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1200076437989686e-06,
+      "loss": 0.9804,
+      "step": 19726
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119912096311867e-06,
+      "loss": 1.065,
+      "step": 19727
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1198165488247662e-06,
+      "loss": 0.9081,
+      "step": 19728
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119721001337665e-06,
+      "loss": 0.5625,
+      "step": 19729
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119625453850564e-06,
+      "loss": 0.9595,
+      "step": 19730
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1195299063634625e-06,
+      "loss": 0.8922,
+      "step": 19731
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119434358876362e-06,
+      "loss": 0.8305,
+      "step": 19732
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.119338811389261e-06,
+      "loss": 0.8493,
+      "step": 19733
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1192432639021596e-06,
+      "loss": 0.8465,
+      "step": 19734
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1191477164150586e-06,
+      "loss": 0.6017,
+      "step": 19735
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1190521689279573e-06,
+      "loss": 0.8856,
+      "step": 19736
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1189566214408563e-06,
+      "loss": 0.7959,
+      "step": 19737
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118861073953755e-06,
+      "loss": 0.7372,
+      "step": 19738
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1187655264666544e-06,
+      "loss": 0.6192,
+      "step": 19739
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1186699789795534e-06,
+      "loss": 0.7047,
+      "step": 19740
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118574431492452e-06,
+      "loss": 0.6693,
+      "step": 19741
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118478884005351e-06,
+      "loss": 1.0328,
+      "step": 19742
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1183833365182497e-06,
+      "loss": 0.6629,
+      "step": 19743
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1182877890311487e-06,
+      "loss": 1.4713,
+      "step": 19744
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1181922415440473e-06,
+      "loss": 1.0598,
+      "step": 19745
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118096694056947e-06,
+      "loss": 0.6239,
+      "step": 19746
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118001146569846e-06,
+      "loss": 0.8503,
+      "step": 19747
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1179055990827444e-06,
+      "loss": 0.897,
+      "step": 19748
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1178100515956435e-06,
+      "loss": 0.2998,
+      "step": 19749
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117714504108542e-06,
+      "loss": 0.8326,
+      "step": 19750
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117618956621441e-06,
+      "loss": 0.6103,
+      "step": 19751
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1175234091343398e-06,
+      "loss": 1.0602,
+      "step": 19752
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1174278616472392e-06,
+      "loss": 0.7982,
+      "step": 19753
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117332314160138e-06,
+      "loss": 0.6647,
+      "step": 19754
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117236766673037e-06,
+      "loss": 0.7358,
+      "step": 19755
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.117141219185936e-06,
+      "loss": 0.7845,
+      "step": 19756
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1170456716988345e-06,
+      "loss": 0.8737,
+      "step": 19757
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1169501242117336e-06,
+      "loss": 1.6619,
+      "step": 19758
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.116854576724632e-06,
+      "loss": 0.6924,
+      "step": 19759
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1167590292375312e-06,
+      "loss": 0.8974,
+      "step": 19760
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1166634817504303e-06,
+      "loss": 0.5902,
+      "step": 19761
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1165679342633293e-06,
+      "loss": 0.8952,
+      "step": 19762
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.116472386776228e-06,
+      "loss": 0.4696,
+      "step": 19763
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.116376839289127e-06,
+      "loss": 0.6751,
+      "step": 19764
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1162812918020256e-06,
+      "loss": 0.692,
+      "step": 19765
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1161857443149246e-06,
+      "loss": 0.7983,
+      "step": 19766
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1160901968278236e-06,
+      "loss": 0.8544,
+      "step": 19767
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1159946493407227e-06,
+      "loss": 0.9981,
+      "step": 19768
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1158991018536217e-06,
+      "loss": 1.0894,
+      "step": 19769
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1158035543665203e-06,
+      "loss": 0.6097,
+      "step": 19770
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1157080068794194e-06,
+      "loss": 0.9764,
+      "step": 19771
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.115612459392318e-06,
+      "loss": 1.0532,
+      "step": 19772
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.115516911905217e-06,
+      "loss": 1.2667,
+      "step": 19773
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1154213644181156e-06,
+      "loss": 1.1615,
+      "step": 19774
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115325816931015e-06,
+      "loss": 0.5648,
+      "step": 19775
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115230269443914e-06,
+      "loss": 1.2643,
+      "step": 19776
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1151347219568128e-06,
+      "loss": 0.7346,
+      "step": 19777
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.115039174469712e-06,
+      "loss": 1.1882,
+      "step": 19778
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1149436269826104e-06,
+      "loss": 0.6357,
+      "step": 19779
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1148480794955094e-06,
+      "loss": 1.0235,
+      "step": 19780
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114752532008408e-06,
+      "loss": 1.0116,
+      "step": 19781
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1146569845213075e-06,
+      "loss": 0.9215,
+      "step": 19782
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1145614370342066e-06,
+      "loss": 1.1018,
+      "step": 19783
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114465889547105e-06,
+      "loss": 0.9894,
+      "step": 19784
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1143703420600042e-06,
+      "loss": 0.7486,
+      "step": 19785
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114274794572903e-06,
+      "loss": 0.854,
+      "step": 19786
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.114179247085802e-06,
+      "loss": 1.1294,
+      "step": 19787
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1140836995987005e-06,
+      "loss": 0.9475,
+      "step": 19788
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1139881521116e-06,
+      "loss": 1.0902,
+      "step": 19789
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.113892604624499e-06,
+      "loss": 1.105,
+      "step": 19790
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1137970571373976e-06,
+      "loss": 0.5236,
+      "step": 19791
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1137015096502966e-06,
+      "loss": 0.7025,
+      "step": 19792
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1136059621631953e-06,
+      "loss": 1.3888,
+      "step": 19793
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1135104146760943e-06,
+      "loss": 0.4988,
+      "step": 19794
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.113414867188993e-06,
+      "loss": 0.9986,
+      "step": 19795
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.113319319701892e-06,
+      "loss": 1.3538,
+      "step": 19796
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1132237722147914e-06,
+      "loss": 0.6394,
+      "step": 19797
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11312822472769e-06,
+      "loss": 1.1311,
+      "step": 19798
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.113032677240589e-06,
+      "loss": 1.6707,
+      "step": 19799
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1129371297534877e-06,
+      "loss": 1.5623,
+      "step": 19800
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1128415822663867e-06,
+      "loss": 1.0674,
+      "step": 19801
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1127460347792853e-06,
+      "loss": 0.4042,
+      "step": 19802
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1126504872921844e-06,
+      "loss": 0.4506,
+      "step": 19803
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1125549398050834e-06,
+      "loss": 0.4581,
+      "step": 19804
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1124593923179824e-06,
+      "loss": 0.7786,
+      "step": 19805
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1123638448308815e-06,
+      "loss": 0.5401,
+      "step": 19806
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11226829734378e-06,
+      "loss": 0.9922,
+      "step": 19807
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.112172749856679e-06,
+      "loss": 0.6723,
+      "step": 19808
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1120772023695778e-06,
+      "loss": 0.6889,
+      "step": 19809
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111981654882477e-06,
+      "loss": 1.3629,
+      "step": 19810
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111886107395376e-06,
+      "loss": 0.9717,
+      "step": 19811
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111790559908275e-06,
+      "loss": 0.9149,
+      "step": 19812
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1116950124211735e-06,
+      "loss": 0.6812,
+      "step": 19813
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1115994649340725e-06,
+      "loss": 1.1787,
+      "step": 19814
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111503917446971e-06,
+      "loss": 0.9778,
+      "step": 19815
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.11140836995987e-06,
+      "loss": 1.1296,
+      "step": 19816
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111312822472769e-06,
+      "loss": 0.7269,
+      "step": 19817
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1112172749856683e-06,
+      "loss": 1.093,
+      "step": 19818
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1111217274985673e-06,
+      "loss": 1.1414,
+      "step": 19819
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.111026180011466e-06,
+      "loss": 0.6988,
+      "step": 19820
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.110930632524365e-06,
+      "loss": 0.7483,
+      "step": 19821
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1108350850372636e-06,
+      "loss": 0.7642,
+      "step": 19822
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1107395375501626e-06,
+      "loss": 1.2324,
+      "step": 19823
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1106439900630612e-06,
+      "loss": 1.364,
+      "step": 19824
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1105484425759607e-06,
+      "loss": 0.9366,
+      "step": 19825
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1104528950888597e-06,
+      "loss": 0.8938,
+      "step": 19826
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1103573476017583e-06,
+      "loss": 0.7808,
+      "step": 19827
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1102618001146574e-06,
+      "loss": 0.9363,
+      "step": 19828
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110166252627556e-06,
+      "loss": 0.6535,
+      "step": 19829
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.110070705140455e-06,
+      "loss": 1.2403,
+      "step": 19830
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1099751576533536e-06,
+      "loss": 0.7034,
+      "step": 19831
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1098796101662527e-06,
+      "loss": 1.2555,
+      "step": 19832
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109784062679152e-06,
+      "loss": 0.7274,
+      "step": 19833
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1096885151920508e-06,
+      "loss": 0.8757,
+      "step": 19834
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.10959296770495e-06,
+      "loss": 1.0882,
+      "step": 19835
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1094974202178484e-06,
+      "loss": 0.4252,
+      "step": 19836
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1094018727307474e-06,
+      "loss": 0.7481,
+      "step": 19837
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109306325243646e-06,
+      "loss": 0.6911,
+      "step": 19838
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109210777756545e-06,
+      "loss": 1.0232,
+      "step": 19839
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1091152302694446e-06,
+      "loss": 0.7969,
+      "step": 19840
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.109019682782343e-06,
+      "loss": 0.9845,
+      "step": 19841
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1089241352952422e-06,
+      "loss": 0.5994,
+      "step": 19842
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.108828587808141e-06,
+      "loss": 1.1387,
+      "step": 19843
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.10873304032104e-06,
+      "loss": 0.9427,
+      "step": 19844
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1086374928339385e-06,
+      "loss": 0.5745,
+      "step": 19845
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1085419453468375e-06,
+      "loss": 0.7799,
+      "step": 19846
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.108446397859737e-06,
+      "loss": 1.377,
+      "step": 19847
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1083508503726356e-06,
+      "loss": 0.8427,
+      "step": 19848
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1082553028855346e-06,
+      "loss": 0.6057,
+      "step": 19849
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1081597553984333e-06,
+      "loss": 0.8683,
+      "step": 19850
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1080642079113323e-06,
+      "loss": 1.1955,
+      "step": 19851
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.107968660424231e-06,
+      "loss": 0.592,
+      "step": 19852
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.10787311293713e-06,
+      "loss": 1.0026,
+      "step": 19853
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.107777565450029e-06,
+      "loss": 0.9458,
+      "step": 19854
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.107682017962928e-06,
+      "loss": 0.913,
+      "step": 19855
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1075864704758266e-06,
+      "loss": 1.1971,
+      "step": 19856
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1074909229887257e-06,
+      "loss": 1.2849,
+      "step": 19857
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1073953755016247e-06,
+      "loss": 1.0338,
+      "step": 19858
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1072998280145233e-06,
+      "loss": 0.7507,
+      "step": 19859
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1072042805274224e-06,
+      "loss": 1.2626,
+      "step": 19860
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1071087330403214e-06,
+      "loss": 0.6106,
+      "step": 19861
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1070131855532204e-06,
+      "loss": 0.6306,
+      "step": 19862
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106917638066119e-06,
+      "loss": 0.7925,
+      "step": 19863
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106822090579018e-06,
+      "loss": 0.878,
+      "step": 19864
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1067265430919167e-06,
+      "loss": 0.9261,
+      "step": 19865
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1066309956048158e-06,
+      "loss": 0.6235,
+      "step": 19866
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1065354481177144e-06,
+      "loss": 0.7091,
+      "step": 19867
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1064399006306134e-06,
+      "loss": 1.0195,
+      "step": 19868
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106344353143513e-06,
+      "loss": 0.9388,
+      "step": 19869
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1062488056564115e-06,
+      "loss": 0.5898,
+      "step": 19870
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1061532581693105e-06,
+      "loss": 0.5861,
+      "step": 19871
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106057710682209e-06,
+      "loss": 0.8734,
+      "step": 19872
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105962163195108e-06,
+      "loss": 0.8867,
+      "step": 19873
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105866615708007e-06,
+      "loss": 0.8839,
+      "step": 19874
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105771068220906e-06,
+      "loss": 0.7016,
+      "step": 19875
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1056755207338053e-06,
+      "loss": 1.1631,
+      "step": 19876
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105579973246704e-06,
+      "loss": 0.9678,
+      "step": 19877
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.105484425759603e-06,
+      "loss": 0.5912,
+      "step": 19878
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1053888782725016e-06,
+      "loss": 1.2278,
+      "step": 19879
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1052933307854006e-06,
+      "loss": 1.0934,
+      "step": 19880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1051977832982992e-06,
+      "loss": 0.9286,
+      "step": 19881
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1051022358111983e-06,
+      "loss": 0.7702,
+      "step": 19882
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1050066883240977e-06,
+      "loss": 0.8064,
+      "step": 19883
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1049111408369963e-06,
+      "loss": 1.0123,
+      "step": 19884
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1048155933498954e-06,
+      "loss": 0.8578,
+      "step": 19885
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.104720045862794e-06,
+      "loss": 1.2675,
+      "step": 19886
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.104624498375693e-06,
+      "loss": 0.9147,
+      "step": 19887
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1045289508885916e-06,
+      "loss": 0.9804,
+      "step": 19888
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1044334034014907e-06,
+      "loss": 0.9261,
+      "step": 19889
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.10433785591439e-06,
+      "loss": 1.1698,
+      "step": 19890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1042423084272888e-06,
+      "loss": 1.0124,
+      "step": 19891
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.104146760940188e-06,
+      "loss": 0.709,
+      "step": 19892
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1040512134530864e-06,
+      "loss": 0.8578,
+      "step": 19893
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1039556659659854e-06,
+      "loss": 0.5477,
+      "step": 19894
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103860118478884e-06,
+      "loss": 1.307,
+      "step": 19895
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103764570991783e-06,
+      "loss": 1.1883,
+      "step": 19896
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1036690235046826e-06,
+      "loss": 0.7414,
+      "step": 19897
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103573476017581e-06,
+      "loss": 0.6622,
+      "step": 19898
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1034779285304802e-06,
+      "loss": 1.1489,
+      "step": 19899
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103382381043379e-06,
+      "loss": 0.6989,
+      "step": 19900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103286833556278e-06,
+      "loss": 0.7102,
+      "step": 19901
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1031912860691765e-06,
+      "loss": 0.795,
+      "step": 19902
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1030957385820755e-06,
+      "loss": 1.0211,
+      "step": 19903
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.103000191094974e-06,
+      "loss": 0.8682,
+      "step": 19904
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1029046436078736e-06,
+      "loss": 0.7257,
+      "step": 19905
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1028090961207722e-06,
+      "loss": 0.8036,
+      "step": 19906
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1027135486336713e-06,
+      "loss": 1.0575,
+      "step": 19907
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1026180011465703e-06,
+      "loss": 0.4763,
+      "step": 19908
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.102522453659469e-06,
+      "loss": 0.8339,
+      "step": 19909
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.102426906172368e-06,
+      "loss": 0.5065,
+      "step": 19910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1023313586852666e-06,
+      "loss": 1.079,
+      "step": 19911
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.102235811198166e-06,
+      "loss": 1.1286,
+      "step": 19912
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1021402637110646e-06,
+      "loss": 0.707,
+      "step": 19913
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1020447162239637e-06,
+      "loss": 1.0862,
+      "step": 19914
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1019491687368623e-06,
+      "loss": 1.2736,
+      "step": 19915
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1018536212497613e-06,
+      "loss": 0.7968,
+      "step": 19916
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.10175807376266e-06,
+      "loss": 0.8181,
+      "step": 19917
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.101662526275559e-06,
+      "loss": 0.5157,
+      "step": 19918
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1015669787884584e-06,
+      "loss": 0.7394,
+      "step": 19919
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.101471431301357e-06,
+      "loss": 0.9359,
+      "step": 19920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.101375883814256e-06,
+      "loss": 0.8641,
+      "step": 19921
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1012803363271547e-06,
+      "loss": 1.1311,
+      "step": 19922
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1011847888400538e-06,
+      "loss": 1.4222,
+      "step": 19923
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1010892413529524e-06,
+      "loss": 0.6694,
+      "step": 19924
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1009936938658514e-06,
+      "loss": 1.1522,
+      "step": 19925
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.100898146378751e-06,
+      "loss": 0.7793,
+      "step": 19926
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1008025988916495e-06,
+      "loss": 1.3063,
+      "step": 19927
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1007070514045485e-06,
+      "loss": 1.1648,
+      "step": 19928
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.100611503917447e-06,
+      "loss": 0.5916,
+      "step": 19929
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.100515956430346e-06,
+      "loss": 0.9452,
+      "step": 19930
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100420408943245e-06,
+      "loss": 0.5494,
+      "step": 19931
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100324861456144e-06,
+      "loss": 1.3678,
+      "step": 19932
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.1002293139690433e-06,
+      "loss": 0.7775,
+      "step": 19933
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100133766481942e-06,
+      "loss": 0.9254,
+      "step": 19934
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.100038218994841e-06,
+      "loss": 0.9646,
+      "step": 19935
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0999426715077396e-06,
+      "loss": 0.8164,
+      "step": 19936
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0998471240206386e-06,
+      "loss": 0.8087,
+      "step": 19937
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0997515765335372e-06,
+      "loss": 1.1108,
+      "step": 19938
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0996560290464363e-06,
+      "loss": 0.8269,
+      "step": 19939
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099560481559335e-06,
+      "loss": 1.1901,
+      "step": 19940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0994649340722343e-06,
+      "loss": 1.1464,
+      "step": 19941
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0993693865851334e-06,
+      "loss": 1.1671,
+      "step": 19942
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099273839098032e-06,
+      "loss": 1.0025,
+      "step": 19943
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.099178291610931e-06,
+      "loss": 1.0212,
+      "step": 19944
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0990827441238296e-06,
+      "loss": 0.8323,
+      "step": 19945
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0989871966367287e-06,
+      "loss": 0.937,
+      "step": 19946
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0988916491496273e-06,
+      "loss": 0.5342,
+      "step": 19947
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0987961016625268e-06,
+      "loss": 0.7398,
+      "step": 19948
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098700554175426e-06,
+      "loss": 0.666,
+      "step": 19949
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0986050066883244e-06,
+      "loss": 0.6061,
+      "step": 19950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0985094592012234e-06,
+      "loss": 1.0193,
+      "step": 19951
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098413911714122e-06,
+      "loss": 1.4938,
+      "step": 19952
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098318364227021e-06,
+      "loss": 1.2612,
+      "step": 19953
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0982228167399197e-06,
+      "loss": 0.9958,
+      "step": 19954
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098127269252819e-06,
+      "loss": 0.7242,
+      "step": 19955
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098031721765718e-06,
+      "loss": 0.5263,
+      "step": 19956
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.097936174278617e-06,
+      "loss": 1.1497,
+      "step": 19957
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0978406267915154e-06,
+      "loss": 0.6151,
+      "step": 19958
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0977450793044145e-06,
+      "loss": 0.6931,
+      "step": 19959
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0976495318173135e-06,
+      "loss": 0.6279,
+      "step": 19960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.097553984330212e-06,
+      "loss": 0.7496,
+      "step": 19961
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0974584368431116e-06,
+      "loss": 0.7351,
+      "step": 19962
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.09736288935601e-06,
+      "loss": 0.8979,
+      "step": 19963
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0972673418689093e-06,
+      "loss": 1.2293,
+      "step": 19964
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.097171794381808e-06,
+      "loss": 1.0278,
+      "step": 19965
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.097076246894707e-06,
+      "loss": 0.8433,
+      "step": 19966
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0969806994076055e-06,
+      "loss": 0.9587,
+      "step": 19967
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0968851519205046e-06,
+      "loss": 1.0095,
+      "step": 19968
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096789604433404e-06,
+      "loss": 0.6275,
+      "step": 19969
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0966940569463026e-06,
+      "loss": 1.0431,
+      "step": 19970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0965985094592017e-06,
+      "loss": 0.7307,
+      "step": 19971
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0965029619721003e-06,
+      "loss": 0.9315,
+      "step": 19972
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0964074144849993e-06,
+      "loss": 0.8039,
+      "step": 19973
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096311866997898e-06,
+      "loss": 0.9224,
+      "step": 19974
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096216319510797e-06,
+      "loss": 0.7891,
+      "step": 19975
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0961207720236956e-06,
+      "loss": 0.8812,
+      "step": 19976
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096025224536595e-06,
+      "loss": 0.7798,
+      "step": 19977
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.095929677049494e-06,
+      "loss": 1.0751,
+      "step": 19978
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0958341295623927e-06,
+      "loss": 0.7799,
+      "step": 19979
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0957385820752917e-06,
+      "loss": 0.8275,
+      "step": 19980
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0956430345881904e-06,
+      "loss": 1.0908,
+      "step": 19981
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0955474871010894e-06,
+      "loss": 0.7518,
+      "step": 19982
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095451939613988e-06,
+      "loss": 0.5248,
+      "step": 19983
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0953563921268875e-06,
+      "loss": 0.7861,
+      "step": 19984
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0952608446397865e-06,
+      "loss": 1.2195,
+      "step": 19985
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095165297152685e-06,
+      "loss": 1.1808,
+      "step": 19986
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.095069749665584e-06,
+      "loss": 0.6639,
+      "step": 19987
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0949742021784828e-06,
+      "loss": 0.7711,
+      "step": 19988
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094878654691382e-06,
+      "loss": 0.8744,
+      "step": 19989
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0947831072042804e-06,
+      "loss": 1.8402,
+      "step": 19990
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09468755971718e-06,
+      "loss": 1.2049,
+      "step": 19991
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094592012230079e-06,
+      "loss": 0.7893,
+      "step": 19992
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0944964647429776e-06,
+      "loss": 0.6862,
+      "step": 19993
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0944009172558766e-06,
+      "loss": 1.0731,
+      "step": 19994
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094305369768775e-06,
+      "loss": 0.9344,
+      "step": 19995
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0942098222816742e-06,
+      "loss": 0.9415,
+      "step": 19996
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.094114274794573e-06,
+      "loss": 0.9316,
+      "step": 19997
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0940187273074723e-06,
+      "loss": 1.0098,
+      "step": 19998
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0939231798203714e-06,
+      "loss": 1.2761,
+      "step": 19999
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09382763233327e-06,
+      "loss": 1.1269,
+      "step": 20000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.093732084846169e-06,
+      "loss": 0.9738,
+      "step": 20001
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0936365373590676e-06,
+      "loss": 0.7495,
+      "step": 20002
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0935409898719667e-06,
+      "loss": 1.0443,
+      "step": 20003
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0934454423848653e-06,
+      "loss": 0.9452,
+      "step": 20004
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0933498948977647e-06,
+      "loss": 0.8339,
+      "step": 20005
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0932543474106634e-06,
+      "loss": 0.64,
+      "step": 20006
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0931587999235624e-06,
+      "loss": 0.7445,
+      "step": 20007
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.093063252436461e-06,
+      "loss": 0.9235,
+      "step": 20008
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09296770494936e-06,
+      "loss": 0.9336,
+      "step": 20009
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.092872157462259e-06,
+      "loss": 0.8447,
+      "step": 20010
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0927766099751577e-06,
+      "loss": 0.743,
+      "step": 20011
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0926810624880567e-06,
+      "loss": 1.227,
+      "step": 20012
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0925855150009558e-06,
+      "loss": 0.6376,
+      "step": 20013
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.092489967513855e-06,
+      "loss": 0.7466,
+      "step": 20014
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0923944200267534e-06,
+      "loss": 0.6428,
+      "step": 20015
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0922988725396525e-06,
+      "loss": 0.9612,
+      "step": 20016
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.092203325052551e-06,
+      "loss": 0.8668,
+      "step": 20017
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.09210777756545e-06,
+      "loss": 1.0882,
+      "step": 20018
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0920122300783487e-06,
+      "loss": 1.2211,
+      "step": 20019
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091916682591248e-06,
+      "loss": 1.0268,
+      "step": 20020
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0918211351041472e-06,
+      "loss": 0.8957,
+      "step": 20021
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091725587617046e-06,
+      "loss": 0.7729,
+      "step": 20022
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091630040129945e-06,
+      "loss": 0.8441,
+      "step": 20023
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0915344926428435e-06,
+      "loss": 0.6385,
+      "step": 20024
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0914389451557426e-06,
+      "loss": 0.7448,
+      "step": 20025
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091343397668641e-06,
+      "loss": 0.5102,
+      "step": 20026
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0912478501815406e-06,
+      "loss": 1.1275,
+      "step": 20027
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0911523026944397e-06,
+      "loss": 1.0848,
+      "step": 20028
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0910567552073383e-06,
+      "loss": 0.8909,
+      "step": 20029
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0909612077202373e-06,
+      "loss": 0.639,
+      "step": 20030
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.090865660233136e-06,
+      "loss": 1.0019,
+      "step": 20031
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.090770112746035e-06,
+      "loss": 0.8406,
+      "step": 20032
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0906745652589336e-06,
+      "loss": 0.6278,
+      "step": 20033
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.090579017771833e-06,
+      "loss": 0.66,
+      "step": 20034
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.090483470284732e-06,
+      "loss": 0.6414,
+      "step": 20035
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0903879227976307e-06,
+      "loss": 0.8069,
+      "step": 20036
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0902923753105297e-06,
+      "loss": 1.0151,
+      "step": 20037
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0901968278234284e-06,
+      "loss": 0.7178,
+      "step": 20038
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0901012803363274e-06,
+      "loss": 0.6912,
+      "step": 20039
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.090005732849226e-06,
+      "loss": 1.119,
+      "step": 20040
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0899101853621255e-06,
+      "loss": 0.7935,
+      "step": 20041
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0898146378750245e-06,
+      "loss": 0.5006,
+      "step": 20042
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.089719090387923e-06,
+      "loss": 0.9913,
+      "step": 20043
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.089623542900822e-06,
+      "loss": 0.5862,
+      "step": 20044
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0895279954137208e-06,
+      "loss": 0.929,
+      "step": 20045
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.08943244792662e-06,
+      "loss": 0.717,
+      "step": 20046
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0893369004395184e-06,
+      "loss": 0.9309,
+      "step": 20047
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0892413529524175e-06,
+      "loss": 0.7762,
+      "step": 20048
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0891458054653165e-06,
+      "loss": 0.4165,
+      "step": 20049
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0890502579782156e-06,
+      "loss": 0.6744,
+      "step": 20050
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0889547104911146e-06,
+      "loss": 1.0278,
+      "step": 20051
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.088859163004013e-06,
+      "loss": 0.4028,
+      "step": 20052
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0887636155169122e-06,
+      "loss": 0.9038,
+      "step": 20053
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.088668068029811e-06,
+      "loss": 0.8411,
+      "step": 20054
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.08857252054271e-06,
+      "loss": 0.6441,
+      "step": 20055
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.088476973055609e-06,
+      "loss": 1.1236,
+      "step": 20056
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.088381425568508e-06,
+      "loss": 0.8938,
+      "step": 20057
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0882858780814066e-06,
+      "loss": 1.0566,
+      "step": 20058
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0881903305943056e-06,
+      "loss": 0.6533,
+      "step": 20059
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0880947831072042e-06,
+      "loss": 0.7788,
+      "step": 20060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0879992356201033e-06,
+      "loss": 0.5075,
+      "step": 20061
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0879036881330023e-06,
+      "loss": 0.8113,
+      "step": 20062
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0878081406459014e-06,
+      "loss": 1.0899,
+      "step": 20063
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0877125931588004e-06,
+      "loss": 0.7513,
+      "step": 20064
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.087617045671699e-06,
+      "loss": 0.7364,
+      "step": 20065
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.087521498184598e-06,
+      "loss": 0.9384,
+      "step": 20066
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0874259506974967e-06,
+      "loss": 1.2527,
+      "step": 20067
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0873304032103957e-06,
+      "loss": 0.5912,
+      "step": 20068
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0872348557232943e-06,
+      "loss": 1.3299,
+      "step": 20069
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0871393082361938e-06,
+      "loss": 1.1387,
+      "step": 20070
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.087043760749093e-06,
+      "loss": 0.7416,
+      "step": 20071
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0869482132619914e-06,
+      "loss": 0.6028,
+      "step": 20072
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0868526657748905e-06,
+      "loss": 1.2337,
+      "step": 20073
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086757118287789e-06,
+      "loss": 0.7365,
+      "step": 20074
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086661570800688e-06,
+      "loss": 1.6735,
+      "step": 20075
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0865660233135867e-06,
+      "loss": 0.8365,
+      "step": 20076
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086470475826486e-06,
+      "loss": 0.8233,
+      "step": 20077
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0863749283393852e-06,
+      "loss": 0.6761,
+      "step": 20078
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086279380852284e-06,
+      "loss": 0.618,
+      "step": 20079
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086183833365183e-06,
+      "loss": 0.7885,
+      "step": 20080
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0860882858780815e-06,
+      "loss": 1.0717,
+      "step": 20081
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0859927383909806e-06,
+      "loss": 0.7552,
+      "step": 20082
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.085897190903879e-06,
+      "loss": 0.3454,
+      "step": 20083
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.085801643416778e-06,
+      "loss": 0.9903,
+      "step": 20084
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0857060959296777e-06,
+      "loss": 0.5314,
+      "step": 20085
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0856105484425763e-06,
+      "loss": 1.2138,
+      "step": 20086
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0855150009554753e-06,
+      "loss": 0.6764,
+      "step": 20087
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.085419453468374e-06,
+      "loss": 0.9816,
+      "step": 20088
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.085323905981273e-06,
+      "loss": 1.3284,
+      "step": 20089
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0852283584941716e-06,
+      "loss": 1.4054,
+      "step": 20090
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0851328110070706e-06,
+      "loss": 0.7062,
+      "step": 20091
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08503726351997e-06,
+      "loss": 0.834,
+      "step": 20092
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0849417160328687e-06,
+      "loss": 0.6024,
+      "step": 20093
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0848461685457677e-06,
+      "loss": 0.594,
+      "step": 20094
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0847506210586664e-06,
+      "loss": 1.0519,
+      "step": 20095
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0846550735715654e-06,
+      "loss": 0.612,
+      "step": 20096
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084559526084464e-06,
+      "loss": 0.9197,
+      "step": 20097
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084463978597363e-06,
+      "loss": 0.596,
+      "step": 20098
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084368431110262e-06,
+      "loss": 0.6393,
+      "step": 20099
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.084272883623161e-06,
+      "loss": 0.5308,
+      "step": 20100
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08417733613606e-06,
+      "loss": 0.7861,
+      "step": 20101
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0840817886489588e-06,
+      "loss": 0.8708,
+      "step": 20102
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.083986241161858e-06,
+      "loss": 0.9261,
+      "step": 20103
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0838906936747564e-06,
+      "loss": 1.1993,
+      "step": 20104
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0837951461876555e-06,
+      "loss": 1.2798,
+      "step": 20105
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0836995987005545e-06,
+      "loss": 1.341,
+      "step": 20106
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0836040512134536e-06,
+      "loss": 1.0761,
+      "step": 20107
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.083508503726352e-06,
+      "loss": 0.8294,
+      "step": 20108
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.083412956239251e-06,
+      "loss": 0.8469,
+      "step": 20109
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08331740875215e-06,
+      "loss": 1.0206,
+      "step": 20110
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.083221861265049e-06,
+      "loss": 0.8534,
+      "step": 20111
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0831263137779475e-06,
+      "loss": 0.5539,
+      "step": 20112
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.083030766290847e-06,
+      "loss": 1.0579,
+      "step": 20113
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.082935218803746e-06,
+      "loss": 0.6639,
+      "step": 20114
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0828396713166446e-06,
+      "loss": 0.8087,
+      "step": 20115
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0827441238295436e-06,
+      "loss": 1.3058,
+      "step": 20116
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0826485763424422e-06,
+      "loss": 0.8425,
+      "step": 20117
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0825530288553413e-06,
+      "loss": 0.6564,
+      "step": 20118
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.08245748136824e-06,
+      "loss": 1.0091,
+      "step": 20119
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.082361933881139e-06,
+      "loss": 0.8153,
+      "step": 20120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0822663863940384e-06,
+      "loss": 0.6516,
+      "step": 20121
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.082170838906937e-06,
+      "loss": 0.5702,
+      "step": 20122
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.082075291419836e-06,
+      "loss": 0.536,
+      "step": 20123
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0819797439327347e-06,
+      "loss": 0.731,
+      "step": 20124
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0818841964456337e-06,
+      "loss": 1.3311,
+      "step": 20125
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0817886489585323e-06,
+      "loss": 0.8703,
+      "step": 20126
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0816931014714314e-06,
+      "loss": 0.6449,
+      "step": 20127
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.081597553984331e-06,
+      "loss": 0.9713,
+      "step": 20128
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0815020064972294e-06,
+      "loss": 0.428,
+      "step": 20129
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0814064590101285e-06,
+      "loss": 1.4413,
+      "step": 20130
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.081310911523027e-06,
+      "loss": 1.0576,
+      "step": 20131
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.081215364035926e-06,
+      "loss": 0.8986,
+      "step": 20132
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0811198165488247e-06,
+      "loss": 1.0052,
+      "step": 20133
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0810242690617238e-06,
+      "loss": 0.7844,
+      "step": 20134
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0809287215746232e-06,
+      "loss": 1.2726,
+      "step": 20135
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080833174087522e-06,
+      "loss": 0.3873,
+      "step": 20136
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080737626600421e-06,
+      "loss": 0.7099,
+      "step": 20137
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0806420791133195e-06,
+      "loss": 0.9124,
+      "step": 20138
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0805465316262186e-06,
+      "loss": 1.0398,
+      "step": 20139
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080450984139117e-06,
+      "loss": 1.0494,
+      "step": 20140
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.080355436652016e-06,
+      "loss": 0.7273,
+      "step": 20141
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0802598891649157e-06,
+      "loss": 0.8947,
+      "step": 20142
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0801643416778143e-06,
+      "loss": 0.5513,
+      "step": 20143
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0800687941907133e-06,
+      "loss": 0.9892,
+      "step": 20144
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079973246703612e-06,
+      "loss": 1.4736,
+      "step": 20145
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079877699216511e-06,
+      "loss": 0.8228,
+      "step": 20146
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0797821517294096e-06,
+      "loss": 0.4356,
+      "step": 20147
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0796866042423086e-06,
+      "loss": 1.1741,
+      "step": 20148
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0795910567552077e-06,
+      "loss": 0.6711,
+      "step": 20149
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0794955092681067e-06,
+      "loss": 1.0781,
+      "step": 20150
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0793999617810053e-06,
+      "loss": 0.7204,
+      "step": 20151
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0793044142939044e-06,
+      "loss": 0.9044,
+      "step": 20152
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0792088668068034e-06,
+      "loss": 0.5847,
+      "step": 20153
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079113319319702e-06,
+      "loss": 0.6075,
+      "step": 20154
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.079017771832601e-06,
+      "loss": 1.2006,
+      "step": 20155
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0789222243454997e-06,
+      "loss": 0.9688,
+      "step": 20156
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.078826676858399e-06,
+      "loss": 0.6166,
+      "step": 20157
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0787311293712977e-06,
+      "loss": 0.6372,
+      "step": 20158
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0786355818841968e-06,
+      "loss": 1.0608,
+      "step": 20159
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0785400343970954e-06,
+      "loss": 0.9792,
+      "step": 20160
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0784444869099944e-06,
+      "loss": 0.5894,
+      "step": 20161
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.078348939422893e-06,
+      "loss": 0.8741,
+      "step": 20162
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.078253391935792e-06,
+      "loss": 0.9006,
+      "step": 20163
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0781578444486916e-06,
+      "loss": 0.7817,
+      "step": 20164
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.07806229696159e-06,
+      "loss": 0.7119,
+      "step": 20165
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077966749474489e-06,
+      "loss": 0.9098,
+      "step": 20166
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077871201987388e-06,
+      "loss": 1.2739,
+      "step": 20167
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077775654500287e-06,
+      "loss": 0.7155,
+      "step": 20168
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0776801070131855e-06,
+      "loss": 0.9007,
+      "step": 20169
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0775845595260845e-06,
+      "loss": 0.7201,
+      "step": 20170
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077489012038984e-06,
+      "loss": 1.3361,
+      "step": 20171
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0773934645518826e-06,
+      "loss": 0.8607,
+      "step": 20172
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0772979170647816e-06,
+      "loss": 1.3376,
+      "step": 20173
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0772023695776802e-06,
+      "loss": 1.1425,
+      "step": 20174
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0771068220905793e-06,
+      "loss": 0.457,
+      "step": 20175
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077011274603478e-06,
+      "loss": 0.9788,
+      "step": 20176
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076915727116377e-06,
+      "loss": 0.7465,
+      "step": 20177
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0768201796292764e-06,
+      "loss": 0.7759,
+      "step": 20178
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076724632142175e-06,
+      "loss": 0.4201,
+      "step": 20179
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076629084655074e-06,
+      "loss": 0.7114,
+      "step": 20180
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0765335371679727e-06,
+      "loss": 0.8854,
+      "step": 20181
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0764379896808717e-06,
+      "loss": 0.773,
+      "step": 20182
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0763424421937703e-06,
+      "loss": 0.8602,
+      "step": 20183
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0762468947066694e-06,
+      "loss": 0.8864,
+      "step": 20184
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076151347219569e-06,
+      "loss": 0.9696,
+      "step": 20185
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0760557997324674e-06,
+      "loss": 1.0236,
+      "step": 20186
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0759602522453665e-06,
+      "loss": 0.8927,
+      "step": 20187
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.075864704758265e-06,
+      "loss": 0.8728,
+      "step": 20188
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.075769157271164e-06,
+      "loss": 0.7393,
+      "step": 20189
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0756736097840627e-06,
+      "loss": 1.3153,
+      "step": 20190
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0755780622969618e-06,
+      "loss": 1.3324,
+      "step": 20191
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0754825148098604e-06,
+      "loss": 1.0077,
+      "step": 20192
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07538696732276e-06,
+      "loss": 0.7218,
+      "step": 20193
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075291419835659e-06,
+      "loss": 0.9019,
+      "step": 20194
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0751958723485575e-06,
+      "loss": 0.9312,
+      "step": 20195
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0751003248614566e-06,
+      "loss": 0.8883,
+      "step": 20196
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.075004777374355e-06,
+      "loss": 1.3286,
+      "step": 20197
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074909229887254e-06,
+      "loss": 0.7875,
+      "step": 20198
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074813682400153e-06,
+      "loss": 0.7774,
+      "step": 20199
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0747181349130523e-06,
+      "loss": 1.2893,
+      "step": 20200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074622587425951e-06,
+      "loss": 1.0468,
+      "step": 20201
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07452703993885e-06,
+      "loss": 0.6473,
+      "step": 20202
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.074431492451749e-06,
+      "loss": 0.6877,
+      "step": 20203
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0743359449646476e-06,
+      "loss": 0.856,
+      "step": 20204
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0742403974775466e-06,
+      "loss": 1.1762,
+      "step": 20205
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0741448499904452e-06,
+      "loss": 1.3968,
+      "step": 20206
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0740493025033447e-06,
+      "loss": 0.341,
+      "step": 20207
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0739537550162433e-06,
+      "loss": 1.1692,
+      "step": 20208
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0738582075291424e-06,
+      "loss": 1.0057,
+      "step": 20209
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.073762660042041e-06,
+      "loss": 0.7246,
+      "step": 20210
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07366711255494e-06,
+      "loss": 1.4267,
+      "step": 20211
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0735715650678386e-06,
+      "loss": 0.7724,
+      "step": 20212
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0734760175807377e-06,
+      "loss": 1.0049,
+      "step": 20213
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.073380470093637e-06,
+      "loss": 1.0188,
+      "step": 20214
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0732849226065357e-06,
+      "loss": 0.923,
+      "step": 20215
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0731893751194348e-06,
+      "loss": 1.2267,
+      "step": 20216
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0730938276323334e-06,
+      "loss": 0.9684,
+      "step": 20217
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0729982801452324e-06,
+      "loss": 0.8328,
+      "step": 20218
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072902732658131e-06,
+      "loss": 0.9521,
+      "step": 20219
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.07280718517103e-06,
+      "loss": 0.7217,
+      "step": 20220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0727116376839296e-06,
+      "loss": 0.7225,
+      "step": 20221
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072616090196828e-06,
+      "loss": 1.2072,
+      "step": 20222
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072520542709727e-06,
+      "loss": 0.7984,
+      "step": 20223
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072424995222626e-06,
+      "loss": 0.855,
+      "step": 20224
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072329447735525e-06,
+      "loss": 0.7338,
+      "step": 20225
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0722339002484235e-06,
+      "loss": 0.6098,
+      "step": 20226
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0721383527613225e-06,
+      "loss": 0.7539,
+      "step": 20227
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072042805274221e-06,
+      "loss": 1.0856,
+      "step": 20228
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0719472577871206e-06,
+      "loss": 0.5815,
+      "step": 20229
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0718517103000196e-06,
+      "loss": 1.1524,
+      "step": 20230
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0717561628129182e-06,
+      "loss": 0.8714,
+      "step": 20231
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0716606153258173e-06,
+      "loss": 1.094,
+      "step": 20232
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071565067838716e-06,
+      "loss": 1.2598,
+      "step": 20233
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071469520351615e-06,
+      "loss": 0.7864,
+      "step": 20234
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0713739728645136e-06,
+      "loss": 0.5749,
+      "step": 20235
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071278425377413e-06,
+      "loss": 0.7987,
+      "step": 20236
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071182877890312e-06,
+      "loss": 1.1424,
+      "step": 20237
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0710873304032107e-06,
+      "loss": 0.8722,
+      "step": 20238
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0709917829161097e-06,
+      "loss": 0.7657,
+      "step": 20239
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0708962354290083e-06,
+      "loss": 0.7633,
+      "step": 20240
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0708006879419074e-06,
+      "loss": 0.4931,
+      "step": 20241
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.070705140454806e-06,
+      "loss": 0.8562,
+      "step": 20242
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0706095929677054e-06,
+      "loss": 0.8846,
+      "step": 20243
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0705140454806045e-06,
+      "loss": 0.6536,
+      "step": 20244
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070418497993503e-06,
+      "loss": 0.959,
+      "step": 20245
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.070322950506402e-06,
+      "loss": 0.7591,
+      "step": 20246
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0702274030193007e-06,
+      "loss": 0.7419,
+      "step": 20247
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0701318555321998e-06,
+      "loss": 0.6902,
+      "step": 20248
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0700363080450984e-06,
+      "loss": 0.4206,
+      "step": 20249
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069940760557998e-06,
+      "loss": 0.8545,
+      "step": 20250
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0698452130708965e-06,
+      "loss": 0.728,
+      "step": 20251
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0697496655837955e-06,
+      "loss": 0.9998,
+      "step": 20252
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069654118096694e-06,
+      "loss": 0.8641,
+      "step": 20253
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069558570609593e-06,
+      "loss": 0.8439,
+      "step": 20254
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069463023122492e-06,
+      "loss": 1.0976,
+      "step": 20255
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069367475635391e-06,
+      "loss": 0.9149,
+      "step": 20256
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0692719281482903e-06,
+      "loss": 1.0825,
+      "step": 20257
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069176380661189e-06,
+      "loss": 0.548,
+      "step": 20258
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.069080833174088e-06,
+      "loss": 0.7644,
+      "step": 20259
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0689852856869866e-06,
+      "loss": 0.9513,
+      "step": 20260
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0688897381998856e-06,
+      "loss": 0.9596,
+      "step": 20261
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.068794190712784e-06,
+      "loss": 1.1174,
+      "step": 20262
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0686986432256832e-06,
+      "loss": 0.8097,
+      "step": 20263
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.068603095738582e-06,
+      "loss": 0.8136,
+      "step": 20264
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0685075482514813e-06,
+      "loss": 0.6603,
+      "step": 20265
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0684120007643804e-06,
+      "loss": 0.8229,
+      "step": 20266
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.068316453277279e-06,
+      "loss": 0.863,
+      "step": 20267
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.068220905790178e-06,
+      "loss": 0.8175,
+      "step": 20268
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0681253583030766e-06,
+      "loss": 1.1422,
+      "step": 20269
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0680298108159757e-06,
+      "loss": 0.6558,
+      "step": 20270
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0679342633288743e-06,
+      "loss": 0.9329,
+      "step": 20271
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0678387158417737e-06,
+      "loss": 0.3429,
+      "step": 20272
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0677431683546728e-06,
+      "loss": 0.7784,
+      "step": 20273
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0676476208675714e-06,
+      "loss": 0.6449,
+      "step": 20274
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0675520733804704e-06,
+      "loss": 0.5301,
+      "step": 20275
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.067456525893369e-06,
+      "loss": 0.5984,
+      "step": 20276
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.067360978406268e-06,
+      "loss": 0.8069,
+      "step": 20277
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0672654309191667e-06,
+      "loss": 1.0718,
+      "step": 20278
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.067169883432066e-06,
+      "loss": 1.3952,
+      "step": 20279
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.067074335944965e-06,
+      "loss": 0.7789,
+      "step": 20280
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066978788457864e-06,
+      "loss": 0.644,
+      "step": 20281
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066883240970763e-06,
+      "loss": 0.9425,
+      "step": 20282
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0667876934836615e-06,
+      "loss": 0.6456,
+      "step": 20283
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0666921459965605e-06,
+      "loss": 0.7819,
+      "step": 20284
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066596598509459e-06,
+      "loss": 1.222,
+      "step": 20285
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0665010510223586e-06,
+      "loss": 0.914,
+      "step": 20286
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0664055035352576e-06,
+      "loss": 0.6288,
+      "step": 20287
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0663099560481562e-06,
+      "loss": 0.8532,
+      "step": 20288
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0662144085610553e-06,
+      "loss": 1.3088,
+      "step": 20289
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066118861073954e-06,
+      "loss": 0.9024,
+      "step": 20290
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066023313586853e-06,
+      "loss": 0.9297,
+      "step": 20291
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0659277660997516e-06,
+      "loss": 0.8593,
+      "step": 20292
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.065832218612651e-06,
+      "loss": 1.0375,
+      "step": 20293
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.06573667112555e-06,
+      "loss": 0.776,
+      "step": 20294
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0656411236384487e-06,
+      "loss": 0.8534,
+      "step": 20295
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0655455761513477e-06,
+      "loss": 0.473,
+      "step": 20296
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0654500286642463e-06,
+      "loss": 0.6637,
+      "step": 20297
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0653544811771454e-06,
+      "loss": 0.7868,
+      "step": 20298
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065258933690044e-06,
+      "loss": 1.2964,
+      "step": 20299
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065163386202943e-06,
+      "loss": 0.9228,
+      "step": 20300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.065067838715842e-06,
+      "loss": 1.0628,
+      "step": 20301
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064972291228741e-06,
+      "loss": 0.54,
+      "step": 20302
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0648767437416397e-06,
+      "loss": 1.3663,
+      "step": 20303
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0647811962545387e-06,
+      "loss": 1.0518,
+      "step": 20304
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0646856487674378e-06,
+      "loss": 0.9706,
+      "step": 20305
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0645901012803364e-06,
+      "loss": 0.7243,
+      "step": 20306
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0644945537932354e-06,
+      "loss": 1.3622,
+      "step": 20307
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0643990063061345e-06,
+      "loss": 0.6676,
+      "step": 20308
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0643034588190335e-06,
+      "loss": 1.0928,
+      "step": 20309
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064207911331932e-06,
+      "loss": 1.005,
+      "step": 20310
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.064112363844831e-06,
+      "loss": 1.1252,
+      "step": 20311
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0640168163577298e-06,
+      "loss": 0.8473,
+      "step": 20312
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.063921268870629e-06,
+      "loss": 0.7654,
+      "step": 20313
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0638257213835274e-06,
+      "loss": 1.2825,
+      "step": 20314
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.063730173896427e-06,
+      "loss": 1.3681,
+      "step": 20315
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.063634626409326e-06,
+      "loss": 0.9088,
+      "step": 20316
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0635390789222246e-06,
+      "loss": 0.9383,
+      "step": 20317
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0634435314351236e-06,
+      "loss": 1.1055,
+      "step": 20318
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.063347983948022e-06,
+      "loss": 1.2357,
+      "step": 20319
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0632524364609212e-06,
+      "loss": 1.6628,
+      "step": 20320
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.06315688897382e-06,
+      "loss": 1.0804,
+      "step": 20321
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0630613414867193e-06,
+      "loss": 0.898,
+      "step": 20322
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0629657939996184e-06,
+      "loss": 0.9696,
+      "step": 20323
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.062870246512517e-06,
+      "loss": 0.6524,
+      "step": 20324
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.062774699025416e-06,
+      "loss": 0.7436,
+      "step": 20325
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0626791515383146e-06,
+      "loss": 1.129,
+      "step": 20326
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0625836040512137e-06,
+      "loss": 0.6016,
+      "step": 20327
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0624880565641123e-06,
+      "loss": 0.8575,
+      "step": 20328
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0623925090770117e-06,
+      "loss": 0.7728,
+      "step": 20329
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0622969615899108e-06,
+      "loss": 1.0655,
+      "step": 20330
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0622014141028094e-06,
+      "loss": 0.7338,
+      "step": 20331
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0621058666157084e-06,
+      "loss": 1.1209,
+      "step": 20332
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.062010319128607e-06,
+      "loss": 1.0857,
+      "step": 20333
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061914771641506e-06,
+      "loss": 0.8729,
+      "step": 20334
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0618192241544047e-06,
+      "loss": 1.004,
+      "step": 20335
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0617236766673037e-06,
+      "loss": 0.9949,
+      "step": 20336
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061628129180203e-06,
+      "loss": 0.9061,
+      "step": 20337
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061532581693102e-06,
+      "loss": 1.0127,
+      "step": 20338
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061437034206001e-06,
+      "loss": 0.8533,
+      "step": 20339
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0613414867188995e-06,
+      "loss": 0.9517,
+      "step": 20340
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0612459392317985e-06,
+      "loss": 0.5815,
+      "step": 20341
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061150391744697e-06,
+      "loss": 1.3322,
+      "step": 20342
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.061054844257596e-06,
+      "loss": 1.0955,
+      "step": 20343
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060959296770495e-06,
+      "loss": 1.1212,
+      "step": 20344
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0608637492833942e-06,
+      "loss": 1.1053,
+      "step": 20345
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0607682017962933e-06,
+      "loss": 0.5595,
+      "step": 20346
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060672654309192e-06,
+      "loss": 0.7315,
+      "step": 20347
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060577106822091e-06,
+      "loss": 0.9054,
+      "step": 20348
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0604815593349895e-06,
+      "loss": 0.5491,
+      "step": 20349
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0603860118478886e-06,
+      "loss": 0.8249,
+      "step": 20350
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0602904643607876e-06,
+      "loss": 1.075,
+      "step": 20351
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0601949168736867e-06,
+      "loss": 1.1312,
+      "step": 20352
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0600993693865853e-06,
+      "loss": 0.8907,
+      "step": 20353
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0600038218994843e-06,
+      "loss": 0.6644,
+      "step": 20354
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059908274412383e-06,
+      "loss": 0.8181,
+      "step": 20355
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059812726925282e-06,
+      "loss": 0.6869,
+      "step": 20356
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059717179438181e-06,
+      "loss": 1.2281,
+      "step": 20357
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.05962163195108e-06,
+      "loss": 0.9479,
+      "step": 20358
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059526084463979e-06,
+      "loss": 1.2046,
+      "step": 20359
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0594305369768777e-06,
+      "loss": 1.0406,
+      "step": 20360
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0593349894897767e-06,
+      "loss": 0.69,
+      "step": 20361
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0592394420026754e-06,
+      "loss": 1.1505,
+      "step": 20362
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0591438945155744e-06,
+      "loss": 0.9925,
+      "step": 20363
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.059048347028473e-06,
+      "loss": 0.5632,
+      "step": 20364
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0589527995413725e-06,
+      "loss": 0.5697,
+      "step": 20365
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0588572520542715e-06,
+      "loss": 0.781,
+      "step": 20366
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.05876170456717e-06,
+      "loss": 1.2722,
+      "step": 20367
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.058666157080069e-06,
+      "loss": 0.864,
+      "step": 20368
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0585706095929678e-06,
+      "loss": 1.4707,
+      "step": 20369
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.058475062105867e-06,
+      "loss": 0.7901,
+      "step": 20370
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0583795146187654e-06,
+      "loss": 0.3937,
+      "step": 20371
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0582839671316645e-06,
+      "loss": 0.8697,
+      "step": 20372
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.058188419644564e-06,
+      "loss": 1.2211,
+      "step": 20373
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0580928721574625e-06,
+      "loss": 0.8178,
+      "step": 20374
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0579973246703616e-06,
+      "loss": 0.911,
+      "step": 20375
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.05790177718326e-06,
+      "loss": 0.7659,
+      "step": 20376
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0578062296961592e-06,
+      "loss": 0.7543,
+      "step": 20377
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.057710682209058e-06,
+      "loss": 0.6428,
+      "step": 20378
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.057615134721957e-06,
+      "loss": 1.0343,
+      "step": 20379
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0575195872348564e-06,
+      "loss": 0.9,
+      "step": 20380
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.057424039747755e-06,
+      "loss": 1.099,
+      "step": 20381
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.057328492260654e-06,
+      "loss": 0.6777,
+      "step": 20382
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0572329447735526e-06,
+      "loss": 0.8946,
+      "step": 20383
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0571373972864517e-06,
+      "loss": 1.2832,
+      "step": 20384
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0570418497993503e-06,
+      "loss": 0.3009,
+      "step": 20385
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0569463023122493e-06,
+      "loss": 0.9743,
+      "step": 20386
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0568507548251488e-06,
+      "loss": 1.4203,
+      "step": 20387
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0567552073380474e-06,
+      "loss": 0.6833,
+      "step": 20388
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0566596598509464e-06,
+      "loss": 1.0164,
+      "step": 20389
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.056564112363845e-06,
+      "loss": 0.664,
+      "step": 20390
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.056468564876744e-06,
+      "loss": 0.7864,
+      "step": 20391
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0563730173896427e-06,
+      "loss": 0.5657,
+      "step": 20392
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0562774699025417e-06,
+      "loss": 1.0046,
+      "step": 20393
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0561819224154408e-06,
+      "loss": 0.5961,
+      "step": 20394
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.05608637492834e-06,
+      "loss": 1.1108,
+      "step": 20395
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055990827441239e-06,
+      "loss": 1.0797,
+      "step": 20396
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0558952799541375e-06,
+      "loss": 1.1051,
+      "step": 20397
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0557997324670365e-06,
+      "loss": 0.8586,
+      "step": 20398
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055704184979935e-06,
+      "loss": 0.5683,
+      "step": 20399
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055608637492834e-06,
+      "loss": 1.1125,
+      "step": 20400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055513090005733e-06,
+      "loss": 1.1317,
+      "step": 20401
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0554175425186322e-06,
+      "loss": 0.8862,
+      "step": 20402
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.055321995031531e-06,
+      "loss": 0.7768,
+      "step": 20403
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.05522644754443e-06,
+      "loss": 0.8239,
+      "step": 20404
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0551309000573285e-06,
+      "loss": 0.472,
+      "step": 20405
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0550353525702275e-06,
+      "loss": 1.1909,
+      "step": 20406
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054939805083126e-06,
+      "loss": 0.9208,
+      "step": 20407
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054844257596025e-06,
+      "loss": 1.7078,
+      "step": 20408
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0547487101089247e-06,
+      "loss": 1.0138,
+      "step": 20409
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0546531626218233e-06,
+      "loss": 0.7088,
+      "step": 20410
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0545576151347223e-06,
+      "loss": 1.3418,
+      "step": 20411
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054462067647621e-06,
+      "loss": 0.6034,
+      "step": 20412
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.05436652016052e-06,
+      "loss": 0.7398,
+      "step": 20413
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0542709726734186e-06,
+      "loss": 1.0577,
+      "step": 20414
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0541754251863176e-06,
+      "loss": 1.1123,
+      "step": 20415
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.054079877699217e-06,
+      "loss": 0.7451,
+      "step": 20416
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0539843302121157e-06,
+      "loss": 0.7369,
+      "step": 20417
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0538887827250147e-06,
+      "loss": 0.9133,
+      "step": 20418
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0537932352379134e-06,
+      "loss": 0.9228,
+      "step": 20419
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0536976877508124e-06,
+      "loss": 1.0076,
+      "step": 20420
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.053602140263711e-06,
+      "loss": 1.4834,
+      "step": 20421
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.05350659277661e-06,
+      "loss": 0.8321,
+      "step": 20422
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0534110452895095e-06,
+      "loss": 0.9276,
+      "step": 20423
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.053315497802408e-06,
+      "loss": 1.1149,
+      "step": 20424
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.053219950315307e-06,
+      "loss": 0.7871,
+      "step": 20425
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0531244028282058e-06,
+      "loss": 0.9258,
+      "step": 20426
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.053028855341105e-06,
+      "loss": 0.6755,
+      "step": 20427
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0529333078540034e-06,
+      "loss": 1.1446,
+      "step": 20428
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0528377603669025e-06,
+      "loss": 1.0128,
+      "step": 20429
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052742212879802e-06,
+      "loss": 1.4022,
+      "step": 20430
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0526466653927005e-06,
+      "loss": 1.2994,
+      "step": 20431
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0525511179055996e-06,
+      "loss": 0.5875,
+      "step": 20432
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052455570418498e-06,
+      "loss": 1.0874,
+      "step": 20433
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0523600229313972e-06,
+      "loss": 0.8854,
+      "step": 20434
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052264475444296e-06,
+      "loss": 0.7858,
+      "step": 20435
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052168927957195e-06,
+      "loss": 0.9247,
+      "step": 20436
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0520733804700944e-06,
+      "loss": 1.1982,
+      "step": 20437
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051977832982993e-06,
+      "loss": 0.4969,
+      "step": 20438
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051882285495892e-06,
+      "loss": 1.0795,
+      "step": 20439
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0517867380087906e-06,
+      "loss": 0.9725,
+      "step": 20440
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0516911905216897e-06,
+      "loss": 0.9144,
+      "step": 20441
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0515956430345883e-06,
+      "loss": 0.8564,
+      "step": 20442
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0515000955474873e-06,
+      "loss": 0.9449,
+      "step": 20443
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051404548060386e-06,
+      "loss": 0.9749,
+      "step": 20444
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0513090005732854e-06,
+      "loss": 0.8067,
+      "step": 20445
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051213453086184e-06,
+      "loss": 0.8302,
+      "step": 20446
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051117905599083e-06,
+      "loss": 1.1304,
+      "step": 20447
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.051022358111982e-06,
+      "loss": 0.7336,
+      "step": 20448
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0509268106248807e-06,
+      "loss": 0.5594,
+      "step": 20449
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0508312631377797e-06,
+      "loss": 0.9355,
+      "step": 20450
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0507357156506784e-06,
+      "loss": 0.9613,
+      "step": 20451
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.050640168163578e-06,
+      "loss": 0.8301,
+      "step": 20452
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0505446206764764e-06,
+      "loss": 0.5731,
+      "step": 20453
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0504490731893755e-06,
+      "loss": 0.7735,
+      "step": 20454
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.050353525702274e-06,
+      "loss": 0.9619,
+      "step": 20455
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.050257978215173e-06,
+      "loss": 0.8583,
+      "step": 20456
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0501624307280717e-06,
+      "loss": 0.9768,
+      "step": 20457
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0500668832409708e-06,
+      "loss": 1.2194,
+      "step": 20458
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0499713357538702e-06,
+      "loss": 1.2777,
+      "step": 20459
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.049875788266769e-06,
+      "loss": 1.1066,
+      "step": 20460
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.049780240779668e-06,
+      "loss": 0.4996,
+      "step": 20461
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0496846932925665e-06,
+      "loss": 0.8922,
+      "step": 20462
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0495891458054655e-06,
+      "loss": 0.9289,
+      "step": 20463
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.049493598318364e-06,
+      "loss": 1.2167,
+      "step": 20464
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.049398050831263e-06,
+      "loss": 0.8425,
+      "step": 20465
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0493025033441627e-06,
+      "loss": 0.551,
+      "step": 20466
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0492069558570613e-06,
+      "loss": 0.5405,
+      "step": 20467
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0491114083699603e-06,
+      "loss": 0.6672,
+      "step": 20468
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.049015860882859e-06,
+      "loss": 0.9319,
+      "step": 20469
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.048920313395758e-06,
+      "loss": 1.1029,
+      "step": 20470
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0488247659086566e-06,
+      "loss": 1.2621,
+      "step": 20471
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0487292184215556e-06,
+      "loss": 1.0199,
+      "step": 20472
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.048633670934455e-06,
+      "loss": 0.7628,
+      "step": 20473
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0485381234473537e-06,
+      "loss": 0.8417,
+      "step": 20474
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0484425759602527e-06,
+      "loss": 0.8052,
+      "step": 20475
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0483470284731514e-06,
+      "loss": 1.2596,
+      "step": 20476
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0482514809860504e-06,
+      "loss": 1.0314,
+      "step": 20477
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.048155933498949e-06,
+      "loss": 0.6158,
+      "step": 20478
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.048060386011848e-06,
+      "loss": 0.6385,
+      "step": 20479
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0479648385247467e-06,
+      "loss": 1.0219,
+      "step": 20480
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047869291037646e-06,
+      "loss": 0.527,
+      "step": 20481
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047773743550545e-06,
+      "loss": 0.9015,
+      "step": 20482
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0476781960634438e-06,
+      "loss": 0.7029,
+      "step": 20483
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047582648576343e-06,
+      "loss": 0.9792,
+      "step": 20484
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0474871010892414e-06,
+      "loss": 0.8734,
+      "step": 20485
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0473915536021405e-06,
+      "loss": 1.1762,
+      "step": 20486
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047296006115039e-06,
+      "loss": 1.0099,
+      "step": 20487
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0472004586279385e-06,
+      "loss": 0.8802,
+      "step": 20488
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0471049111408376e-06,
+      "loss": 0.9297,
+      "step": 20489
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047009363653736e-06,
+      "loss": 1.4538,
+      "step": 20490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0469138161666352e-06,
+      "loss": 0.8818,
+      "step": 20491
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.046818268679534e-06,
+      "loss": 0.7934,
+      "step": 20492
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.046722721192433e-06,
+      "loss": 0.9632,
+      "step": 20493
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0466271737053315e-06,
+      "loss": 0.9263,
+      "step": 20494
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.046531626218231e-06,
+      "loss": 0.9845,
+      "step": 20495
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0464360787311296e-06,
+      "loss": 1.255,
+      "step": 20496
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0463405312440286e-06,
+      "loss": 0.8231,
+      "step": 20497
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0462449837569277e-06,
+      "loss": 0.7896,
+      "step": 20498
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0461494362698263e-06,
+      "loss": 0.5727,
+      "step": 20499
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0460538887827253e-06,
+      "loss": 0.7177,
+      "step": 20500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.045958341295624e-06,
+      "loss": 1.079,
+      "step": 20501
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0458627938085234e-06,
+      "loss": 0.7061,
+      "step": 20502
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.045767246321422e-06,
+      "loss": 1.2325,
+      "step": 20503
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.045671698834321e-06,
+      "loss": 0.6774,
+      "step": 20504
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0455761513472197e-06,
+      "loss": 0.5801,
+      "step": 20505
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0454806038601187e-06,
+      "loss": 0.9235,
+      "step": 20506
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0453850563730173e-06,
+      "loss": 0.9235,
+      "step": 20507
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0452895088859164e-06,
+      "loss": 0.8098,
+      "step": 20508
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.045193961398816e-06,
+      "loss": 0.641,
+      "step": 20509
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0450984139117144e-06,
+      "loss": 0.4797,
+      "step": 20510
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0450028664246135e-06,
+      "loss": 0.6875,
+      "step": 20511
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044907318937512e-06,
+      "loss": 0.7975,
+      "step": 20512
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044811771450411e-06,
+      "loss": 0.6382,
+      "step": 20513
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0447162239633097e-06,
+      "loss": 0.9245,
+      "step": 20514
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0446206764762088e-06,
+      "loss": 0.7772,
+      "step": 20515
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0445251289891074e-06,
+      "loss": 0.7847,
+      "step": 20516
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044429581502007e-06,
+      "loss": 0.9946,
+      "step": 20517
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044334034014906e-06,
+      "loss": 0.8826,
+      "step": 20518
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0442384865278045e-06,
+      "loss": 0.9887,
+      "step": 20519
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0441429390407035e-06,
+      "loss": 0.9118,
+      "step": 20520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.044047391553602e-06,
+      "loss": 1.3198,
+      "step": 20521
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.043951844066501e-06,
+      "loss": 0.8814,
+      "step": 20522
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0438562965794e-06,
+      "loss": 0.8159,
+      "step": 20523
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0437607490922993e-06,
+      "loss": 0.6869,
+      "step": 20524
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0436652016051983e-06,
+      "loss": 1.2088,
+      "step": 20525
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.043569654118097e-06,
+      "loss": 0.9471,
+      "step": 20526
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.043474106630996e-06,
+      "loss": 0.4878,
+      "step": 20527
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0433785591438946e-06,
+      "loss": 1.4447,
+      "step": 20528
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0432830116567936e-06,
+      "loss": 0.6539,
+      "step": 20529
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0431874641696922e-06,
+      "loss": 0.9621,
+      "step": 20530
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0430919166825917e-06,
+      "loss": 0.6227,
+      "step": 20531
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0429963691954907e-06,
+      "loss": 0.453,
+      "step": 20532
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0429008217083894e-06,
+      "loss": 0.8172,
+      "step": 20533
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0428052742212884e-06,
+      "loss": 0.8257,
+      "step": 20534
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.042709726734187e-06,
+      "loss": 0.676,
+      "step": 20535
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.042614179247086e-06,
+      "loss": 0.5999,
+      "step": 20536
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0425186317599847e-06,
+      "loss": 0.4613,
+      "step": 20537
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.042423084272884e-06,
+      "loss": 0.6636,
+      "step": 20538
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.042327536785783e-06,
+      "loss": 0.6269,
+      "step": 20539
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0422319892986818e-06,
+      "loss": 0.9098,
+      "step": 20540
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.042136441811581e-06,
+      "loss": 1.1539,
+      "step": 20541
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0420408943244794e-06,
+      "loss": 0.7359,
+      "step": 20542
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0419453468373785e-06,
+      "loss": 0.6775,
+      "step": 20543
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041849799350277e-06,
+      "loss": 0.9583,
+      "step": 20544
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0417542518631765e-06,
+      "loss": 1.0374,
+      "step": 20545
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041658704376075e-06,
+      "loss": 1.1564,
+      "step": 20546
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041563156888974e-06,
+      "loss": 0.6615,
+      "step": 20547
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041467609401873e-06,
+      "loss": 1.0455,
+      "step": 20548
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041372061914772e-06,
+      "loss": 0.8575,
+      "step": 20549
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.041276514427671e-06,
+      "loss": 1.2994,
+      "step": 20550
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0411809669405695e-06,
+      "loss": 0.8535,
+      "step": 20551
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0410854194534685e-06,
+      "loss": 0.7858,
+      "step": 20552
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0409898719663676e-06,
+      "loss": 0.9372,
+      "step": 20553
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0408943244792666e-06,
+      "loss": 0.8356,
+      "step": 20554
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0407987769921652e-06,
+      "loss": 0.8815,
+      "step": 20555
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0407032295050643e-06,
+      "loss": 0.7519,
+      "step": 20556
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.040607682017963e-06,
+      "loss": 1.0499,
+      "step": 20557
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.040512134530862e-06,
+      "loss": 0.8966,
+      "step": 20558
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0404165870437605e-06,
+      "loss": 0.872,
+      "step": 20559
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.04032103955666e-06,
+      "loss": 0.5795,
+      "step": 20560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.040225492069559e-06,
+      "loss": 1.0013,
+      "step": 20561
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0401299445824577e-06,
+      "loss": 0.5856,
+      "step": 20562
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0400343970953567e-06,
+      "loss": 0.9825,
+      "step": 20563
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0399388496082553e-06,
+      "loss": 0.8834,
+      "step": 20564
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0398433021211544e-06,
+      "loss": 0.9695,
+      "step": 20565
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.039747754634053e-06,
+      "loss": 1.2149,
+      "step": 20566
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0396522071469524e-06,
+      "loss": 0.9663,
+      "step": 20567
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0395566596598515e-06,
+      "loss": 1.1972,
+      "step": 20568
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.03946111217275e-06,
+      "loss": 1.0508,
+      "step": 20569
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.039365564685649e-06,
+      "loss": 0.7065,
+      "step": 20570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0392700171985477e-06,
+      "loss": 1.2493,
+      "step": 20571
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0391744697114468e-06,
+      "loss": 1.4076,
+      "step": 20572
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0390789222243454e-06,
+      "loss": 0.8527,
+      "step": 20573
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038983374737245e-06,
+      "loss": 1.1366,
+      "step": 20574
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038887827250144e-06,
+      "loss": 0.6533,
+      "step": 20575
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0387922797630425e-06,
+      "loss": 1.1765,
+      "step": 20576
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0386967322759415e-06,
+      "loss": 0.9176,
+      "step": 20577
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.03860118478884e-06,
+      "loss": 1.1575,
+      "step": 20578
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038505637301739e-06,
+      "loss": 0.8632,
+      "step": 20579
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038410089814638e-06,
+      "loss": 0.8171,
+      "step": 20580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0383145423275373e-06,
+      "loss": 0.9524,
+      "step": 20581
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0382189948404363e-06,
+      "loss": 1.0333,
+      "step": 20582
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038123447353335e-06,
+      "loss": 0.5433,
+      "step": 20583
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.038027899866234e-06,
+      "loss": 0.8282,
+      "step": 20584
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0379323523791326e-06,
+      "loss": 0.6532,
+      "step": 20585
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0378368048920316e-06,
+      "loss": 1.1017,
+      "step": 20586
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0377412574049302e-06,
+      "loss": 0.8674,
+      "step": 20587
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0376457099178293e-06,
+      "loss": 0.8365,
+      "step": 20588
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0375501624307287e-06,
+      "loss": 0.8192,
+      "step": 20589
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0374546149436274e-06,
+      "loss": 0.8834,
+      "step": 20590
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0373590674565264e-06,
+      "loss": 1.0917,
+      "step": 20591
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.037263519969425e-06,
+      "loss": 0.7868,
+      "step": 20592
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.037167972482324e-06,
+      "loss": 0.4562,
+      "step": 20593
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0370724249952227e-06,
+      "loss": 0.6101,
+      "step": 20594
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0369768775081217e-06,
+      "loss": 0.6123,
+      "step": 20595
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0368813300210207e-06,
+      "loss": 0.8898,
+      "step": 20596
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0367857825339198e-06,
+      "loss": 0.8327,
+      "step": 20597
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0366902350468184e-06,
+      "loss": 1.3889,
+      "step": 20598
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0365946875597174e-06,
+      "loss": 0.8832,
+      "step": 20599
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0364991400726165e-06,
+      "loss": 1.0431,
+      "step": 20600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.036403592585515e-06,
+      "loss": 0.7664,
+      "step": 20601
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.036308045098414e-06,
+      "loss": 1.0445,
+      "step": 20602
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.036212497611313e-06,
+      "loss": 0.8348,
+      "step": 20603
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.036116950124212e-06,
+      "loss": 0.5781,
+      "step": 20604
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.036021402637111e-06,
+      "loss": 0.9404,
+      "step": 20605
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.03592585515001e-06,
+      "loss": 1.0194,
+      "step": 20606
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0358303076629085e-06,
+      "loss": 0.6084,
+      "step": 20607
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0357347601758075e-06,
+      "loss": 0.7037,
+      "step": 20608
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.035639212688706e-06,
+      "loss": 0.9085,
+      "step": 20609
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0355436652016056e-06,
+      "loss": 1.0614,
+      "step": 20610
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0354481177145046e-06,
+      "loss": 0.7278,
+      "step": 20611
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0353525702274032e-06,
+      "loss": 0.6782,
+      "step": 20612
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0352570227403023e-06,
+      "loss": 0.9738,
+      "step": 20613
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.035161475253201e-06,
+      "loss": 1.2113,
+      "step": 20614
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0350659277661e-06,
+      "loss": 0.5755,
+      "step": 20615
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0349703802789985e-06,
+      "loss": 0.6648,
+      "step": 20616
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.034874832791898e-06,
+      "loss": 0.9492,
+      "step": 20617
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.034779285304797e-06,
+      "loss": 0.8384,
+      "step": 20618
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0346837378176957e-06,
+      "loss": 1.0574,
+      "step": 20619
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0345881903305947e-06,
+      "loss": 0.5041,
+      "step": 20620
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0344926428434933e-06,
+      "loss": 1.0046,
+      "step": 20621
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0343970953563924e-06,
+      "loss": 0.9978,
+      "step": 20622
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.034301547869291e-06,
+      "loss": 0.7549,
+      "step": 20623
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.03420600038219e-06,
+      "loss": 1.0363,
+      "step": 20624
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0341104528950895e-06,
+      "loss": 0.7836,
+      "step": 20625
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.034014905407988e-06,
+      "loss": 1.0241,
+      "step": 20626
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.033919357920887e-06,
+      "loss": 1.0906,
+      "step": 20627
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0338238104337857e-06,
+      "loss": 0.5048,
+      "step": 20628
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0337282629466848e-06,
+      "loss": 0.7076,
+      "step": 20629
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0336327154595834e-06,
+      "loss": 0.9177,
+      "step": 20630
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0335371679724824e-06,
+      "loss": 0.684,
+      "step": 20631
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.033441620485382e-06,
+      "loss": 0.8594,
+      "step": 20632
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0333460729982805e-06,
+      "loss": 0.6112,
+      "step": 20633
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0332505255111795e-06,
+      "loss": 1.1498,
+      "step": 20634
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.033154978024078e-06,
+      "loss": 0.8793,
+      "step": 20635
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.033059430536977e-06,
+      "loss": 0.9706,
+      "step": 20636
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.032963883049876e-06,
+      "loss": 0.48,
+      "step": 20637
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.032868335562775e-06,
+      "loss": 0.6828,
+      "step": 20638
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.032772788075674e-06,
+      "loss": 1.1376,
+      "step": 20639
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.032677240588573e-06,
+      "loss": 1.0211,
+      "step": 20640
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.032581693101472e-06,
+      "loss": 1.2153,
+      "step": 20641
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0324861456143706e-06,
+      "loss": 0.7118,
+      "step": 20642
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0323905981272696e-06,
+      "loss": 1.097,
+      "step": 20643
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0322950506401682e-06,
+      "loss": 1.0274,
+      "step": 20644
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0321995031530673e-06,
+      "loss": 0.9647,
+      "step": 20645
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0321039556659663e-06,
+      "loss": 0.8552,
+      "step": 20646
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0320084081788654e-06,
+      "loss": 1.0128,
+      "step": 20647
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.031912860691764e-06,
+      "loss": 1.1603,
+      "step": 20648
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.031817313204663e-06,
+      "loss": 0.9589,
+      "step": 20649
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0317217657175616e-06,
+      "loss": 1.241,
+      "step": 20650
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0316262182304607e-06,
+      "loss": 0.6503,
+      "step": 20651
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0315306707433597e-06,
+      "loss": 0.9028,
+      "step": 20652
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0314351232562587e-06,
+      "loss": 1.0008,
+      "step": 20653
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0313395757691578e-06,
+      "loss": 0.8284,
+      "step": 20654
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0312440282820564e-06,
+      "loss": 1.1974,
+      "step": 20655
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0311484807949554e-06,
+      "loss": 0.8768,
+      "step": 20656
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.031052933307854e-06,
+      "loss": 0.5224,
+      "step": 20657
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.030957385820753e-06,
+      "loss": 0.9517,
+      "step": 20658
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0308618383336517e-06,
+      "loss": 0.7184,
+      "step": 20659
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0307662908465507e-06,
+      "loss": 0.5212,
+      "step": 20660
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.03067074335945e-06,
+      "loss": 1.0577,
+      "step": 20661
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.030575195872349e-06,
+      "loss": 0.8821,
+      "step": 20662
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.030479648385248e-06,
+      "loss": 0.7213,
+      "step": 20663
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0303841008981465e-06,
+      "loss": 0.5952,
+      "step": 20664
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0302885534110455e-06,
+      "loss": 1.4529,
+      "step": 20665
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030193005923944e-06,
+      "loss": 0.5736,
+      "step": 20666
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.030097458436843e-06,
+      "loss": 0.5167,
+      "step": 20667
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0300019109497426e-06,
+      "loss": 0.8344,
+      "step": 20668
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0299063634626412e-06,
+      "loss": 0.8887,
+      "step": 20669
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0298108159755403e-06,
+      "loss": 1.0897,
+      "step": 20670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.029715268488439e-06,
+      "loss": 0.6694,
+      "step": 20671
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.029619721001338e-06,
+      "loss": 0.4371,
+      "step": 20672
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0295241735142365e-06,
+      "loss": 0.6943,
+      "step": 20673
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0294286260271356e-06,
+      "loss": 0.5571,
+      "step": 20674
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.029333078540035e-06,
+      "loss": 0.785,
+      "step": 20675
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0292375310529337e-06,
+      "loss": 0.8273,
+      "step": 20676
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0291419835658327e-06,
+      "loss": 0.938,
+      "step": 20677
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0290464360787313e-06,
+      "loss": 1.18,
+      "step": 20678
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0289508885916304e-06,
+      "loss": 0.7251,
+      "step": 20679
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.028855341104529e-06,
+      "loss": 0.8756,
+      "step": 20680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.028759793617428e-06,
+      "loss": 1.0686,
+      "step": 20681
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0286642461303275e-06,
+      "loss": 0.7269,
+      "step": 20682
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.028568698643226e-06,
+      "loss": 0.8377,
+      "step": 20683
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.028473151156125e-06,
+      "loss": 1.0182,
+      "step": 20684
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0283776036690237e-06,
+      "loss": 1.0098,
+      "step": 20685
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0282820561819228e-06,
+      "loss": 0.8952,
+      "step": 20686
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0281865086948214e-06,
+      "loss": 1.0103,
+      "step": 20687
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0280909612077204e-06,
+      "loss": 0.6103,
+      "step": 20688
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0279954137206195e-06,
+      "loss": 1.0726,
+      "step": 20689
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0278998662335185e-06,
+      "loss": 0.429,
+      "step": 20690
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0278043187464175e-06,
+      "loss": 0.8146,
+      "step": 20691
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027708771259316e-06,
+      "loss": 0.983,
+      "step": 20692
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027613223772215e-06,
+      "loss": 0.4605,
+      "step": 20693
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027517676285114e-06,
+      "loss": 0.4569,
+      "step": 20694
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027422128798013e-06,
+      "loss": 0.8845,
+      "step": 20695
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0273265813109115e-06,
+      "loss": 0.8143,
+      "step": 20696
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027231033823811e-06,
+      "loss": 0.9916,
+      "step": 20697
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0271354863367095e-06,
+      "loss": 1.3148,
+      "step": 20698
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0270399388496086e-06,
+      "loss": 0.9011,
+      "step": 20699
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.026944391362507e-06,
+      "loss": 1.0397,
+      "step": 20700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0268488438754062e-06,
+      "loss": 0.924,
+      "step": 20701
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0267532963883053e-06,
+      "loss": 0.4687,
+      "step": 20702
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.026657748901204e-06,
+      "loss": 0.7716,
+      "step": 20703
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0265622014141034e-06,
+      "loss": 1.2682,
+      "step": 20704
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.026466653927002e-06,
+      "loss": 0.7166,
+      "step": 20705
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.026371106439901e-06,
+      "loss": 0.8996,
+      "step": 20706
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0262755589527996e-06,
+      "loss": 1.3647,
+      "step": 20707
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0261800114656987e-06,
+      "loss": 1.7911,
+      "step": 20708
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0260844639785973e-06,
+      "loss": 0.881,
+      "step": 20709
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0259889164914963e-06,
+      "loss": 1.0656,
+      "step": 20710
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0258933690043958e-06,
+      "loss": 0.6579,
+      "step": 20711
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0257978215172944e-06,
+      "loss": 0.5037,
+      "step": 20712
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0257022740301934e-06,
+      "loss": 1.0784,
+      "step": 20713
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.025606726543092e-06,
+      "loss": 1.0412,
+      "step": 20714
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.025511179055991e-06,
+      "loss": 1.0388,
+      "step": 20715
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0254156315688897e-06,
+      "loss": 1.0019,
+      "step": 20716
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0253200840817887e-06,
+      "loss": 1.0321,
+      "step": 20717
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.025224536594688e-06,
+      "loss": 0.8065,
+      "step": 20718
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.025128989107587e-06,
+      "loss": 0.7114,
+      "step": 20719
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.025033441620486e-06,
+      "loss": 1.0924,
+      "step": 20720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0249378941333845e-06,
+      "loss": 0.8604,
+      "step": 20721
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0248423466462835e-06,
+      "loss": 0.8549,
+      "step": 20722
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024746799159182e-06,
+      "loss": 0.4819,
+      "step": 20723
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024651251672081e-06,
+      "loss": 0.9877,
+      "step": 20724
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0245557041849806e-06,
+      "loss": 0.9713,
+      "step": 20725
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0244601566978792e-06,
+      "loss": 0.4583,
+      "step": 20726
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0243646092107783e-06,
+      "loss": 0.5983,
+      "step": 20727
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024269061723677e-06,
+      "loss": 0.8893,
+      "step": 20728
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.024173514236576e-06,
+      "loss": 1.2372,
+      "step": 20729
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0240779667494745e-06,
+      "loss": 0.7464,
+      "step": 20730
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0239824192623736e-06,
+      "loss": 0.7404,
+      "step": 20731
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.023886871775272e-06,
+      "loss": 1.0326,
+      "step": 20732
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0237913242881717e-06,
+      "loss": 0.624,
+      "step": 20733
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0236957768010707e-06,
+      "loss": 0.638,
+      "step": 20734
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0236002293139693e-06,
+      "loss": 0.8961,
+      "step": 20735
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0235046818268684e-06,
+      "loss": 0.9799,
+      "step": 20736
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.023409134339767e-06,
+      "loss": 0.9365,
+      "step": 20737
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.023313586852666e-06,
+      "loss": 0.3901,
+      "step": 20738
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0232180393655646e-06,
+      "loss": 0.7539,
+      "step": 20739
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.023122491878464e-06,
+      "loss": 0.7772,
+      "step": 20740
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0230269443913627e-06,
+      "loss": 0.9897,
+      "step": 20741
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0229313969042617e-06,
+      "loss": 1.0963,
+      "step": 20742
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0228358494171608e-06,
+      "loss": 0.8418,
+      "step": 20743
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0227403019300594e-06,
+      "loss": 0.9545,
+      "step": 20744
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0226447544429584e-06,
+      "loss": 0.7997,
+      "step": 20745
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.022549206955857e-06,
+      "loss": 0.684,
+      "step": 20746
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0224536594687565e-06,
+      "loss": 0.8177,
+      "step": 20747
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.022358111981655e-06,
+      "loss": 1.2159,
+      "step": 20748
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.022262564494554e-06,
+      "loss": 0.9748,
+      "step": 20749
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0221670170074528e-06,
+      "loss": 1.1407,
+      "step": 20750
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.022071469520352e-06,
+      "loss": 0.9437,
+      "step": 20751
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0219759220332504e-06,
+      "loss": 0.7925,
+      "step": 20752
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0218803745461495e-06,
+      "loss": 0.7001,
+      "step": 20753
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.021784827059049e-06,
+      "loss": 0.6739,
+      "step": 20754
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0216892795719475e-06,
+      "loss": 0.8771,
+      "step": 20755
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0215937320848466e-06,
+      "loss": 1.0655,
+      "step": 20756
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.021498184597745e-06,
+      "loss": 1.4736,
+      "step": 20757
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0214026371106442e-06,
+      "loss": 0.9659,
+      "step": 20758
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.021307089623543e-06,
+      "loss": 0.4657,
+      "step": 20759
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.021211542136442e-06,
+      "loss": 0.7859,
+      "step": 20760
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0211159946493414e-06,
+      "loss": 0.8115,
+      "step": 20761
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.02102044716224e-06,
+      "loss": 1.0981,
+      "step": 20762
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.020924899675139e-06,
+      "loss": 0.951,
+      "step": 20763
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0208293521880376e-06,
+      "loss": 0.9649,
+      "step": 20764
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0207338047009367e-06,
+      "loss": 1.3804,
+      "step": 20765
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0206382572138353e-06,
+      "loss": 1.3023,
+      "step": 20766
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0205427097267343e-06,
+      "loss": 0.908,
+      "step": 20767
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.020447162239633e-06,
+      "loss": 0.7701,
+      "step": 20768
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0203516147525324e-06,
+      "loss": 1.0802,
+      "step": 20769
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0202560672654314e-06,
+      "loss": 1.3791,
+      "step": 20770
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.02016051977833e-06,
+      "loss": 1.435,
+      "step": 20771
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.020064972291229e-06,
+      "loss": 0.7472,
+      "step": 20772
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0199694248041277e-06,
+      "loss": 0.8851,
+      "step": 20773
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0198738773170267e-06,
+      "loss": 1.0599,
+      "step": 20774
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0197783298299253e-06,
+      "loss": 0.5909,
+      "step": 20775
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.019682782342825e-06,
+      "loss": 0.6474,
+      "step": 20776
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.019587234855724e-06,
+      "loss": 0.82,
+      "step": 20777
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0194916873686225e-06,
+      "loss": 1.0765,
+      "step": 20778
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0193961398815215e-06,
+      "loss": 0.8703,
+      "step": 20779
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.01930059239442e-06,
+      "loss": 0.7486,
+      "step": 20780
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.019205044907319e-06,
+      "loss": 0.434,
+      "step": 20781
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0191094974202178e-06,
+      "loss": 0.8778,
+      "step": 20782
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0190139499331172e-06,
+      "loss": 0.6904,
+      "step": 20783
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0189184024460163e-06,
+      "loss": 0.9102,
+      "step": 20784
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.018822854958915e-06,
+      "loss": 1.1303,
+      "step": 20785
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.018727307471814e-06,
+      "loss": 1.1468,
+      "step": 20786
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0186317599847125e-06,
+      "loss": 0.8042,
+      "step": 20787
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0185362124976116e-06,
+      "loss": 0.6336,
+      "step": 20788
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.01844066501051e-06,
+      "loss": 0.4952,
+      "step": 20789
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0183451175234097e-06,
+      "loss": 1.0934,
+      "step": 20790
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0182495700363083e-06,
+      "loss": 0.7965,
+      "step": 20791
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0181540225492073e-06,
+      "loss": 0.5161,
+      "step": 20792
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0180584750621063e-06,
+      "loss": 1.0046,
+      "step": 20793
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017962927575005e-06,
+      "loss": 1.1019,
+      "step": 20794
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017867380087904e-06,
+      "loss": 0.9058,
+      "step": 20795
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0177718326008026e-06,
+      "loss": 0.8901,
+      "step": 20796
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017676285113702e-06,
+      "loss": 0.8606,
+      "step": 20797
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0175807376266007e-06,
+      "loss": 0.8033,
+      "step": 20798
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0174851901394997e-06,
+      "loss": 1.5114,
+      "step": 20799
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0173896426523983e-06,
+      "loss": 0.7759,
+      "step": 20800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0172940951652974e-06,
+      "loss": 0.7092,
+      "step": 20801
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017198547678196e-06,
+      "loss": 0.8403,
+      "step": 20802
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.017103000191095e-06,
+      "loss": 0.6893,
+      "step": 20803
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0170074527039937e-06,
+      "loss": 0.5468,
+      "step": 20804
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.016911905216893e-06,
+      "loss": 1.0142,
+      "step": 20805
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.016816357729792e-06,
+      "loss": 0.4316,
+      "step": 20806
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0167208102426908e-06,
+      "loss": 1.2453,
+      "step": 20807
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.01662526275559e-06,
+      "loss": 1.1294,
+      "step": 20808
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0165297152684884e-06,
+      "loss": 0.7369,
+      "step": 20809
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0164341677813875e-06,
+      "loss": 1.2245,
+      "step": 20810
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.016338620294286e-06,
+      "loss": 0.6544,
+      "step": 20811
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0162430728071855e-06,
+      "loss": 1.022,
+      "step": 20812
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0161475253200846e-06,
+      "loss": 0.9212,
+      "step": 20813
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.016051977832983e-06,
+      "loss": 0.9804,
+      "step": 20814
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0159564303458822e-06,
+      "loss": 1.0453,
+      "step": 20815
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.015860882858781e-06,
+      "loss": 1.0773,
+      "step": 20816
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.01576533537168e-06,
+      "loss": 0.8674,
+      "step": 20817
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0156697878845785e-06,
+      "loss": 0.6264,
+      "step": 20818
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.015574240397478e-06,
+      "loss": 1.2573,
+      "step": 20819
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.015478692910377e-06,
+      "loss": 0.8877,
+      "step": 20820
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0153831454232756e-06,
+      "loss": 0.7082,
+      "step": 20821
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0152875979361747e-06,
+      "loss": 1.6569,
+      "step": 20822
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0151920504490733e-06,
+      "loss": 0.9591,
+      "step": 20823
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0150965029619723e-06,
+      "loss": 0.9858,
+      "step": 20824
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.015000955474871e-06,
+      "loss": 0.958,
+      "step": 20825
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0149054079877704e-06,
+      "loss": 0.9176,
+      "step": 20826
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0148098605006694e-06,
+      "loss": 0.9243,
+      "step": 20827
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.014714313013568e-06,
+      "loss": 0.7956,
+      "step": 20828
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.014618765526467e-06,
+      "loss": 1.1763,
+      "step": 20829
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0145232180393657e-06,
+      "loss": 0.7936,
+      "step": 20830
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0144276705522647e-06,
+      "loss": 0.8881,
+      "step": 20831
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0143321230651633e-06,
+      "loss": 0.7915,
+      "step": 20832
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.014236575578063e-06,
+      "loss": 1.1957,
+      "step": 20833
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.014141028090962e-06,
+      "loss": 0.5931,
+      "step": 20834
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0140454806038605e-06,
+      "loss": 0.8461,
+      "step": 20835
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0139499331167595e-06,
+      "loss": 0.6907,
+      "step": 20836
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.013854385629658e-06,
+      "loss": 0.898,
+      "step": 20837
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.013758838142557e-06,
+      "loss": 0.9287,
+      "step": 20838
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0136632906554558e-06,
+      "loss": 0.6457,
+      "step": 20839
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.013567743168355e-06,
+      "loss": 1.0459,
+      "step": 20840
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.013472195681254e-06,
+      "loss": 0.8913,
+      "step": 20841
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.013376648194153e-06,
+      "loss": 0.9452,
+      "step": 20842
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0132811007070515e-06,
+      "loss": 0.9522,
+      "step": 20843
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0131855532199505e-06,
+      "loss": 0.8463,
+      "step": 20844
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0130900057328496e-06,
+      "loss": 1.0277,
+      "step": 20845
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.012994458245748e-06,
+      "loss": 0.9969,
+      "step": 20846
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0128989107586472e-06,
+      "loss": 0.9233,
+      "step": 20847
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0128033632715463e-06,
+      "loss": 0.8691,
+      "step": 20848
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0127078157844453e-06,
+      "loss": 1.1258,
+      "step": 20849
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.012612268297344e-06,
+      "loss": 0.7063,
+      "step": 20850
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.012516720810243e-06,
+      "loss": 0.537,
+      "step": 20851
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0124211733231416e-06,
+      "loss": 0.8055,
+      "step": 20852
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0123256258360406e-06,
+      "loss": 0.657,
+      "step": 20853
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0122300783489392e-06,
+      "loss": 1.3038,
+      "step": 20854
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0121345308618387e-06,
+      "loss": 0.8125,
+      "step": 20855
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0120389833747377e-06,
+      "loss": 0.7683,
+      "step": 20856
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0119434358876363e-06,
+      "loss": 1.02,
+      "step": 20857
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0118478884005354e-06,
+      "loss": 0.8219,
+      "step": 20858
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.011752340913434e-06,
+      "loss": 0.5125,
+      "step": 20859
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.011656793426333e-06,
+      "loss": 0.9167,
+      "step": 20860
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0115612459392317e-06,
+      "loss": 0.8158,
+      "step": 20861
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.011465698452131e-06,
+      "loss": 1.155,
+      "step": 20862
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.01137015096503e-06,
+      "loss": 0.8926,
+      "step": 20863
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0112746034779288e-06,
+      "loss": 0.6814,
+      "step": 20864
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.011179055990828e-06,
+      "loss": 0.885,
+      "step": 20865
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0110835085037264e-06,
+      "loss": 1.1848,
+      "step": 20866
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0109879610166255e-06,
+      "loss": 0.6308,
+      "step": 20867
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010892413529524e-06,
+      "loss": 1.0363,
+      "step": 20868
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0107968660424235e-06,
+      "loss": 0.7756,
+      "step": 20869
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0107013185553226e-06,
+      "loss": 1.0796,
+      "step": 20870
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010605771068221e-06,
+      "loss": 1.1796,
+      "step": 20871
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0105102235811202e-06,
+      "loss": 0.7158,
+      "step": 20872
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010414676094019e-06,
+      "loss": 0.4148,
+      "step": 20873
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.010319128606918e-06,
+      "loss": 0.595,
+      "step": 20874
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0102235811198165e-06,
+      "loss": 0.8548,
+      "step": 20875
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0101280336327155e-06,
+      "loss": 1.0898,
+      "step": 20876
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.010032486145615e-06,
+      "loss": 0.9537,
+      "step": 20877
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0099369386585136e-06,
+      "loss": 0.733,
+      "step": 20878
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0098413911714127e-06,
+      "loss": 1.0969,
+      "step": 20879
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0097458436843113e-06,
+      "loss": 1.104,
+      "step": 20880
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0096502961972103e-06,
+      "loss": 1.0956,
+      "step": 20881
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009554748710109e-06,
+      "loss": 0.731,
+      "step": 20882
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009459201223008e-06,
+      "loss": 0.9818,
+      "step": 20883
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0093636537359074e-06,
+      "loss": 0.4901,
+      "step": 20884
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009268106248806e-06,
+      "loss": 0.8736,
+      "step": 20885
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.009172558761705e-06,
+      "loss": 1.0087,
+      "step": 20886
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0090770112746037e-06,
+      "loss": 0.5204,
+      "step": 20887
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0089814637875027e-06,
+      "loss": 0.6244,
+      "step": 20888
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0088859163004013e-06,
+      "loss": 0.7826,
+      "step": 20889
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0087903688133004e-06,
+      "loss": 0.858,
+      "step": 20890
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0086948213261994e-06,
+      "loss": 0.8606,
+      "step": 20891
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0085992738390985e-06,
+      "loss": 1.2516,
+      "step": 20892
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.008503726351997e-06,
+      "loss": 1.1678,
+      "step": 20893
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.008408178864896e-06,
+      "loss": 1.134,
+      "step": 20894
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.008312631377795e-06,
+      "loss": 1.0518,
+      "step": 20895
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0082170838906938e-06,
+      "loss": 1.0318,
+      "step": 20896
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.008121536403593e-06,
+      "loss": 1.4636,
+      "step": 20897
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.008025988916492e-06,
+      "loss": 1.2782,
+      "step": 20898
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007930441429391e-06,
+      "loss": 0.9948,
+      "step": 20899
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0078348939422895e-06,
+      "loss": 0.6764,
+      "step": 20900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0077393464551885e-06,
+      "loss": 0.6849,
+      "step": 20901
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007643798968087e-06,
+      "loss": 0.4974,
+      "step": 20902
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007548251480986e-06,
+      "loss": 1.1554,
+      "step": 20903
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007452703993885e-06,
+      "loss": 0.8869,
+      "step": 20904
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0073571565067843e-06,
+      "loss": 1.0055,
+      "step": 20905
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0072616090196833e-06,
+      "loss": 1.121,
+      "step": 20906
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007166061532582e-06,
+      "loss": 0.8738,
+      "step": 20907
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007070514045481e-06,
+      "loss": 1.4793,
+      "step": 20908
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0069749665583796e-06,
+      "loss": 0.7945,
+      "step": 20909
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0068794190712786e-06,
+      "loss": 1.3913,
+      "step": 20910
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0067838715841772e-06,
+      "loss": 0.7297,
+      "step": 20911
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0066883240970763e-06,
+      "loss": 0.9688,
+      "step": 20912
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0065927766099757e-06,
+      "loss": 0.9062,
+      "step": 20913
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0064972291228743e-06,
+      "loss": 0.7205,
+      "step": 20914
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0064016816357734e-06,
+      "loss": 0.7661,
+      "step": 20915
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.006306134148672e-06,
+      "loss": 0.6201,
+      "step": 20916
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.006210586661571e-06,
+      "loss": 1.1373,
+      "step": 20917
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0061150391744697e-06,
+      "loss": 0.6213,
+      "step": 20918
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0060194916873687e-06,
+      "loss": 1.0644,
+      "step": 20919
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.005923944200268e-06,
+      "loss": 1.1026,
+      "step": 20920
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0058283967131668e-06,
+      "loss": 0.5804,
+      "step": 20921
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.005732849226066e-06,
+      "loss": 0.4547,
+      "step": 20922
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0056373017389644e-06,
+      "loss": 0.7238,
+      "step": 20923
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0055417542518635e-06,
+      "loss": 0.7925,
+      "step": 20924
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.005446206764762e-06,
+      "loss": 1.3599,
+      "step": 20925
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.005350659277661e-06,
+      "loss": 1.1016,
+      "step": 20926
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0052551117905606e-06,
+      "loss": 0.7582,
+      "step": 20927
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.005159564303459e-06,
+      "loss": 1.0957,
+      "step": 20928
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0050640168163582e-06,
+      "loss": 0.6548,
+      "step": 20929
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004968469329257e-06,
+      "loss": 1.4126,
+      "step": 20930
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004872921842156e-06,
+      "loss": 0.6611,
+      "step": 20931
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0047773743550545e-06,
+      "loss": 1.0177,
+      "step": 20932
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0046818268679535e-06,
+      "loss": 1.076,
+      "step": 20933
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0045862793808526e-06,
+      "loss": 0.972,
+      "step": 20934
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0044907318937516e-06,
+      "loss": 0.977,
+      "step": 20935
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0043951844066507e-06,
+      "loss": 1.0892,
+      "step": 20936
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0042996369195493e-06,
+      "loss": 1.0088,
+      "step": 20937
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0042040894324483e-06,
+      "loss": 0.788,
+      "step": 20938
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004108541945347e-06,
+      "loss": 1.0325,
+      "step": 20939
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.004012994458246e-06,
+      "loss": 1.0173,
+      "step": 20940
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.003917446971145e-06,
+      "loss": 1.0513,
+      "step": 20941
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.003821899484044e-06,
+      "loss": 1.0849,
+      "step": 20942
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0037263519969427e-06,
+      "loss": 0.786,
+      "step": 20943
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0036308045098417e-06,
+      "loss": 1.2647,
+      "step": 20944
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0035352570227403e-06,
+      "loss": 0.7901,
+      "step": 20945
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0034397095356393e-06,
+      "loss": 0.9901,
+      "step": 20946
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0033441620485384e-06,
+      "loss": 1.0381,
+      "step": 20947
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.003248614561437e-06,
+      "loss": 1.4562,
+      "step": 20948
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0031530670743365e-06,
+      "loss": 0.8855,
+      "step": 20949
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.003057519587235e-06,
+      "loss": 0.632,
+      "step": 20950
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.002961972100134e-06,
+      "loss": 0.9976,
+      "step": 20951
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0028664246130327e-06,
+      "loss": 0.7833,
+      "step": 20952
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7037920912583825,
+      "eval_loss": 1.0298326015472412,
+      "eval_runtime": 140.7576,
+      "eval_samples_per_second": 297.696,
+      "eval_steps_per_second": 2.33,
+      "step": 20952
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0027708771259318e-06,
+      "loss": 0.9261,
+      "step": 20953
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0026753296388304e-06,
+      "loss": 0.687,
+      "step": 20954
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0025797821517294e-06,
+      "loss": 1.0324,
+      "step": 20955
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.002484234664629e-06,
+      "loss": 0.8478,
+      "step": 20956
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0023886871775275e-06,
+      "loss": 0.7892,
+      "step": 20957
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0022931396904265e-06,
+      "loss": 1.0976,
+      "step": 20958
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.002197592203325e-06,
+      "loss": 0.8837,
+      "step": 20959
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.002102044716224e-06,
+      "loss": 0.5713,
+      "step": 20960
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.002006497229123e-06,
+      "loss": 1.0068,
+      "step": 20961
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.001910949742022e-06,
+      "loss": 0.8503,
+      "step": 20962
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0018154022549213e-06,
+      "loss": 0.6143,
+      "step": 20963
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.00171985476782e-06,
+      "loss": 0.8988,
+      "step": 20964
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.001624307280719e-06,
+      "loss": 0.8835,
+      "step": 20965
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0015287597936176e-06,
+      "loss": 0.8222,
+      "step": 20966
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0014332123065166e-06,
+      "loss": 0.8973,
+      "step": 20967
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0013376648194152e-06,
+      "loss": 0.8078,
+      "step": 20968
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0012421173323143e-06,
+      "loss": 1.1231,
+      "step": 20969
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0011465698452137e-06,
+      "loss": 0.6811,
+      "step": 20970
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0010510223581123e-06,
+      "loss": 0.4432,
+      "step": 20971
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0009554748710114e-06,
+      "loss": 0.6332,
+      "step": 20972
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.00085992738391e-06,
+      "loss": 0.8796,
+      "step": 20973
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.000764379896809e-06,
+      "loss": 0.488,
+      "step": 20974
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0006688324097077e-06,
+      "loss": 0.8446,
+      "step": 20975
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0005732849226067e-06,
+      "loss": 0.6211,
+      "step": 20976
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.000477737435506e-06,
+      "loss": 0.6425,
+      "step": 20977
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0003821899484048e-06,
+      "loss": 0.6099,
+      "step": 20978
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.000286642461304e-06,
+      "loss": 1.0701,
+      "step": 20979
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0001910949742024e-06,
+      "loss": 0.669,
+      "step": 20980
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.0000955474871015e-06,
+      "loss": 0.8823,
+      "step": 20981
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3e-06,
+      "loss": 1.2585,
+      "step": 20982
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.999904452512899e-06,
+      "loss": 0.9989,
+      "step": 20983
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9998089050257977e-06,
+      "loss": 1.2166,
+      "step": 20984
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.999713357538697e-06,
+      "loss": 0.7208,
+      "step": 20985
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9996178100515962e-06,
+      "loss": 0.7144,
+      "step": 20986
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.999522262564495e-06,
+      "loss": 0.9567,
+      "step": 20987
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.999426715077394e-06,
+      "loss": 0.7956,
+      "step": 20988
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9993311675902925e-06,
+      "loss": 0.9393,
+      "step": 20989
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9992356201031915e-06,
+      "loss": 0.7195,
+      "step": 20990
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.99914007261609e-06,
+      "loss": 0.8476,
+      "step": 20991
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9990445251289896e-06,
+      "loss": 0.9632,
+      "step": 20992
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9989489776418882e-06,
+      "loss": 0.5203,
+      "step": 20993
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9988534301547873e-06,
+      "loss": 0.5074,
+      "step": 20994
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.998757882667686e-06,
+      "loss": 0.6332,
+      "step": 20995
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.998662335180585e-06,
+      "loss": 0.6475,
+      "step": 20996
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.998566787693484e-06,
+      "loss": 1.2579,
+      "step": 20997
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9984712402063826e-06,
+      "loss": 0.703,
+      "step": 20998
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.998375692719282e-06,
+      "loss": 0.4932,
+      "step": 20999
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9982801452321807e-06,
+      "loss": 0.837,
+      "step": 21000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9981845977450797e-06,
+      "loss": 1.1019,
+      "step": 21001
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9980890502579783e-06,
+      "loss": 0.7013,
+      "step": 21002
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9979935027708773e-06,
+      "loss": 0.5853,
+      "step": 21003
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997897955283776e-06,
+      "loss": 0.6162,
+      "step": 21004
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997802407796675e-06,
+      "loss": 0.822,
+      "step": 21005
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9977068603095745e-06,
+      "loss": 0.9626,
+      "step": 21006
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997611312822473e-06,
+      "loss": 1.0777,
+      "step": 21007
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997515765335372e-06,
+      "loss": 1.1565,
+      "step": 21008
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9974202178482707e-06,
+      "loss": 1.4242,
+      "step": 21009
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9973246703611698e-06,
+      "loss": 1.3376,
+      "step": 21010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9972291228740684e-06,
+      "loss": 1.1452,
+      "step": 21011
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9971335753869674e-06,
+      "loss": 0.7158,
+      "step": 21012
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997038027899867e-06,
+      "loss": 0.7905,
+      "step": 21013
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9969424804127655e-06,
+      "loss": 0.563,
+      "step": 21014
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9968469329256645e-06,
+      "loss": 1.0214,
+      "step": 21015
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996751385438563e-06,
+      "loss": 0.6619,
+      "step": 21016
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996655837951462e-06,
+      "loss": 0.8664,
+      "step": 21017
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996560290464361e-06,
+      "loss": 1.2392,
+      "step": 21018
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.99646474297726e-06,
+      "loss": 0.8662,
+      "step": 21019
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9963691954901585e-06,
+      "loss": 0.6635,
+      "step": 21020
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996273648003058e-06,
+      "loss": 1.0563,
+      "step": 21021
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.996178100515957e-06,
+      "loss": 0.6114,
+      "step": 21022
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9960825530288556e-06,
+      "loss": 0.5608,
+      "step": 21023
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9959870055417546e-06,
+      "loss": 0.6121,
+      "step": 21024
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9958914580546532e-06,
+      "loss": 1.107,
+      "step": 21025
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9957959105675523e-06,
+      "loss": 0.9723,
+      "step": 21026
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.995700363080451e-06,
+      "loss": 0.6793,
+      "step": 21027
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9956048155933503e-06,
+      "loss": 0.4751,
+      "step": 21028
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9955092681062494e-06,
+      "loss": 1.0369,
+      "step": 21029
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.995413720619148e-06,
+      "loss": 0.9188,
+      "step": 21030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.995318173132047e-06,
+      "loss": 0.9039,
+      "step": 21031
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9952226256449457e-06,
+      "loss": 0.9967,
+      "step": 21032
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9951270781578447e-06,
+      "loss": 0.8773,
+      "step": 21033
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9950315306707433e-06,
+      "loss": 0.9888,
+      "step": 21034
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9949359831836428e-06,
+      "loss": 0.7818,
+      "step": 21035
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9948404356965414e-06,
+      "loss": 0.7937,
+      "step": 21036
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9947448882094404e-06,
+      "loss": 0.5603,
+      "step": 21037
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9946493407223395e-06,
+      "loss": 0.4565,
+      "step": 21038
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.994553793235238e-06,
+      "loss": 0.6749,
+      "step": 21039
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.994458245748137e-06,
+      "loss": 1.3743,
+      "step": 21040
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9943626982610357e-06,
+      "loss": 0.8494,
+      "step": 21041
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.994267150773935e-06,
+      "loss": 0.8885,
+      "step": 21042
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.994171603286834e-06,
+      "loss": 1.0001,
+      "step": 21043
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.994076055799733e-06,
+      "loss": 0.7208,
+      "step": 21044
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9939805083126315e-06,
+      "loss": 0.6961,
+      "step": 21045
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9938849608255305e-06,
+      "loss": 1.1277,
+      "step": 21046
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.993789413338429e-06,
+      "loss": 1.0486,
+      "step": 21047
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.993693865851328e-06,
+      "loss": 1.0009,
+      "step": 21048
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9935983183642276e-06,
+      "loss": 0.5795,
+      "step": 21049
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9935027708771262e-06,
+      "loss": 0.8607,
+      "step": 21050
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9934072233900253e-06,
+      "loss": 0.5246,
+      "step": 21051
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.993311675902924e-06,
+      "loss": 1.1104,
+      "step": 21052
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.993216128415823e-06,
+      "loss": 0.7962,
+      "step": 21053
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9931205809287215e-06,
+      "loss": 0.7269,
+      "step": 21054
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9930250334416206e-06,
+      "loss": 0.8972,
+      "step": 21055
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992929485954519e-06,
+      "loss": 1.1156,
+      "step": 21056
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9928339384674187e-06,
+      "loss": 0.5991,
+      "step": 21057
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9927383909803177e-06,
+      "loss": 0.9092,
+      "step": 21058
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9926428434932163e-06,
+      "loss": 0.7919,
+      "step": 21059
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9925472960061153e-06,
+      "loss": 0.7145,
+      "step": 21060
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992451748519014e-06,
+      "loss": 0.8847,
+      "step": 21061
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992356201031913e-06,
+      "loss": 0.9862,
+      "step": 21062
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9922606535448116e-06,
+      "loss": 1.0246,
+      "step": 21063
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992165106057711e-06,
+      "loss": 1.1609,
+      "step": 21064
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.99206955857061e-06,
+      "loss": 0.8094,
+      "step": 21065
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9919740110835087e-06,
+      "loss": 0.5347,
+      "step": 21066
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9918784635964078e-06,
+      "loss": 0.4151,
+      "step": 21067
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9917829161093064e-06,
+      "loss": 0.6998,
+      "step": 21068
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9916873686222054e-06,
+      "loss": 0.685,
+      "step": 21069
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.991591821135104e-06,
+      "loss": 1.1648,
+      "step": 21070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9914962736480035e-06,
+      "loss": 0.9068,
+      "step": 21071
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9914007261609025e-06,
+      "loss": 0.8442,
+      "step": 21072
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.991305178673801e-06,
+      "loss": 0.8235,
+      "step": 21073
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9912096311867e-06,
+      "loss": 0.8826,
+      "step": 21074
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.991114083699599e-06,
+      "loss": 0.6959,
+      "step": 21075
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.991018536212498e-06,
+      "loss": 1.0914,
+      "step": 21076
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9909229887253965e-06,
+      "loss": 0.9552,
+      "step": 21077
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.990827441238296e-06,
+      "loss": 0.5762,
+      "step": 21078
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.990731893751195e-06,
+      "loss": 0.9904,
+      "step": 21079
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9906363462640936e-06,
+      "loss": 0.925,
+      "step": 21080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9905407987769926e-06,
+      "loss": 0.8119,
+      "step": 21081
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9904452512898912e-06,
+      "loss": 0.6659,
+      "step": 21082
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9903497038027903e-06,
+      "loss": 0.5286,
+      "step": 21083
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.990254156315689e-06,
+      "loss": 0.96,
+      "step": 21084
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9901586088285883e-06,
+      "loss": 0.6393,
+      "step": 21085
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.990063061341487e-06,
+      "loss": 0.8207,
+      "step": 21086
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.989967513854386e-06,
+      "loss": 0.5932,
+      "step": 21087
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.989871966367285e-06,
+      "loss": 0.661,
+      "step": 21088
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9897764188801837e-06,
+      "loss": 0.6947,
+      "step": 21089
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9896808713930827e-06,
+      "loss": 0.6963,
+      "step": 21090
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9895853239059813e-06,
+      "loss": 1.0148,
+      "step": 21091
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9894897764188803e-06,
+      "loss": 1.2064,
+      "step": 21092
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9893942289317794e-06,
+      "loss": 0.9408,
+      "step": 21093
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9892986814446784e-06,
+      "loss": 0.7335,
+      "step": 21094
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.989203133957577e-06,
+      "loss": 1.0548,
+      "step": 21095
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.989107586470476e-06,
+      "loss": 0.8479,
+      "step": 21096
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9890120389833747e-06,
+      "loss": 1.1033,
+      "step": 21097
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9889164914962737e-06,
+      "loss": 0.9737,
+      "step": 21098
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9888209440091723e-06,
+      "loss": 0.8778,
+      "step": 21099
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.988725396522072e-06,
+      "loss": 0.8523,
+      "step": 21100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.988629849034971e-06,
+      "loss": 0.3658,
+      "step": 21101
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9885343015478695e-06,
+      "loss": 0.7936,
+      "step": 21102
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9884387540607685e-06,
+      "loss": 0.9829,
+      "step": 21103
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.988343206573667e-06,
+      "loss": 0.6859,
+      "step": 21104
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.988247659086566e-06,
+      "loss": 0.7033,
+      "step": 21105
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9881521115994648e-06,
+      "loss": 0.6607,
+      "step": 21106
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9880565641123642e-06,
+      "loss": 0.9169,
+      "step": 21107
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9879610166252633e-06,
+      "loss": 0.5519,
+      "step": 21108
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.987865469138162e-06,
+      "loss": 0.7012,
+      "step": 21109
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.987769921651061e-06,
+      "loss": 0.7015,
+      "step": 21110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9876743741639595e-06,
+      "loss": 0.5975,
+      "step": 21111
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9875788266768586e-06,
+      "loss": 0.6227,
+      "step": 21112
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.987483279189757e-06,
+      "loss": 0.9101,
+      "step": 21113
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9873877317026567e-06,
+      "loss": 1.0799,
+      "step": 21114
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9872921842155557e-06,
+      "loss": 0.5083,
+      "step": 21115
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9871966367284543e-06,
+      "loss": 0.9003,
+      "step": 21116
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9871010892413533e-06,
+      "loss": 0.8408,
+      "step": 21117
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.987005541754252e-06,
+      "loss": 1.131,
+      "step": 21118
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986909994267151e-06,
+      "loss": 0.6362,
+      "step": 21119
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9868144467800496e-06,
+      "loss": 1.0392,
+      "step": 21120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986718899292949e-06,
+      "loss": 1.0665,
+      "step": 21121
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986623351805848e-06,
+      "loss": 0.6443,
+      "step": 21122
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9865278043187467e-06,
+      "loss": 0.7146,
+      "step": 21123
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9864322568316458e-06,
+      "loss": 1.1636,
+      "step": 21124
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9863367093445444e-06,
+      "loss": 0.482,
+      "step": 21125
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9862411618574434e-06,
+      "loss": 0.9414,
+      "step": 21126
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986145614370342e-06,
+      "loss": 0.9153,
+      "step": 21127
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.986050066883241e-06,
+      "loss": 0.8712,
+      "step": 21128
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9859545193961405e-06,
+      "loss": 0.6989,
+      "step": 21129
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.985858971909039e-06,
+      "loss": 0.8662,
+      "step": 21130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.985763424421938e-06,
+      "loss": 0.8019,
+      "step": 21131
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.985667876934837e-06,
+      "loss": 0.454,
+      "step": 21132
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.985572329447736e-06,
+      "loss": 1.2547,
+      "step": 21133
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9854767819606345e-06,
+      "loss": 1.2575,
+      "step": 21134
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9853812344735335e-06,
+      "loss": 0.4008,
+      "step": 21135
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9852856869864325e-06,
+      "loss": 0.7203,
+      "step": 21136
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9851901394993316e-06,
+      "loss": 0.8755,
+      "step": 21137
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.98509459201223e-06,
+      "loss": 1.1432,
+      "step": 21138
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9849990445251292e-06,
+      "loss": 0.7939,
+      "step": 21139
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9849034970380283e-06,
+      "loss": 0.9018,
+      "step": 21140
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984807949550927e-06,
+      "loss": 0.6533,
+      "step": 21141
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984712402063826e-06,
+      "loss": 0.7264,
+      "step": 21142
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984616854576725e-06,
+      "loss": 0.9373,
+      "step": 21143
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984521307089624e-06,
+      "loss": 0.9486,
+      "step": 21144
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9844257596025226e-06,
+      "loss": 1.3,
+      "step": 21145
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9843302121154216e-06,
+      "loss": 0.9207,
+      "step": 21146
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9842346646283203e-06,
+      "loss": 0.98,
+      "step": 21147
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9841391171412193e-06,
+      "loss": 1.0188,
+      "step": 21148
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.984043569654118e-06,
+      "loss": 0.6134,
+      "step": 21149
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9839480221670174e-06,
+      "loss": 0.5462,
+      "step": 21150
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9838524746799164e-06,
+      "loss": 0.7232,
+      "step": 21151
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.983756927192815e-06,
+      "loss": 0.5157,
+      "step": 21152
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.983661379705714e-06,
+      "loss": 0.6483,
+      "step": 21153
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9835658322186127e-06,
+      "loss": 0.627,
+      "step": 21154
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9834702847315117e-06,
+      "loss": 0.5236,
+      "step": 21155
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9833747372444103e-06,
+      "loss": 0.5613,
+      "step": 21156
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.98327918975731e-06,
+      "loss": 0.9813,
+      "step": 21157
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.983183642270209e-06,
+      "loss": 0.7209,
+      "step": 21158
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9830880947831075e-06,
+      "loss": 0.6997,
+      "step": 21159
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9829925472960065e-06,
+      "loss": 0.5794,
+      "step": 21160
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.982896999808905e-06,
+      "loss": 0.7847,
+      "step": 21161
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.982801452321804e-06,
+      "loss": 1.3825,
+      "step": 21162
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9827059048347028e-06,
+      "loss": 0.9584,
+      "step": 21163
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.982610357347602e-06,
+      "loss": 0.471,
+      "step": 21164
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9825148098605013e-06,
+      "loss": 0.5509,
+      "step": 21165
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9824192623734e-06,
+      "loss": 0.9941,
+      "step": 21166
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.982323714886299e-06,
+      "loss": 0.6735,
+      "step": 21167
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9822281673991975e-06,
+      "loss": 0.3607,
+      "step": 21168
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9821326199120966e-06,
+      "loss": 0.688,
+      "step": 21169
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.982037072424995e-06,
+      "loss": 0.6062,
+      "step": 21170
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9819415249378942e-06,
+      "loss": 0.7653,
+      "step": 21171
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9818459774507937e-06,
+      "loss": 0.4453,
+      "step": 21172
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9817504299636923e-06,
+      "loss": 0.7932,
+      "step": 21173
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9816548824765913e-06,
+      "loss": 0.7669,
+      "step": 21174
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.98155933498949e-06,
+      "loss": 0.9176,
+      "step": 21175
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.981463787502389e-06,
+      "loss": 1.1123,
+      "step": 21176
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9813682400152876e-06,
+      "loss": 0.6604,
+      "step": 21177
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9812726925281866e-06,
+      "loss": 0.8366,
+      "step": 21178
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.981177145041086e-06,
+      "loss": 1.0143,
+      "step": 21179
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9810815975539847e-06,
+      "loss": 1.1735,
+      "step": 21180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9809860500668838e-06,
+      "loss": 0.7035,
+      "step": 21181
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9808905025797824e-06,
+      "loss": 0.6784,
+      "step": 21182
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9807949550926814e-06,
+      "loss": 0.964,
+      "step": 21183
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.98069940760558e-06,
+      "loss": 0.5832,
+      "step": 21184
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.980603860118479e-06,
+      "loss": 1.1181,
+      "step": 21185
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.980508312631378e-06,
+      "loss": 0.8018,
+      "step": 21186
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.980412765144277e-06,
+      "loss": 0.9346,
+      "step": 21187
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9803172176571758e-06,
+      "loss": 1.3988,
+      "step": 21188
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.980221670170075e-06,
+      "loss": 1.0152,
+      "step": 21189
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.980126122682974e-06,
+      "loss": 0.7389,
+      "step": 21190
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9800305751958725e-06,
+      "loss": 0.7166,
+      "step": 21191
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9799350277087715e-06,
+      "loss": 1.487,
+      "step": 21192
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9798394802216705e-06,
+      "loss": 0.9017,
+      "step": 21193
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9797439327345696e-06,
+      "loss": 0.9308,
+      "step": 21194
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.979648385247468e-06,
+      "loss": 1.1498,
+      "step": 21195
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9795528377603672e-06,
+      "loss": 1.005,
+      "step": 21196
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.979457290273266e-06,
+      "loss": 0.4846,
+      "step": 21197
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.979361742786165e-06,
+      "loss": 0.8328,
+      "step": 21198
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9792661952990635e-06,
+      "loss": 0.8675,
+      "step": 21199
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9791706478119625e-06,
+      "loss": 0.7162,
+      "step": 21200
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.979075100324862e-06,
+      "loss": 0.5042,
+      "step": 21201
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9789795528377606e-06,
+      "loss": 0.6632,
+      "step": 21202
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9788840053506596e-06,
+      "loss": 0.7839,
+      "step": 21203
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9787884578635583e-06,
+      "loss": 0.6868,
+      "step": 21204
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9786929103764573e-06,
+      "loss": 0.5789,
+      "step": 21205
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.978597362889356e-06,
+      "loss": 0.9124,
+      "step": 21206
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.978501815402255e-06,
+      "loss": 0.8724,
+      "step": 21207
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9784062679151544e-06,
+      "loss": 0.8504,
+      "step": 21208
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.978310720428053e-06,
+      "loss": 1.0485,
+      "step": 21209
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.978215172940952e-06,
+      "loss": 0.636,
+      "step": 21210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9781196254538507e-06,
+      "loss": 0.9719,
+      "step": 21211
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9780240779667497e-06,
+      "loss": 0.9699,
+      "step": 21212
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9779285304796483e-06,
+      "loss": 0.7249,
+      "step": 21213
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9778329829925474e-06,
+      "loss": 1.0016,
+      "step": 21214
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.977737435505447e-06,
+      "loss": 0.9454,
+      "step": 21215
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9776418880183455e-06,
+      "loss": 0.6621,
+      "step": 21216
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9775463405312445e-06,
+      "loss": 0.4904,
+      "step": 21217
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.977450793044143e-06,
+      "loss": 1.0757,
+      "step": 21218
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.977355245557042e-06,
+      "loss": 0.6183,
+      "step": 21219
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9772596980699408e-06,
+      "loss": 0.7914,
+      "step": 21220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.97716415058284e-06,
+      "loss": 1.0429,
+      "step": 21221
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9770686030957393e-06,
+      "loss": 1.1531,
+      "step": 21222
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.976973055608638e-06,
+      "loss": 0.7531,
+      "step": 21223
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.976877508121537e-06,
+      "loss": 0.662,
+      "step": 21224
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9767819606344355e-06,
+      "loss": 1.2532,
+      "step": 21225
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9766864131473346e-06,
+      "loss": 1.0211,
+      "step": 21226
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.976590865660233e-06,
+      "loss": 0.555,
+      "step": 21227
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9764953181731322e-06,
+      "loss": 1.0229,
+      "step": 21228
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9763997706860317e-06,
+      "loss": 0.741,
+      "step": 21229
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9763042231989303e-06,
+      "loss": 0.5138,
+      "step": 21230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9762086757118293e-06,
+      "loss": 0.8797,
+      "step": 21231
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.976113128224728e-06,
+      "loss": 1.1887,
+      "step": 21232
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.976017580737627e-06,
+      "loss": 1.0335,
+      "step": 21233
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9759220332505256e-06,
+      "loss": 0.6866,
+      "step": 21234
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9758264857634246e-06,
+      "loss": 0.9377,
+      "step": 21235
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9757309382763233e-06,
+      "loss": 0.8339,
+      "step": 21236
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9756353907892227e-06,
+      "loss": 1.0571,
+      "step": 21237
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9755398433021213e-06,
+      "loss": 1.0611,
+      "step": 21238
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9754442958150204e-06,
+      "loss": 0.7949,
+      "step": 21239
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.975348748327919e-06,
+      "loss": 1.062,
+      "step": 21240
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.975253200840818e-06,
+      "loss": 0.7934,
+      "step": 21241
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.975157653353717e-06,
+      "loss": 0.7962,
+      "step": 21242
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9750621058666157e-06,
+      "loss": 0.651,
+      "step": 21243
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974966558379515e-06,
+      "loss": 1.0618,
+      "step": 21244
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9748710108924138e-06,
+      "loss": 0.6563,
+      "step": 21245
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974775463405313e-06,
+      "loss": 0.6748,
+      "step": 21246
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9746799159182114e-06,
+      "loss": 0.3988,
+      "step": 21247
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9745843684311105e-06,
+      "loss": 0.63,
+      "step": 21248
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974488820944009e-06,
+      "loss": 0.7891,
+      "step": 21249
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974393273456908e-06,
+      "loss": 0.9044,
+      "step": 21250
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9742977259698076e-06,
+      "loss": 1.0187,
+      "step": 21251
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974202178482706e-06,
+      "loss": 0.8085,
+      "step": 21252
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9741066309956052e-06,
+      "loss": 0.8269,
+      "step": 21253
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.974011083508504e-06,
+      "loss": 0.8668,
+      "step": 21254
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.973915536021403e-06,
+      "loss": 0.5924,
+      "step": 21255
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9738199885343015e-06,
+      "loss": 1.0523,
+      "step": 21256
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9737244410472005e-06,
+      "loss": 0.6575,
+      "step": 21257
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9736288935601e-06,
+      "loss": 1.1656,
+      "step": 21258
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9735333460729986e-06,
+      "loss": 1.1687,
+      "step": 21259
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9734377985858976e-06,
+      "loss": 0.5451,
+      "step": 21260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9733422510987963e-06,
+      "loss": 1.0566,
+      "step": 21261
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9732467036116953e-06,
+      "loss": 0.7803,
+      "step": 21262
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.973151156124594e-06,
+      "loss": 1.1549,
+      "step": 21263
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.973055608637493e-06,
+      "loss": 0.5844,
+      "step": 21264
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9729600611503924e-06,
+      "loss": 0.5829,
+      "step": 21265
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.972864513663291e-06,
+      "loss": 0.2491,
+      "step": 21266
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.97276896617619e-06,
+      "loss": 0.5763,
+      "step": 21267
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9726734186890887e-06,
+      "loss": 0.5673,
+      "step": 21268
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9725778712019877e-06,
+      "loss": 1.1615,
+      "step": 21269
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9724823237148863e-06,
+      "loss": 0.6786,
+      "step": 21270
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9723867762277854e-06,
+      "loss": 0.7687,
+      "step": 21271
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.972291228740684e-06,
+      "loss": 0.4798,
+      "step": 21272
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9721956812535835e-06,
+      "loss": 0.6335,
+      "step": 21273
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9721001337664825e-06,
+      "loss": 0.8109,
+      "step": 21274
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.972004586279381e-06,
+      "loss": 0.6031,
+      "step": 21275
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.97190903879228e-06,
+      "loss": 0.4758,
+      "step": 21276
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9718134913051788e-06,
+      "loss": 0.6838,
+      "step": 21277
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.971717943818078e-06,
+      "loss": 1.1244,
+      "step": 21278
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9716223963309764e-06,
+      "loss": 0.8538,
+      "step": 21279
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.971526848843876e-06,
+      "loss": 1.4179,
+      "step": 21280
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.971431301356775e-06,
+      "loss": 0.9997,
+      "step": 21281
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9713357538696735e-06,
+      "loss": 0.7859,
+      "step": 21282
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9712402063825726e-06,
+      "loss": 0.5491,
+      "step": 21283
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.971144658895471e-06,
+      "loss": 0.6756,
+      "step": 21284
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9710491114083702e-06,
+      "loss": 0.8722,
+      "step": 21285
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970953563921269e-06,
+      "loss": 0.7737,
+      "step": 21286
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9708580164341683e-06,
+      "loss": 0.9806,
+      "step": 21287
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970762468947067e-06,
+      "loss": 0.2846,
+      "step": 21288
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970666921459966e-06,
+      "loss": 1.0497,
+      "step": 21289
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9705713739728646e-06,
+      "loss": 0.5727,
+      "step": 21290
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9704758264857636e-06,
+      "loss": 0.8574,
+      "step": 21291
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9703802789986626e-06,
+      "loss": 0.8667,
+      "step": 21292
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9702847315115613e-06,
+      "loss": 1.0453,
+      "step": 21293
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9701891840244607e-06,
+      "loss": 0.7182,
+      "step": 21294
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9700936365373593e-06,
+      "loss": 0.8949,
+      "step": 21295
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9699980890502584e-06,
+      "loss": 1.2644,
+      "step": 21296
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969902541563157e-06,
+      "loss": 0.7878,
+      "step": 21297
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969806994076056e-06,
+      "loss": 0.8334,
+      "step": 21298
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9697114465889546e-06,
+      "loss": 0.9322,
+      "step": 21299
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9696158991018537e-06,
+      "loss": 0.8306,
+      "step": 21300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969520351614753e-06,
+      "loss": 1.1811,
+      "step": 21301
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9694248041276518e-06,
+      "loss": 0.8644,
+      "step": 21302
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969329256640551e-06,
+      "loss": 1.1868,
+      "step": 21303
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9692337091534494e-06,
+      "loss": 0.6699,
+      "step": 21304
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9691381616663485e-06,
+      "loss": 0.7952,
+      "step": 21305
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.969042614179247e-06,
+      "loss": 1.4093,
+      "step": 21306
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.968947066692146e-06,
+      "loss": 0.6138,
+      "step": 21307
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9688515192050447e-06,
+      "loss": 0.8676,
+      "step": 21308
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.968755971717944e-06,
+      "loss": 0.5469,
+      "step": 21309
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9686604242308432e-06,
+      "loss": 0.727,
+      "step": 21310
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.968564876743742e-06,
+      "loss": 0.7609,
+      "step": 21311
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.968469329256641e-06,
+      "loss": 1.2679,
+      "step": 21312
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9683737817695395e-06,
+      "loss": 0.9233,
+      "step": 21313
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9682782342824385e-06,
+      "loss": 0.6921,
+      "step": 21314
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.968182686795337e-06,
+      "loss": 1.0836,
+      "step": 21315
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9680871393082366e-06,
+      "loss": 1.1077,
+      "step": 21316
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9679915918211356e-06,
+      "loss": 0.938,
+      "step": 21317
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9678960443340343e-06,
+      "loss": 0.882,
+      "step": 21318
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9678004968469333e-06,
+      "loss": 0.5362,
+      "step": 21319
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.967704949359832e-06,
+      "loss": 0.8151,
+      "step": 21320
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.967609401872731e-06,
+      "loss": 0.7629,
+      "step": 21321
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9675138543856296e-06,
+      "loss": 0.6948,
+      "step": 21322
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.967418306898529e-06,
+      "loss": 0.8077,
+      "step": 21323
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.967322759411428e-06,
+      "loss": 1.3298,
+      "step": 21324
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9672272119243267e-06,
+      "loss": 0.743,
+      "step": 21325
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9671316644372257e-06,
+      "loss": 0.4855,
+      "step": 21326
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9670361169501243e-06,
+      "loss": 0.5604,
+      "step": 21327
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9669405694630234e-06,
+      "loss": 0.7764,
+      "step": 21328
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966845021975922e-06,
+      "loss": 0.797,
+      "step": 21329
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9667494744888215e-06,
+      "loss": 0.591,
+      "step": 21330
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.96665392700172e-06,
+      "loss": 0.9536,
+      "step": 21331
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966558379514619e-06,
+      "loss": 0.3791,
+      "step": 21332
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966462832027518e-06,
+      "loss": 0.9196,
+      "step": 21333
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9663672845404168e-06,
+      "loss": 0.9594,
+      "step": 21334
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966271737053316e-06,
+      "loss": 0.581,
+      "step": 21335
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9661761895662144e-06,
+      "loss": 0.6382,
+      "step": 21336
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.966080642079114e-06,
+      "loss": 0.6365,
+      "step": 21337
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9659850945920125e-06,
+      "loss": 0.6907,
+      "step": 21338
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9658895471049115e-06,
+      "loss": 0.777,
+      "step": 21339
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.96579399961781e-06,
+      "loss": 0.7776,
+      "step": 21340
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965698452130709e-06,
+      "loss": 0.6711,
+      "step": 21341
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965602904643608e-06,
+      "loss": 0.5922,
+      "step": 21342
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965507357156507e-06,
+      "loss": 0.688,
+      "step": 21343
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965411809669406e-06,
+      "loss": 0.9161,
+      "step": 21344
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.965316262182305e-06,
+      "loss": 0.8562,
+      "step": 21345
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.965220714695204e-06,
+      "loss": 1.0195,
+      "step": 21346
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9651251672081026e-06,
+      "loss": 0.5965,
+      "step": 21347
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9650296197210016e-06,
+      "loss": 0.7816,
+      "step": 21348
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9649340722339002e-06,
+      "loss": 1.2588,
+      "step": 21349
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9648385247467993e-06,
+      "loss": 0.7064,
+      "step": 21350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.964742977259698e-06,
+      "loss": 0.9521,
+      "step": 21351
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9646474297725973e-06,
+      "loss": 0.7679,
+      "step": 21352
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9645518822854964e-06,
+      "loss": 1.3513,
+      "step": 21353
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.964456334798395e-06,
+      "loss": 1.0903,
+      "step": 21354
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.964360787311294e-06,
+      "loss": 0.6614,
+      "step": 21355
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9642652398241926e-06,
+      "loss": 0.6242,
+      "step": 21356
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9641696923370917e-06,
+      "loss": 0.9442,
+      "step": 21357
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9640741448499903e-06,
+      "loss": 0.8168,
+      "step": 21358
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9639785973628898e-06,
+      "loss": 1.1574,
+      "step": 21359
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.963883049875789e-06,
+      "loss": 0.8351,
+      "step": 21360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9637875023886874e-06,
+      "loss": 0.7843,
+      "step": 21361
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9636919549015865e-06,
+      "loss": 0.674,
+      "step": 21362
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.963596407414485e-06,
+      "loss": 0.6512,
+      "step": 21363
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.963500859927384e-06,
+      "loss": 1.148,
+      "step": 21364
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9634053124402827e-06,
+      "loss": 0.9662,
+      "step": 21365
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.963309764953182e-06,
+      "loss": 0.7898,
+      "step": 21366
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9632142174660812e-06,
+      "loss": 0.8682,
+      "step": 21367
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96311866997898e-06,
+      "loss": 1.1701,
+      "step": 21368
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.963023122491879e-06,
+      "loss": 0.5541,
+      "step": 21369
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9629275750047775e-06,
+      "loss": 0.6249,
+      "step": 21370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9628320275176765e-06,
+      "loss": 0.8675,
+      "step": 21371
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.962736480030575e-06,
+      "loss": 0.8097,
+      "step": 21372
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9626409325434746e-06,
+      "loss": 0.8242,
+      "step": 21373
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9625453850563736e-06,
+      "loss": 0.8195,
+      "step": 21374
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9624498375692723e-06,
+      "loss": 0.8078,
+      "step": 21375
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9623542900821713e-06,
+      "loss": 0.7983,
+      "step": 21376
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96225874259507e-06,
+      "loss": 0.8935,
+      "step": 21377
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.962163195107969e-06,
+      "loss": 1.0103,
+      "step": 21378
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9620676476208676e-06,
+      "loss": 0.7736,
+      "step": 21379
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9619721001337666e-06,
+      "loss": 0.7351,
+      "step": 21380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9618765526466656e-06,
+      "loss": 0.3799,
+      "step": 21381
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9617810051595647e-06,
+      "loss": 0.8562,
+      "step": 21382
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9616854576724637e-06,
+      "loss": 0.9773,
+      "step": 21383
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9615899101853623e-06,
+      "loss": 0.8033,
+      "step": 21384
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9614943626982614e-06,
+      "loss": 0.7745,
+      "step": 21385
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.96139881521116e-06,
+      "loss": 0.5306,
+      "step": 21386
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.961303267724059e-06,
+      "loss": 0.6473,
+      "step": 21387
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.961207720236958e-06,
+      "loss": 1.0404,
+      "step": 21388
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.961112172749857e-06,
+      "loss": 1.0473,
+      "step": 21389
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9610166252627557e-06,
+      "loss": 0.6964,
+      "step": 21390
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9609210777756548e-06,
+      "loss": 0.6643,
+      "step": 21391
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9608255302885534e-06,
+      "loss": 0.5247,
+      "step": 21392
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9607299828014524e-06,
+      "loss": 0.6762,
+      "step": 21393
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.960634435314351e-06,
+      "loss": 1.1158,
+      "step": 21394
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9605388878272505e-06,
+      "loss": 0.9834,
+      "step": 21395
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9604433403401495e-06,
+      "loss": 0.6234,
+      "step": 21396
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.960347792853048e-06,
+      "loss": 0.7274,
+      "step": 21397
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.960252245365947e-06,
+      "loss": 0.5267,
+      "step": 21398
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.960156697878846e-06,
+      "loss": 0.6741,
+      "step": 21399
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.960061150391745e-06,
+      "loss": 0.9206,
+      "step": 21400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9599656029046435e-06,
+      "loss": 0.8399,
+      "step": 21401
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.959870055417543e-06,
+      "loss": 0.4448,
+      "step": 21402
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.959774507930442e-06,
+      "loss": 0.8937,
+      "step": 21403
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9596789604433406e-06,
+      "loss": 1.1249,
+      "step": 21404
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9595834129562396e-06,
+      "loss": 0.9872,
+      "step": 21405
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9594878654691382e-06,
+      "loss": 0.7941,
+      "step": 21406
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9593923179820373e-06,
+      "loss": 0.6283,
+      "step": 21407
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.959296770494936e-06,
+      "loss": 0.6787,
+      "step": 21408
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9592012230078353e-06,
+      "loss": 1.1126,
+      "step": 21409
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9591056755207344e-06,
+      "loss": 0.7892,
+      "step": 21410
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.959010128033633e-06,
+      "loss": 0.6515,
+      "step": 21411
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.958914580546532e-06,
+      "loss": 0.7907,
+      "step": 21412
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9588190330594306e-06,
+      "loss": 0.54,
+      "step": 21413
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9587234855723297e-06,
+      "loss": 0.559,
+      "step": 21414
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9586279380852283e-06,
+      "loss": 0.9,
+      "step": 21415
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9585323905981273e-06,
+      "loss": 0.7339,
+      "step": 21416
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.958436843111027e-06,
+      "loss": 0.5547,
+      "step": 21417
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9583412956239254e-06,
+      "loss": 1.1369,
+      "step": 21418
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9582457481368245e-06,
+      "loss": 0.6606,
+      "step": 21419
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.958150200649723e-06,
+      "loss": 0.9374,
+      "step": 21420
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.958054653162622e-06,
+      "loss": 1.1841,
+      "step": 21421
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9579591056755207e-06,
+      "loss": 0.6034,
+      "step": 21422
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9578635581884198e-06,
+      "loss": 0.9016,
+      "step": 21423
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9577680107013192e-06,
+      "loss": 0.7514,
+      "step": 21424
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.957672463214218e-06,
+      "loss": 0.7291,
+      "step": 21425
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.957576915727117e-06,
+      "loss": 1.0166,
+      "step": 21426
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9574813682400155e-06,
+      "loss": 0.9626,
+      "step": 21427
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9573858207529145e-06,
+      "loss": 0.921,
+      "step": 21428
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.957290273265813e-06,
+      "loss": 1.1839,
+      "step": 21429
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.957194725778712e-06,
+      "loss": 0.9097,
+      "step": 21430
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9570991782916112e-06,
+      "loss": 0.8825,
+      "step": 21431
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9570036308045103e-06,
+      "loss": 0.6177,
+      "step": 21432
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.956908083317409e-06,
+      "loss": 1.0367,
+      "step": 21433
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.956812535830308e-06,
+      "loss": 0.7663,
+      "step": 21434
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.956716988343207e-06,
+      "loss": 0.9134,
+      "step": 21435
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9566214408561056e-06,
+      "loss": 0.9093,
+      "step": 21436
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9565258933690046e-06,
+      "loss": 0.9253,
+      "step": 21437
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9564303458819036e-06,
+      "loss": 1.2159,
+      "step": 21438
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9563347983948027e-06,
+      "loss": 1.2177,
+      "step": 21439
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9562392509077013e-06,
+      "loss": 0.7314,
+      "step": 21440
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9561437034206003e-06,
+      "loss": 1.0266,
+      "step": 21441
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.956048155933499e-06,
+      "loss": 0.6119,
+      "step": 21442
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.955952608446398e-06,
+      "loss": 1.0475,
+      "step": 21443
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9558570609592966e-06,
+      "loss": 0.6966,
+      "step": 21444
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.955761513472196e-06,
+      "loss": 0.5125,
+      "step": 21445
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.955665965985095e-06,
+      "loss": 1.049,
+      "step": 21446
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9555704184979937e-06,
+      "loss": 0.5951,
+      "step": 21447
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9554748710108928e-06,
+      "loss": 0.593,
+      "step": 21448
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9553793235237914e-06,
+      "loss": 0.6883,
+      "step": 21449
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9552837760366904e-06,
+      "loss": 0.9415,
+      "step": 21450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.955188228549589e-06,
+      "loss": 0.8055,
+      "step": 21451
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.955092681062488e-06,
+      "loss": 0.3614,
+      "step": 21452
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9549971335753875e-06,
+      "loss": 0.436,
+      "step": 21453
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.954901586088286e-06,
+      "loss": 0.6742,
+      "step": 21454
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.954806038601185e-06,
+      "loss": 0.712,
+      "step": 21455
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.954710491114084e-06,
+      "loss": 0.5785,
+      "step": 21456
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.954614943626983e-06,
+      "loss": 0.9495,
+      "step": 21457
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9545193961398815e-06,
+      "loss": 1.2569,
+      "step": 21458
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9544238486527805e-06,
+      "loss": 1.0453,
+      "step": 21459
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.95432830116568e-06,
+      "loss": 0.6506,
+      "step": 21460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9542327536785786e-06,
+      "loss": 0.7289,
+      "step": 21461
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9541372061914776e-06,
+      "loss": 0.9514,
+      "step": 21462
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9540416587043762e-06,
+      "loss": 1.1231,
+      "step": 21463
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9539461112172753e-06,
+      "loss": 0.9976,
+      "step": 21464
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.953850563730174e-06,
+      "loss": 1.3401,
+      "step": 21465
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.953755016243073e-06,
+      "loss": 0.7739,
+      "step": 21466
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9536594687559724e-06,
+      "loss": 0.5919,
+      "step": 21467
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.953563921268871e-06,
+      "loss": 0.6745,
+      "step": 21468
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.95346837378177e-06,
+      "loss": 0.7115,
+      "step": 21469
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9533728262946686e-06,
+      "loss": 1.1079,
+      "step": 21470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9532772788075677e-06,
+      "loss": 0.9868,
+      "step": 21471
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9531817313204663e-06,
+      "loss": 1.0235,
+      "step": 21472
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9530861838333653e-06,
+      "loss": 0.5419,
+      "step": 21473
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.952990636346265e-06,
+      "loss": 0.945,
+      "step": 21474
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9528950888591634e-06,
+      "loss": 0.5081,
+      "step": 21475
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9527995413720625e-06,
+      "loss": 0.66,
+      "step": 21476
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.952703993884961e-06,
+      "loss": 0.6894,
+      "step": 21477
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.95260844639786e-06,
+      "loss": 0.874,
+      "step": 21478
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9525128989107587e-06,
+      "loss": 0.6384,
+      "step": 21479
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9524173514236578e-06,
+      "loss": 0.9061,
+      "step": 21480
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.952321803936557e-06,
+      "loss": 0.9081,
+      "step": 21481
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.952226256449456e-06,
+      "loss": 0.571,
+      "step": 21482
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9521307089623545e-06,
+      "loss": 1.0802,
+      "step": 21483
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9520351614752535e-06,
+      "loss": 0.5899,
+      "step": 21484
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9519396139881525e-06,
+      "loss": 0.6852,
+      "step": 21485
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.951844066501051e-06,
+      "loss": 0.8304,
+      "step": 21486
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.95174851901395e-06,
+      "loss": 0.6177,
+      "step": 21487
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.951652971526849e-06,
+      "loss": 0.5668,
+      "step": 21488
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9515574240397483e-06,
+      "loss": 1.0291,
+      "step": 21489
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.951461876552647e-06,
+      "loss": 0.9171,
+      "step": 21490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.951366329065546e-06,
+      "loss": 1.1018,
+      "step": 21491
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9512707815784445e-06,
+      "loss": 0.9517,
+      "step": 21492
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9511752340913436e-06,
+      "loss": 0.8749,
+      "step": 21493
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.951079686604242e-06,
+      "loss": 0.7214,
+      "step": 21494
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9509841391171412e-06,
+      "loss": 0.6116,
+      "step": 21495
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9508885916300407e-06,
+      "loss": 1.1468,
+      "step": 21496
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9507930441429393e-06,
+      "loss": 0.9273,
+      "step": 21497
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9506974966558383e-06,
+      "loss": 0.9274,
+      "step": 21498
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.950601949168737e-06,
+      "loss": 0.7632,
+      "step": 21499
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.950506401681636e-06,
+      "loss": 0.8865,
+      "step": 21500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9504108541945346e-06,
+      "loss": 0.9487,
+      "step": 21501
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9503153067074336e-06,
+      "loss": 1.0219,
+      "step": 21502
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.950219759220333e-06,
+      "loss": 0.6475,
+      "step": 21503
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9501242117332317e-06,
+      "loss": 1.0692,
+      "step": 21504
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9500286642461308e-06,
+      "loss": 0.4492,
+      "step": 21505
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9499331167590294e-06,
+      "loss": 0.8282,
+      "step": 21506
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9498375692719284e-06,
+      "loss": 0.8386,
+      "step": 21507
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949742021784827e-06,
+      "loss": 0.9295,
+      "step": 21508
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949646474297726e-06,
+      "loss": 0.6675,
+      "step": 21509
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9495509268106255e-06,
+      "loss": 0.8251,
+      "step": 21510
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949455379323524e-06,
+      "loss": 0.7734,
+      "step": 21511
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949359831836423e-06,
+      "loss": 0.975,
+      "step": 21512
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949264284349322e-06,
+      "loss": 1.2199,
+      "step": 21513
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.949168736862221e-06,
+      "loss": 0.6033,
+      "step": 21514
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9490731893751195e-06,
+      "loss": 0.9341,
+      "step": 21515
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9489776418880185e-06,
+      "loss": 0.8686,
+      "step": 21516
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948882094400918e-06,
+      "loss": 0.6761,
+      "step": 21517
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9487865469138166e-06,
+      "loss": 0.8873,
+      "step": 21518
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9486909994267156e-06,
+      "loss": 0.9229,
+      "step": 21519
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9485954519396142e-06,
+      "loss": 0.8945,
+      "step": 21520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9484999044525133e-06,
+      "loss": 1.0794,
+      "step": 21521
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948404356965412e-06,
+      "loss": 0.4373,
+      "step": 21522
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948308809478311e-06,
+      "loss": 0.7649,
+      "step": 21523
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9482132619912095e-06,
+      "loss": 0.8785,
+      "step": 21524
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948117714504109e-06,
+      "loss": 0.9343,
+      "step": 21525
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948022167017008e-06,
+      "loss": 0.9383,
+      "step": 21526
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9479266195299066e-06,
+      "loss": 0.5605,
+      "step": 21527
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9478310720428057e-06,
+      "loss": 0.6256,
+      "step": 21528
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9477355245557043e-06,
+      "loss": 0.7744,
+      "step": 21529
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9476399770686033e-06,
+      "loss": 0.7432,
+      "step": 21530
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.947544429581502e-06,
+      "loss": 0.9515,
+      "step": 21531
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9474488820944014e-06,
+      "loss": 1.1892,
+      "step": 21532
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9473533346073e-06,
+      "loss": 1.1319,
+      "step": 21533
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.947257787120199e-06,
+      "loss": 0.6651,
+      "step": 21534
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9471622396330977e-06,
+      "loss": 0.9403,
+      "step": 21535
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9470666921459967e-06,
+      "loss": 0.8032,
+      "step": 21536
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9469711446588958e-06,
+      "loss": 0.7699,
+      "step": 21537
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9468755971717944e-06,
+      "loss": 0.7224,
+      "step": 21538
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.946780049684694e-06,
+      "loss": 0.9544,
+      "step": 21539
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9466845021975924e-06,
+      "loss": 0.6783,
+      "step": 21540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9465889547104915e-06,
+      "loss": 0.7065,
+      "step": 21541
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.94649340722339e-06,
+      "loss": 0.5042,
+      "step": 21542
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.946397859736289e-06,
+      "loss": 0.732,
+      "step": 21543
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9463023122491878e-06,
+      "loss": 0.9895,
+      "step": 21544
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.946206764762087e-06,
+      "loss": 1.0825,
+      "step": 21545
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9461112172749863e-06,
+      "loss": 0.7725,
+      "step": 21546
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.946015669787885e-06,
+      "loss": 0.8768,
+      "step": 21547
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.945920122300784e-06,
+      "loss": 0.702,
+      "step": 21548
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9458245748136825e-06,
+      "loss": 0.8705,
+      "step": 21549
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9457290273265816e-06,
+      "loss": 0.655,
+      "step": 21550
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.94563347983948e-06,
+      "loss": 0.61,
+      "step": 21551
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9455379323523792e-06,
+      "loss": 0.7785,
+      "step": 21552
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9454423848652787e-06,
+      "loss": 0.9773,
+      "step": 21553
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9453468373781773e-06,
+      "loss": 0.978,
+      "step": 21554
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9452512898910763e-06,
+      "loss": 0.733,
+      "step": 21555
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.945155742403975e-06,
+      "loss": 0.8354,
+      "step": 21556
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.945060194916874e-06,
+      "loss": 0.5081,
+      "step": 21557
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9449646474297726e-06,
+      "loss": 1.1779,
+      "step": 21558
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9448690999426716e-06,
+      "loss": 1.0392,
+      "step": 21559
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9447735524555703e-06,
+      "loss": 0.7979,
+      "step": 21560
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9446780049684697e-06,
+      "loss": 0.874,
+      "step": 21561
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9445824574813688e-06,
+      "loss": 0.5901,
+      "step": 21562
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9444869099942674e-06,
+      "loss": 0.6256,
+      "step": 21563
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9443913625071664e-06,
+      "loss": 0.7239,
+      "step": 21564
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.944295815020065e-06,
+      "loss": 0.6011,
+      "step": 21565
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.944200267532964e-06,
+      "loss": 0.9429,
+      "step": 21566
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9441047200458627e-06,
+      "loss": 0.4832,
+      "step": 21567
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.944009172558762e-06,
+      "loss": 0.7727,
+      "step": 21568
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.943913625071661e-06,
+      "loss": 1.0549,
+      "step": 21569
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.94381807758456e-06,
+      "loss": 0.9899,
+      "step": 21570
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.943722530097459e-06,
+      "loss": 0.7959,
+      "step": 21571
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9436269826103574e-06,
+      "loss": 0.4808,
+      "step": 21572
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9435314351232565e-06,
+      "loss": 0.8396,
+      "step": 21573
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.943435887636155e-06,
+      "loss": 0.6885,
+      "step": 21574
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9433403401490546e-06,
+      "loss": 0.6888,
+      "step": 21575
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9432447926619536e-06,
+      "loss": 1.0418,
+      "step": 21576
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9431492451748522e-06,
+      "loss": 0.9863,
+      "step": 21577
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9430536976877513e-06,
+      "loss": 0.7463,
+      "step": 21578
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.94295815020065e-06,
+      "loss": 0.635,
+      "step": 21579
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.942862602713549e-06,
+      "loss": 1.6385,
+      "step": 21580
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9427670552264475e-06,
+      "loss": 0.7653,
+      "step": 21581
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.942671507739347e-06,
+      "loss": 0.8965,
+      "step": 21582
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9425759602522456e-06,
+      "loss": 0.87,
+      "step": 21583
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9424804127651446e-06,
+      "loss": 1.1041,
+      "step": 21584
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9423848652780433e-06,
+      "loss": 1.1458,
+      "step": 21585
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9422893177909423e-06,
+      "loss": 0.9088,
+      "step": 21586
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9421937703038413e-06,
+      "loss": 1.049,
+      "step": 21587
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.94209822281674e-06,
+      "loss": 0.5671,
+      "step": 21588
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9420026753296394e-06,
+      "loss": 0.7536,
+      "step": 21589
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.941907127842538e-06,
+      "loss": 1.0351,
+      "step": 21590
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.941811580355437e-06,
+      "loss": 0.6236,
+      "step": 21591
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9417160328683357e-06,
+      "loss": 0.8619,
+      "step": 21592
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9416204853812347e-06,
+      "loss": 0.5814,
+      "step": 21593
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9415249378941333e-06,
+      "loss": 0.611,
+      "step": 21594
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9414293904070324e-06,
+      "loss": 0.7931,
+      "step": 21595
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.941333842919931e-06,
+      "loss": 0.7505,
+      "step": 21596
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9412382954328304e-06,
+      "loss": 0.5771,
+      "step": 21597
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9411427479457295e-06,
+      "loss": 0.5904,
+      "step": 21598
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.941047200458628e-06,
+      "loss": 1.1218,
+      "step": 21599
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.940951652971527e-06,
+      "loss": 0.9014,
+      "step": 21600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9408561054844258e-06,
+      "loss": 1.1853,
+      "step": 21601
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.940760557997325e-06,
+      "loss": 0.7945,
+      "step": 21602
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9406650105102234e-06,
+      "loss": 1.0361,
+      "step": 21603
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.940569463023123e-06,
+      "loss": 0.694,
+      "step": 21604
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.940473915536022e-06,
+      "loss": 0.8546,
+      "step": 21605
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9403783680489205e-06,
+      "loss": 0.9062,
+      "step": 21606
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9402828205618196e-06,
+      "loss": 0.8037,
+      "step": 21607
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.940187273074718e-06,
+      "loss": 0.872,
+      "step": 21608
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9400917255876172e-06,
+      "loss": 1.0234,
+      "step": 21609
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.939996178100516e-06,
+      "loss": 1.1869,
+      "step": 21610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9399006306134153e-06,
+      "loss": 1.0778,
+      "step": 21611
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9398050831263143e-06,
+      "loss": 1.1639,
+      "step": 21612
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.939709535639213e-06,
+      "loss": 1.0793,
+      "step": 21613
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.939613988152112e-06,
+      "loss": 0.5185,
+      "step": 21614
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9395184406650106e-06,
+      "loss": 0.615,
+      "step": 21615
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9394228931779096e-06,
+      "loss": 0.8732,
+      "step": 21616
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9393273456908083e-06,
+      "loss": 0.6314,
+      "step": 21617
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9392317982037077e-06,
+      "loss": 0.997,
+      "step": 21618
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9391362507166068e-06,
+      "loss": 0.8563,
+      "step": 21619
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9390407032295054e-06,
+      "loss": 0.8572,
+      "step": 21620
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9389451557424044e-06,
+      "loss": 0.7526,
+      "step": 21621
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.938849608255303e-06,
+      "loss": 0.9592,
+      "step": 21622
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.938754060768202e-06,
+      "loss": 0.6903,
+      "step": 21623
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9386585132811007e-06,
+      "loss": 0.8711,
+      "step": 21624
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.938562965794e-06,
+      "loss": 0.7315,
+      "step": 21625
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9384674183068988e-06,
+      "loss": 0.6828,
+      "step": 21626
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.938371870819798e-06,
+      "loss": 0.764,
+      "step": 21627
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.938276323332697e-06,
+      "loss": 0.8267,
+      "step": 21628
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9381807758455954e-06,
+      "loss": 0.7241,
+      "step": 21629
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9380852283584945e-06,
+      "loss": 1.0931,
+      "step": 21630
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937989680871393e-06,
+      "loss": 0.6299,
+      "step": 21631
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937894133384292e-06,
+      "loss": 1.3242,
+      "step": 21632
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937798585897191e-06,
+      "loss": 0.4814,
+      "step": 21633
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9377030384100902e-06,
+      "loss": 0.5559,
+      "step": 21634
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937607490922989e-06,
+      "loss": 0.6964,
+      "step": 21635
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937511943435888e-06,
+      "loss": 0.5485,
+      "step": 21636
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9374163959487865e-06,
+      "loss": 0.8655,
+      "step": 21637
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9373208484616855e-06,
+      "loss": 0.6054,
+      "step": 21638
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9372253009745846e-06,
+      "loss": 1.164,
+      "step": 21639
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9371297534874836e-06,
+      "loss": 0.6876,
+      "step": 21640
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9370342060003826e-06,
+      "loss": 1.05,
+      "step": 21641
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9369386585132813e-06,
+      "loss": 0.8625,
+      "step": 21642
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9368431110261803e-06,
+      "loss": 0.5791,
+      "step": 21643
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.936747563539079e-06,
+      "loss": 0.6613,
+      "step": 21644
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.936652016051978e-06,
+      "loss": 0.9817,
+      "step": 21645
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9365564685648766e-06,
+      "loss": 0.6747,
+      "step": 21646
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.936460921077776e-06,
+      "loss": 0.645,
+      "step": 21647
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.936365373590675e-06,
+      "loss": 1.0271,
+      "step": 21648
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9362698261035737e-06,
+      "loss": 1.1801,
+      "step": 21649
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9361742786164727e-06,
+      "loss": 1.3406,
+      "step": 21650
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9360787311293713e-06,
+      "loss": 1.4002,
+      "step": 21651
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9359831836422704e-06,
+      "loss": 0.7702,
+      "step": 21652
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.935887636155169e-06,
+      "loss": 0.5477,
+      "step": 21653
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9357920886680684e-06,
+      "loss": 1.0042,
+      "step": 21654
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9356965411809675e-06,
+      "loss": 0.6672,
+      "step": 21655
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.935600993693866e-06,
+      "loss": 0.7177,
+      "step": 21656
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.935505446206765e-06,
+      "loss": 1.1453,
+      "step": 21657
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9354098987196638e-06,
+      "loss": 0.635,
+      "step": 21658
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.935314351232563e-06,
+      "loss": 0.6608,
+      "step": 21659
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9352188037454614e-06,
+      "loss": 0.5773,
+      "step": 21660
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.935123256258361e-06,
+      "loss": 0.7735,
+      "step": 21661
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.93502770877126e-06,
+      "loss": 0.7735,
+      "step": 21662
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9349321612841585e-06,
+      "loss": 0.6943,
+      "step": 21663
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9348366137970576e-06,
+      "loss": 0.7192,
+      "step": 21664
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.934741066309956e-06,
+      "loss": 0.7098,
+      "step": 21665
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9346455188228552e-06,
+      "loss": 0.7592,
+      "step": 21666
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.934549971335754e-06,
+      "loss": 0.5151,
+      "step": 21667
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.934454423848653e-06,
+      "loss": 0.7666,
+      "step": 21668
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9343588763615523e-06,
+      "loss": 0.6733,
+      "step": 21669
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.934263328874451e-06,
+      "loss": 0.6845,
+      "step": 21670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.93416778138735e-06,
+      "loss": 0.7235,
+      "step": 21671
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9340722339002486e-06,
+      "loss": 1.1809,
+      "step": 21672
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9339766864131476e-06,
+      "loss": 1.2127,
+      "step": 21673
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9338811389260463e-06,
+      "loss": 1.2804,
+      "step": 21674
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9337855914389453e-06,
+      "loss": 1.1162,
+      "step": 21675
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9336900439518443e-06,
+      "loss": 0.6651,
+      "step": 21676
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9335944964647434e-06,
+      "loss": 0.8646,
+      "step": 21677
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9334989489776424e-06,
+      "loss": 0.5589,
+      "step": 21678
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.933403401490541e-06,
+      "loss": 0.7084,
+      "step": 21679
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.93330785400344e-06,
+      "loss": 0.7398,
+      "step": 21680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9332123065163387e-06,
+      "loss": 1.0391,
+      "step": 21681
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9331167590292377e-06,
+      "loss": 0.511,
+      "step": 21682
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9330212115421368e-06,
+      "loss": 0.8559,
+      "step": 21683
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932925664055036e-06,
+      "loss": 0.8966,
+      "step": 21684
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9328301165679344e-06,
+      "loss": 1.0075,
+      "step": 21685
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9327345690808334e-06,
+      "loss": 0.8752,
+      "step": 21686
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932639021593732e-06,
+      "loss": 1.132,
+      "step": 21687
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932543474106631e-06,
+      "loss": 0.5707,
+      "step": 21688
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.93244792661953e-06,
+      "loss": 1.2497,
+      "step": 21689
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932352379132429e-06,
+      "loss": 0.7448,
+      "step": 21690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9322568316453282e-06,
+      "loss": 0.7664,
+      "step": 21691
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932161284158227e-06,
+      "loss": 0.6534,
+      "step": 21692
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.932065736671126e-06,
+      "loss": 0.7569,
+      "step": 21693
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9319701891840245e-06,
+      "loss": 0.9462,
+      "step": 21694
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9318746416969235e-06,
+      "loss": 0.7776,
+      "step": 21695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.931779094209822e-06,
+      "loss": 0.757,
+      "step": 21696
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9316835467227216e-06,
+      "loss": 0.7804,
+      "step": 21697
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9315879992356206e-06,
+      "loss": 0.9553,
+      "step": 21698
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9314924517485193e-06,
+      "loss": 0.7772,
+      "step": 21699
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9313969042614183e-06,
+      "loss": 1.0335,
+      "step": 21700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.931301356774317e-06,
+      "loss": 0.9141,
+      "step": 21701
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.931205809287216e-06,
+      "loss": 0.6775,
+      "step": 21702
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9311102618001146e-06,
+      "loss": 0.4702,
+      "step": 21703
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9310147143130136e-06,
+      "loss": 0.898,
+      "step": 21704
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.930919166825913e-06,
+      "loss": 0.6662,
+      "step": 21705
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9308236193388117e-06,
+      "loss": 0.8944,
+      "step": 21706
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9307280718517107e-06,
+      "loss": 1.1021,
+      "step": 21707
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9306325243646093e-06,
+      "loss": 1.1103,
+      "step": 21708
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9305369768775084e-06,
+      "loss": 0.8587,
+      "step": 21709
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.930441429390407e-06,
+      "loss": 0.9421,
+      "step": 21710
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.930345881903306e-06,
+      "loss": 0.9721,
+      "step": 21711
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9302503344162055e-06,
+      "loss": 0.9148,
+      "step": 21712
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.930154786929104e-06,
+      "loss": 1.1244,
+      "step": 21713
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.930059239442003e-06,
+      "loss": 0.6234,
+      "step": 21714
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9299636919549018e-06,
+      "loss": 0.7781,
+      "step": 21715
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.929868144467801e-06,
+      "loss": 1.2333,
+      "step": 21716
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9297725969806994e-06,
+      "loss": 0.6584,
+      "step": 21717
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9296770494935984e-06,
+      "loss": 0.8933,
+      "step": 21718
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.929581502006498e-06,
+      "loss": 0.7309,
+      "step": 21719
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9294859545193965e-06,
+      "loss": 0.8281,
+      "step": 21720
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9293904070322956e-06,
+      "loss": 0.7018,
+      "step": 21721
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.929294859545194e-06,
+      "loss": 0.8643,
+      "step": 21722
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9291993120580932e-06,
+      "loss": 1.0173,
+      "step": 21723
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.929103764570992e-06,
+      "loss": 1.1936,
+      "step": 21724
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.929008217083891e-06,
+      "loss": 0.7597,
+      "step": 21725
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.92891266959679e-06,
+      "loss": 0.5902,
+      "step": 21726
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.928817122109689e-06,
+      "loss": 0.8305,
+      "step": 21727
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9287215746225876e-06,
+      "loss": 1.0824,
+      "step": 21728
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9286260271354866e-06,
+      "loss": 0.7854,
+      "step": 21729
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9285304796483856e-06,
+      "loss": 1.1477,
+      "step": 21730
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9284349321612843e-06,
+      "loss": 0.6366,
+      "step": 21731
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9283393846741833e-06,
+      "loss": 0.8658,
+      "step": 21732
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9282438371870823e-06,
+      "loss": 1.0687,
+      "step": 21733
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9281482896999814e-06,
+      "loss": 1.0131,
+      "step": 21734
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.92805274221288e-06,
+      "loss": 0.7956,
+      "step": 21735
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.927957194725779e-06,
+      "loss": 0.7031,
+      "step": 21736
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9278616472386776e-06,
+      "loss": 0.8823,
+      "step": 21737
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9277660997515767e-06,
+      "loss": 0.919,
+      "step": 21738
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9276705522644753e-06,
+      "loss": 0.9363,
+      "step": 21739
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9275750047773743e-06,
+      "loss": 0.6061,
+      "step": 21740
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.927479457290274e-06,
+      "loss": 0.8724,
+      "step": 21741
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9273839098031724e-06,
+      "loss": 0.4785,
+      "step": 21742
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9272883623160714e-06,
+      "loss": 0.7183,
+      "step": 21743
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.92719281482897e-06,
+      "loss": 1.0322,
+      "step": 21744
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.927097267341869e-06,
+      "loss": 0.7329,
+      "step": 21745
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9270017198547677e-06,
+      "loss": 0.4629,
+      "step": 21746
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9269061723676668e-06,
+      "loss": 0.4442,
+      "step": 21747
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9268106248805662e-06,
+      "loss": 0.8981,
+      "step": 21748
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.926715077393465e-06,
+      "loss": 0.7611,
+      "step": 21749
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.926619529906364e-06,
+      "loss": 0.6896,
+      "step": 21750
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9265239824192625e-06,
+      "loss": 0.7803,
+      "step": 21751
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9264284349321615e-06,
+      "loss": 0.7602,
+      "step": 21752
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.92633288744506e-06,
+      "loss": 1.0345,
+      "step": 21753
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.926237339957959e-06,
+      "loss": 0.9967,
+      "step": 21754
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9261417924708586e-06,
+      "loss": 0.6211,
+      "step": 21755
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9260462449837573e-06,
+      "loss": 0.8293,
+      "step": 21756
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9259506974966563e-06,
+      "loss": 0.8613,
+      "step": 21757
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.925855150009555e-06,
+      "loss": 0.9326,
+      "step": 21758
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.925759602522454e-06,
+      "loss": 1.1197,
+      "step": 21759
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9256640550353526e-06,
+      "loss": 1.4326,
+      "step": 21760
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9255685075482516e-06,
+      "loss": 0.6734,
+      "step": 21761
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.925472960061151e-06,
+      "loss": 0.5777,
+      "step": 21762
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9253774125740497e-06,
+      "loss": 0.7336,
+      "step": 21763
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9252818650869487e-06,
+      "loss": 0.9669,
+      "step": 21764
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9251863175998473e-06,
+      "loss": 0.7849,
+      "step": 21765
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9250907701127464e-06,
+      "loss": 0.9802,
+      "step": 21766
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924995222625645e-06,
+      "loss": 0.8291,
+      "step": 21767
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924899675138544e-06,
+      "loss": 1.4621,
+      "step": 21768
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9248041276514435e-06,
+      "loss": 1.1723,
+      "step": 21769
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924708580164342e-06,
+      "loss": 0.7465,
+      "step": 21770
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924613032677241e-06,
+      "loss": 1.0926,
+      "step": 21771
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9245174851901398e-06,
+      "loss": 1.1498,
+      "step": 21772
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924421937703039e-06,
+      "loss": 1.079,
+      "step": 21773
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9243263902159374e-06,
+      "loss": 1.132,
+      "step": 21774
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9242308427288364e-06,
+      "loss": 0.6489,
+      "step": 21775
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.924135295241735e-06,
+      "loss": 0.6567,
+      "step": 21776
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9240397477546345e-06,
+      "loss": 0.583,
+      "step": 21777
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.923944200267533e-06,
+      "loss": 1.1646,
+      "step": 21778
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.923848652780432e-06,
+      "loss": 0.5927,
+      "step": 21779
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9237531052933312e-06,
+      "loss": 1.2999,
+      "step": 21780
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.92365755780623e-06,
+      "loss": 0.8363,
+      "step": 21781
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.923562010319129e-06,
+      "loss": 0.6978,
+      "step": 21782
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9234664628320275e-06,
+      "loss": 0.6931,
+      "step": 21783
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.923370915344927e-06,
+      "loss": 0.6247,
+      "step": 21784
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9232753678578256e-06,
+      "loss": 1.088,
+      "step": 21785
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9231798203707246e-06,
+      "loss": 0.8293,
+      "step": 21786
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9230842728836232e-06,
+      "loss": 1.0388,
+      "step": 21787
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9229887253965223e-06,
+      "loss": 0.7916,
+      "step": 21788
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.922893177909421e-06,
+      "loss": 0.826,
+      "step": 21789
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.92279763042232e-06,
+      "loss": 0.6931,
+      "step": 21790
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9227020829352194e-06,
+      "loss": 0.6226,
+      "step": 21791
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.922606535448118e-06,
+      "loss": 0.6,
+      "step": 21792
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.922510987961017e-06,
+      "loss": 0.8338,
+      "step": 21793
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9224154404739156e-06,
+      "loss": 1.2035,
+      "step": 21794
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9223198929868147e-06,
+      "loss": 0.816,
+      "step": 21795
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9222243454997133e-06,
+      "loss": 0.9366,
+      "step": 21796
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9221287980126123e-06,
+      "loss": 0.933,
+      "step": 21797
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.922033250525512e-06,
+      "loss": 0.9661,
+      "step": 21798
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9219377030384104e-06,
+      "loss": 0.6809,
+      "step": 21799
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9218421555513094e-06,
+      "loss": 0.5884,
+      "step": 21800
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.921746608064208e-06,
+      "loss": 0.5846,
+      "step": 21801
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.921651060577107e-06,
+      "loss": 0.9187,
+      "step": 21802
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9215555130900057e-06,
+      "loss": 0.9708,
+      "step": 21803
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9214599656029048e-06,
+      "loss": 0.756,
+      "step": 21804
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9213644181158042e-06,
+      "loss": 0.7817,
+      "step": 21805
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.921268870628703e-06,
+      "loss": 0.6761,
+      "step": 21806
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.921173323141602e-06,
+      "loss": 0.9475,
+      "step": 21807
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9210777756545005e-06,
+      "loss": 0.6001,
+      "step": 21808
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9209822281673995e-06,
+      "loss": 0.6232,
+      "step": 21809
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920886680680298e-06,
+      "loss": 0.8837,
+      "step": 21810
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920791133193197e-06,
+      "loss": 0.8895,
+      "step": 21811
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920695585706096e-06,
+      "loss": 0.934,
+      "step": 21812
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9206000382189953e-06,
+      "loss": 0.8674,
+      "step": 21813
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9205044907318943e-06,
+      "loss": 0.8869,
+      "step": 21814
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920408943244793e-06,
+      "loss": 0.7984,
+      "step": 21815
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920313395757692e-06,
+      "loss": 0.9027,
+      "step": 21816
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9202178482705906e-06,
+      "loss": 0.6344,
+      "step": 21817
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9201223007834896e-06,
+      "loss": 0.862,
+      "step": 21818
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9200267532963882e-06,
+      "loss": 1.0537,
+      "step": 21819
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9199312058092877e-06,
+      "loss": 0.8599,
+      "step": 21820
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9198356583221867e-06,
+      "loss": 1.1238,
+      "step": 21821
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9197401108350853e-06,
+      "loss": 0.4935,
+      "step": 21822
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9196445633479844e-06,
+      "loss": 0.5444,
+      "step": 21823
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.919549015860883e-06,
+      "loss": 0.8217,
+      "step": 21824
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.919453468373782e-06,
+      "loss": 0.7497,
+      "step": 21825
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9193579208866806e-06,
+      "loss": 1.2269,
+      "step": 21826
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.91926237339958e-06,
+      "loss": 0.7521,
+      "step": 21827
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9191668259124787e-06,
+      "loss": 0.7191,
+      "step": 21828
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9190712784253778e-06,
+      "loss": 0.6501,
+      "step": 21829
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9189757309382764e-06,
+      "loss": 1.0287,
+      "step": 21830
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9188801834511754e-06,
+      "loss": 1.3998,
+      "step": 21831
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9187846359640744e-06,
+      "loss": 0.6383,
+      "step": 21832
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.918689088476973e-06,
+      "loss": 0.9036,
+      "step": 21833
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9185935409898725e-06,
+      "loss": 0.7707,
+      "step": 21834
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.918497993502771e-06,
+      "loss": 0.7034,
+      "step": 21835
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.91840244601567e-06,
+      "loss": 1.2283,
+      "step": 21836
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.918306898528569e-06,
+      "loss": 0.4616,
+      "step": 21837
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.918211351041468e-06,
+      "loss": 0.5798,
+      "step": 21838
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9181158035543664e-06,
+      "loss": 0.9046,
+      "step": 21839
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9180202560672655e-06,
+      "loss": 1.0997,
+      "step": 21840
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917924708580165e-06,
+      "loss": 0.729,
+      "step": 21841
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9178291610930636e-06,
+      "loss": 0.8369,
+      "step": 21842
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9177336136059626e-06,
+      "loss": 0.7132,
+      "step": 21843
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9176380661188612e-06,
+      "loss": 0.5899,
+      "step": 21844
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9175425186317603e-06,
+      "loss": 0.7012,
+      "step": 21845
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917446971144659e-06,
+      "loss": 0.5822,
+      "step": 21846
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917351423657558e-06,
+      "loss": 1.0888,
+      "step": 21847
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9172558761704565e-06,
+      "loss": 0.4179,
+      "step": 21848
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917160328683356e-06,
+      "loss": 1.1856,
+      "step": 21849
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917064781196255e-06,
+      "loss": 1.2365,
+      "step": 21850
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9169692337091536e-06,
+      "loss": 1.0552,
+      "step": 21851
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9168736862220527e-06,
+      "loss": 1.0084,
+      "step": 21852
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9167781387349513e-06,
+      "loss": 0.6083,
+      "step": 21853
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9166825912478503e-06,
+      "loss": 0.9233,
+      "step": 21854
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.916587043760749e-06,
+      "loss": 0.5783,
+      "step": 21855
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9164914962736484e-06,
+      "loss": 0.6708,
+      "step": 21856
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9163959487865474e-06,
+      "loss": 0.7751,
+      "step": 21857
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.916300401299446e-06,
+      "loss": 1.0043,
+      "step": 21858
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.916204853812345e-06,
+      "loss": 0.4142,
+      "step": 21859
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9161093063252437e-06,
+      "loss": 0.8521,
+      "step": 21860
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9160137588381428e-06,
+      "loss": 0.9988,
+      "step": 21861
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9159182113510414e-06,
+      "loss": 0.619,
+      "step": 21862
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.915822663863941e-06,
+      "loss": 0.627,
+      "step": 21863
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.91572711637684e-06,
+      "loss": 0.8,
+      "step": 21864
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9156315688897385e-06,
+      "loss": 0.5539,
+      "step": 21865
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9155360214026375e-06,
+      "loss": 0.8582,
+      "step": 21866
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.915440473915536e-06,
+      "loss": 0.6162,
+      "step": 21867
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.915344926428435e-06,
+      "loss": 0.4865,
+      "step": 21868
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.915249378941334e-06,
+      "loss": 0.7576,
+      "step": 21869
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9151538314542333e-06,
+      "loss": 0.6572,
+      "step": 21870
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9150582839671323e-06,
+      "loss": 0.7539,
+      "step": 21871
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.914962736480031e-06,
+      "loss": 0.7611,
+      "step": 21872
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.91486718899293e-06,
+      "loss": 0.8223,
+      "step": 21873
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9147716415058286e-06,
+      "loss": 0.4616,
+      "step": 21874
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9146760940187276e-06,
+      "loss": 0.6741,
+      "step": 21875
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9145805465316262e-06,
+      "loss": 0.7497,
+      "step": 21876
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9144849990445257e-06,
+      "loss": 0.9432,
+      "step": 21877
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9143894515574243e-06,
+      "loss": 0.625,
+      "step": 21878
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9142939040703233e-06,
+      "loss": 1.0816,
+      "step": 21879
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.914198356583222e-06,
+      "loss": 0.7373,
+      "step": 21880
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.914102809096121e-06,
+      "loss": 0.9752,
+      "step": 21881
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.91400726160902e-06,
+      "loss": 0.9903,
+      "step": 21882
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9139117141219186e-06,
+      "loss": 1.2767,
+      "step": 21883
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9138161666348177e-06,
+      "loss": 0.817,
+      "step": 21884
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9137206191477167e-06,
+      "loss": 1.0745,
+      "step": 21885
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9136250716606158e-06,
+      "loss": 0.4575,
+      "step": 21886
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9135295241735144e-06,
+      "loss": 0.9509,
+      "step": 21887
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9134339766864134e-06,
+      "loss": 0.7639,
+      "step": 21888
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.913338429199312e-06,
+      "loss": 0.7971,
+      "step": 21889
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.913242881712211e-06,
+      "loss": 0.729,
+      "step": 21890
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9131473342251097e-06,
+      "loss": 1.2353,
+      "step": 21891
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.913051786738009e-06,
+      "loss": 0.7231,
+      "step": 21892
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.912956239250908e-06,
+      "loss": 0.7813,
+      "step": 21893
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.912860691763807e-06,
+      "loss": 1.1089,
+      "step": 21894
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.912765144276706e-06,
+      "loss": 0.8248,
+      "step": 21895
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9126695967896044e-06,
+      "loss": 0.9238,
+      "step": 21896
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9125740493025035e-06,
+      "loss": 0.4878,
+      "step": 21897
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.912478501815402e-06,
+      "loss": 0.5715,
+      "step": 21898
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9123829543283016e-06,
+      "loss": 0.5991,
+      "step": 21899
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9122874068412006e-06,
+      "loss": 0.5939,
+      "step": 21900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9121918593540992e-06,
+      "loss": 0.3243,
+      "step": 21901
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9120963118669983e-06,
+      "loss": 0.6696,
+      "step": 21902
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.912000764379897e-06,
+      "loss": 0.8192,
+      "step": 21903
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911905216892796e-06,
+      "loss": 1.0894,
+      "step": 21904
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9118096694056945e-06,
+      "loss": 0.4399,
+      "step": 21905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911714121918594e-06,
+      "loss": 1.3274,
+      "step": 21906
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911618574431493e-06,
+      "loss": 0.6893,
+      "step": 21907
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9115230269443916e-06,
+      "loss": 0.5163,
+      "step": 21908
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9114274794572907e-06,
+      "loss": 1.0347,
+      "step": 21909
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9113319319701893e-06,
+      "loss": 0.5593,
+      "step": 21910
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9112363844830883e-06,
+      "loss": 0.7038,
+      "step": 21911
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911140836995987e-06,
+      "loss": 1.0602,
+      "step": 21912
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9110452895088864e-06,
+      "loss": 0.6468,
+      "step": 21913
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9109497420217854e-06,
+      "loss": 0.952,
+      "step": 21914
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.910854194534684e-06,
+      "loss": 0.919,
+      "step": 21915
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.910758647047583e-06,
+      "loss": 0.8181,
+      "step": 21916
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9106630995604817e-06,
+      "loss": 0.6943,
+      "step": 21917
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9105675520733808e-06,
+      "loss": 1.0363,
+      "step": 21918
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9104720045862794e-06,
+      "loss": 0.8696,
+      "step": 21919
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9103764570991784e-06,
+      "loss": 0.9689,
+      "step": 21920
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9102809096120774e-06,
+      "loss": 1.1065,
+      "step": 21921
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9101853621249765e-06,
+      "loss": 0.738,
+      "step": 21922
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9100898146378755e-06,
+      "loss": 0.697,
+      "step": 21923
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909994267150774e-06,
+      "loss": 1.3124,
+      "step": 21924
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909898719663673e-06,
+      "loss": 0.9981,
+      "step": 21925
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909803172176572e-06,
+      "loss": 0.7086,
+      "step": 21926
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909707624689471e-06,
+      "loss": 1.0212,
+      "step": 21927
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.90961207720237e-06,
+      "loss": 0.4058,
+      "step": 21928
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909516529715269e-06,
+      "loss": 0.7355,
+      "step": 21929
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9094209822281675e-06,
+      "loss": 1.222,
+      "step": 21930
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9093254347410666e-06,
+      "loss": 1.0831,
+      "step": 21931
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909229887253965e-06,
+      "loss": 0.6562,
+      "step": 21932
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.909134339766864e-06,
+      "loss": 1.1632,
+      "step": 21933
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9090387922797632e-06,
+      "loss": 0.6066,
+      "step": 21934
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9089432447926623e-06,
+      "loss": 0.5433,
+      "step": 21935
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9088476973055613e-06,
+      "loss": 1.1434,
+      "step": 21936
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.90875214981846e-06,
+      "loss": 0.9334,
+      "step": 21937
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.908656602331359e-06,
+      "loss": 0.7601,
+      "step": 21938
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9085610548442576e-06,
+      "loss": 0.5542,
+      "step": 21939
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9084655073571566e-06,
+      "loss": 0.4433,
+      "step": 21940
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9083699598700552e-06,
+      "loss": 0.9823,
+      "step": 21941
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9082744123829547e-06,
+      "loss": 0.5436,
+      "step": 21942
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9081788648958537e-06,
+      "loss": 0.4762,
+      "step": 21943
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9080833174087524e-06,
+      "loss": 0.4771,
+      "step": 21944
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9079877699216514e-06,
+      "loss": 0.7388,
+      "step": 21945
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.90789222243455e-06,
+      "loss": 0.8953,
+      "step": 21946
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.907796674947449e-06,
+      "loss": 0.4744,
+      "step": 21947
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9077011274603477e-06,
+      "loss": 0.7755,
+      "step": 21948
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.907605579973247e-06,
+      "loss": 0.3557,
+      "step": 21949
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.907510032486146e-06,
+      "loss": 0.6829,
+      "step": 21950
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9074144849990448e-06,
+      "loss": 1.1388,
+      "step": 21951
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.907318937511944e-06,
+      "loss": 0.9542,
+      "step": 21952
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9072233900248424e-06,
+      "loss": 0.7598,
+      "step": 21953
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9071278425377415e-06,
+      "loss": 0.9869,
+      "step": 21954
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.90703229505064e-06,
+      "loss": 0.955,
+      "step": 21955
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906936747563539e-06,
+      "loss": 0.5529,
+      "step": 21956
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9068412000764386e-06,
+      "loss": 0.8551,
+      "step": 21957
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906745652589337e-06,
+      "loss": 0.9776,
+      "step": 21958
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9066501051022362e-06,
+      "loss": 0.8167,
+      "step": 21959
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906554557615135e-06,
+      "loss": 0.7181,
+      "step": 21960
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906459010128034e-06,
+      "loss": 1.2832,
+      "step": 21961
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9063634626409325e-06,
+      "loss": 1.0039,
+      "step": 21962
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9062679151538316e-06,
+      "loss": 0.7195,
+      "step": 21963
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.906172367666731e-06,
+      "loss": 0.6151,
+      "step": 21964
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9060768201796296e-06,
+      "loss": 1.17,
+      "step": 21965
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9059812726925287e-06,
+      "loss": 1.2256,
+      "step": 21966
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9058857252054273e-06,
+      "loss": 1.3579,
+      "step": 21967
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9057901777183263e-06,
+      "loss": 0.6376,
+      "step": 21968
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905694630231225e-06,
+      "loss": 0.7383,
+      "step": 21969
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905599082744124e-06,
+      "loss": 0.7799,
+      "step": 21970
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905503535257023e-06,
+      "loss": 0.6549,
+      "step": 21971
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905407987769922e-06,
+      "loss": 0.6003,
+      "step": 21972
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.905312440282821e-06,
+      "loss": 0.5729,
+      "step": 21973
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9052168927957197e-06,
+      "loss": 0.4259,
+      "step": 21974
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9051213453086187e-06,
+      "loss": 1.0645,
+      "step": 21975
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9050257978215174e-06,
+      "loss": 1.2041,
+      "step": 21976
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9049302503344164e-06,
+      "loss": 0.815,
+      "step": 21977
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9048347028473154e-06,
+      "loss": 0.7714,
+      "step": 21978
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9047391553602145e-06,
+      "loss": 0.5615,
+      "step": 21979
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904643607873113e-06,
+      "loss": 0.5539,
+      "step": 21980
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904548060386012e-06,
+      "loss": 0.8079,
+      "step": 21981
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9044525128989107e-06,
+      "loss": 0.9025,
+      "step": 21982
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9043569654118098e-06,
+      "loss": 0.8007,
+      "step": 21983
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904261417924709e-06,
+      "loss": 1.3703,
+      "step": 21984
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904165870437608e-06,
+      "loss": 0.5225,
+      "step": 21985
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.904070322950507e-06,
+      "loss": 0.5151,
+      "step": 21986
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9039747754634055e-06,
+      "loss": 1.1603,
+      "step": 21987
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9038792279763046e-06,
+      "loss": 0.6652,
+      "step": 21988
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.903783680489203e-06,
+      "loss": 0.8445,
+      "step": 21989
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.903688133002102e-06,
+      "loss": 0.6856,
+      "step": 21990
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.903592585515001e-06,
+      "loss": 0.8626,
+      "step": 21991
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9034970380279e-06,
+      "loss": 0.7927,
+      "step": 21992
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9034014905407993e-06,
+      "loss": 0.6259,
+      "step": 21993
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.903305943053698e-06,
+      "loss": 0.7248,
+      "step": 21994
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.903210395566597e-06,
+      "loss": 0.5217,
+      "step": 21995
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9031148480794956e-06,
+      "loss": 0.9243,
+      "step": 21996
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9030193005923946e-06,
+      "loss": 0.9308,
+      "step": 21997
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9029237531052932e-06,
+      "loss": 0.73,
+      "step": 21998
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9028282056181923e-06,
+      "loss": 0.6011,
+      "step": 21999
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9027326581310917e-06,
+      "loss": 0.7988,
+      "step": 22000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9026371106439904e-06,
+      "loss": 0.8547,
+      "step": 22001
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9025415631568894e-06,
+      "loss": 0.7459,
+      "step": 22002
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.902446015669788e-06,
+      "loss": 0.7483,
+      "step": 22003
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.902350468182687e-06,
+      "loss": 0.5755,
+      "step": 22004
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9022549206955857e-06,
+      "loss": 0.7306,
+      "step": 22005
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9021593732084847e-06,
+      "loss": 0.7237,
+      "step": 22006
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.902063825721384e-06,
+      "loss": 0.7652,
+      "step": 22007
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9019682782342828e-06,
+      "loss": 0.7321,
+      "step": 22008
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901872730747182e-06,
+      "loss": 0.6313,
+      "step": 22009
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9017771832600804e-06,
+      "loss": 0.9438,
+      "step": 22010
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9016816357729795e-06,
+      "loss": 0.6256,
+      "step": 22011
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901586088285878e-06,
+      "loss": 0.9109,
+      "step": 22012
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901490540798777e-06,
+      "loss": 0.8304,
+      "step": 22013
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9013949933116766e-06,
+      "loss": 0.8111,
+      "step": 22014
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901299445824575e-06,
+      "loss": 0.8925,
+      "step": 22015
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9012038983374742e-06,
+      "loss": 0.803,
+      "step": 22016
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901108350850373e-06,
+      "loss": 0.9566,
+      "step": 22017
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901012803363272e-06,
+      "loss": 0.9233,
+      "step": 22018
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9009172558761705e-06,
+      "loss": 0.8464,
+      "step": 22019
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9008217083890696e-06,
+      "loss": 1.1487,
+      "step": 22020
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9007261609019686e-06,
+      "loss": 0.9166,
+      "step": 22021
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9006306134148676e-06,
+      "loss": 1.1191,
+      "step": 22022
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9005350659277662e-06,
+      "loss": 0.879,
+      "step": 22023
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9004395184406653e-06,
+      "loss": 0.6885,
+      "step": 22024
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9003439709535643e-06,
+      "loss": 0.8965,
+      "step": 22025
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.900248423466463e-06,
+      "loss": 0.8194,
+      "step": 22026
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.900152875979362e-06,
+      "loss": 0.6474,
+      "step": 22027
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.9000573284922606e-06,
+      "loss": 0.8694,
+      "step": 22028
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89996178100516e-06,
+      "loss": 0.7634,
+      "step": 22029
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8998662335180587e-06,
+      "loss": 0.7201,
+      "step": 22030
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8997706860309577e-06,
+      "loss": 0.9837,
+      "step": 22031
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8996751385438563e-06,
+      "loss": 1.0627,
+      "step": 22032
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8995795910567554e-06,
+      "loss": 0.8099,
+      "step": 22033
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.899484043569654e-06,
+      "loss": 1.0312,
+      "step": 22034
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.899388496082553e-06,
+      "loss": 0.5897,
+      "step": 22035
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8992929485954525e-06,
+      "loss": 1.0083,
+      "step": 22036
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.899197401108351e-06,
+      "loss": 0.8374,
+      "step": 22037
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89910185362125e-06,
+      "loss": 0.6428,
+      "step": 22038
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8990063061341487e-06,
+      "loss": 1.0225,
+      "step": 22039
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8989107586470478e-06,
+      "loss": 0.6949,
+      "step": 22040
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8988152111599464e-06,
+      "loss": 0.8368,
+      "step": 22041
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8987196636728454e-06,
+      "loss": 0.8768,
+      "step": 22042
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.898624116185745e-06,
+      "loss": 0.5523,
+      "step": 22043
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8985285686986435e-06,
+      "loss": 0.7307,
+      "step": 22044
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8984330212115426e-06,
+      "loss": 0.9355,
+      "step": 22045
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.898337473724441e-06,
+      "loss": 0.9433,
+      "step": 22046
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89824192623734e-06,
+      "loss": 1.2435,
+      "step": 22047
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.898146378750239e-06,
+      "loss": 0.8911,
+      "step": 22048
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.898050831263138e-06,
+      "loss": 0.6785,
+      "step": 22049
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8979552837760373e-06,
+      "loss": 0.8765,
+      "step": 22050
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.897859736288936e-06,
+      "loss": 0.7455,
+      "step": 22051
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.897764188801835e-06,
+      "loss": 0.6722,
+      "step": 22052
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8976686413147336e-06,
+      "loss": 0.888,
+      "step": 22053
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8975730938276326e-06,
+      "loss": 0.5831,
+      "step": 22054
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8974775463405312e-06,
+      "loss": 0.6564,
+      "step": 22055
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8973819988534303e-06,
+      "loss": 0.9463,
+      "step": 22056
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8972864513663297e-06,
+      "loss": 1.2983,
+      "step": 22057
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8971909038792284e-06,
+      "loss": 0.5498,
+      "step": 22058
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8970953563921274e-06,
+      "loss": 1.1031,
+      "step": 22059
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.896999808905026e-06,
+      "loss": 0.9639,
+      "step": 22060
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.896904261417925e-06,
+      "loss": 1.0884,
+      "step": 22061
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8968087139308237e-06,
+      "loss": 0.7982,
+      "step": 22062
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8967131664437227e-06,
+      "loss": 1.1162,
+      "step": 22063
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8966176189566213e-06,
+      "loss": 0.8175,
+      "step": 22064
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8965220714695208e-06,
+      "loss": 1.0897,
+      "step": 22065
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89642652398242e-06,
+      "loss": 0.7266,
+      "step": 22066
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8963309764953184e-06,
+      "loss": 0.5734,
+      "step": 22067
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8962354290082175e-06,
+      "loss": 0.9147,
+      "step": 22068
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.896139881521116e-06,
+      "loss": 0.8848,
+      "step": 22069
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.896044334034015e-06,
+      "loss": 0.9343,
+      "step": 22070
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8959487865469137e-06,
+      "loss": 0.803,
+      "step": 22071
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895853239059813e-06,
+      "loss": 0.8208,
+      "step": 22072
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895757691572712e-06,
+      "loss": 0.5858,
+      "step": 22073
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895662144085611e-06,
+      "loss": 0.7099,
+      "step": 22074
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.89556659659851e-06,
+      "loss": 1.1328,
+      "step": 22075
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8954710491114085e-06,
+      "loss": 0.9246,
+      "step": 22076
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8953755016243076e-06,
+      "loss": 0.5853,
+      "step": 22077
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895279954137206e-06,
+      "loss": 0.4484,
+      "step": 22078
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8951844066501056e-06,
+      "loss": 0.645,
+      "step": 22079
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8950888591630042e-06,
+      "loss": 1.1595,
+      "step": 22080
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8949933116759033e-06,
+      "loss": 1.1031,
+      "step": 22081
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.894897764188802e-06,
+      "loss": 0.9618,
+      "step": 22082
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.894802216701701e-06,
+      "loss": 0.6273,
+      "step": 22083
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8947066692145996e-06,
+      "loss": 1.1424,
+      "step": 22084
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8946111217274986e-06,
+      "loss": 0.9311,
+      "step": 22085
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.894515574240398e-06,
+      "loss": 0.9174,
+      "step": 22086
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8944200267532967e-06,
+      "loss": 0.3941,
+      "step": 22087
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8943244792661957e-06,
+      "loss": 0.7291,
+      "step": 22088
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8942289317790943e-06,
+      "loss": 0.8906,
+      "step": 22089
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8941333842919934e-06,
+      "loss": 0.7717,
+      "step": 22090
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.894037836804892e-06,
+      "loss": 0.7667,
+      "step": 22091
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.893942289317791e-06,
+      "loss": 1.0436,
+      "step": 22092
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8938467418306905e-06,
+      "loss": 1.2509,
+      "step": 22093
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.893751194343589e-06,
+      "loss": 1.1088,
+      "step": 22094
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.893655646856488e-06,
+      "loss": 0.7417,
+      "step": 22095
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8935600993693867e-06,
+      "loss": 0.6432,
+      "step": 22096
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8934645518822858e-06,
+      "loss": 0.742,
+      "step": 22097
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8933690043951844e-06,
+      "loss": 0.4454,
+      "step": 22098
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8932734569080834e-06,
+      "loss": 0.9683,
+      "step": 22099
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.893177909420982e-06,
+      "loss": 0.587,
+      "step": 22100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8930823619338815e-06,
+      "loss": 0.6355,
+      "step": 22101
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8929868144467806e-06,
+      "loss": 0.3471,
+      "step": 22102
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892891266959679e-06,
+      "loss": 0.6255,
+      "step": 22103
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892795719472578e-06,
+      "loss": 0.4708,
+      "step": 22104
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892700171985477e-06,
+      "loss": 0.8645,
+      "step": 22105
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892604624498376e-06,
+      "loss": 0.5453,
+      "step": 22106
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8925090770112745e-06,
+      "loss": 0.4334,
+      "step": 22107
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892413529524174e-06,
+      "loss": 0.9199,
+      "step": 22108
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.892317982037073e-06,
+      "loss": 0.3434,
+      "step": 22109
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8922224345499716e-06,
+      "loss": 0.4889,
+      "step": 22110
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8921268870628706e-06,
+      "loss": 0.5454,
+      "step": 22111
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8920313395757692e-06,
+      "loss": 0.6306,
+      "step": 22112
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8919357920886683e-06,
+      "loss": 0.6718,
+      "step": 22113
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.891840244601567e-06,
+      "loss": 1.0364,
+      "step": 22114
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8917446971144664e-06,
+      "loss": 0.596,
+      "step": 22115
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8916491496273654e-06,
+      "loss": 0.9918,
+      "step": 22116
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.891553602140264e-06,
+      "loss": 0.7294,
+      "step": 22117
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.891458054653163e-06,
+      "loss": 0.8224,
+      "step": 22118
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8913625071660617e-06,
+      "loss": 0.6941,
+      "step": 22119
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8912669596789607e-06,
+      "loss": 0.89,
+      "step": 22120
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8911714121918593e-06,
+      "loss": 0.7073,
+      "step": 22121
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8910758647047588e-06,
+      "loss": 0.8363,
+      "step": 22122
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8909803172176574e-06,
+      "loss": 0.5647,
+      "step": 22123
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8908847697305564e-06,
+      "loss": 0.4311,
+      "step": 22124
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.890789222243455e-06,
+      "loss": 1.0538,
+      "step": 22125
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.890693674756354e-06,
+      "loss": 0.6944,
+      "step": 22126
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.890598127269253e-06,
+      "loss": 0.7372,
+      "step": 22127
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8905025797821517e-06,
+      "loss": 0.9669,
+      "step": 22128
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.890407032295051e-06,
+      "loss": 0.4462,
+      "step": 22129
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.89031148480795e-06,
+      "loss": 0.8306,
+      "step": 22130
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.890215937320849e-06,
+      "loss": 0.6433,
+      "step": 22131
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8901203898337475e-06,
+      "loss": 0.8684,
+      "step": 22132
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8900248423466465e-06,
+      "loss": 0.8034,
+      "step": 22133
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.889929294859545e-06,
+      "loss": 0.6099,
+      "step": 22134
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.889833747372444e-06,
+      "loss": 0.9769,
+      "step": 22135
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8897381998853428e-06,
+      "loss": 0.6893,
+      "step": 22136
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8896426523982422e-06,
+      "loss": 0.9648,
+      "step": 22137
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8895471049111413e-06,
+      "loss": 0.9235,
+      "step": 22138
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.88945155742404e-06,
+      "loss": 0.659,
+      "step": 22139
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.889356009936939e-06,
+      "loss": 0.9503,
+      "step": 22140
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8892604624498376e-06,
+      "loss": 0.9458,
+      "step": 22141
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8891649149627366e-06,
+      "loss": 1.2269,
+      "step": 22142
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.889069367475635e-06,
+      "loss": 0.7794,
+      "step": 22143
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8889738199885347e-06,
+      "loss": 1.0432,
+      "step": 22144
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8888782725014337e-06,
+      "loss": 0.6503,
+      "step": 22145
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8887827250143323e-06,
+      "loss": 0.5325,
+      "step": 22146
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8886871775272314e-06,
+      "loss": 0.6096,
+      "step": 22147
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.88859163004013e-06,
+      "loss": 1.0,
+      "step": 22148
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.888496082553029e-06,
+      "loss": 0.8504,
+      "step": 22149
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8884005350659276e-06,
+      "loss": 1.0724,
+      "step": 22150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.888304987578827e-06,
+      "loss": 0.907,
+      "step": 22151
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.888209440091726e-06,
+      "loss": 1.2837,
+      "step": 22152
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8881138926046247e-06,
+      "loss": 0.8464,
+      "step": 22153
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8880183451175238e-06,
+      "loss": 0.2773,
+      "step": 22154
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8879227976304224e-06,
+      "loss": 1.1598,
+      "step": 22155
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8878272501433214e-06,
+      "loss": 0.7132,
+      "step": 22156
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.88773170265622e-06,
+      "loss": 0.5821,
+      "step": 22157
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8876361551691195e-06,
+      "loss": 0.8546,
+      "step": 22158
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8875406076820186e-06,
+      "loss": 0.3744,
+      "step": 22159
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887445060194917e-06,
+      "loss": 0.4725,
+      "step": 22160
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887349512707816e-06,
+      "loss": 0.599,
+      "step": 22161
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887253965220715e-06,
+      "loss": 0.6009,
+      "step": 22162
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887158417733614e-06,
+      "loss": 0.7423,
+      "step": 22163
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8870628702465125e-06,
+      "loss": 0.8245,
+      "step": 22164
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886967322759412e-06,
+      "loss": 0.5487,
+      "step": 22165
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886871775272311e-06,
+      "loss": 0.6515,
+      "step": 22166
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8867762277852096e-06,
+      "loss": 0.7693,
+      "step": 22167
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8866806802981086e-06,
+      "loss": 0.604,
+      "step": 22168
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8865851328110072e-06,
+      "loss": 1.3915,
+      "step": 22169
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8864895853239063e-06,
+      "loss": 1.3498,
+      "step": 22170
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886394037836805e-06,
+      "loss": 0.814,
+      "step": 22171
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886298490349704e-06,
+      "loss": 0.808,
+      "step": 22172
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886202942862603e-06,
+      "loss": 1.0782,
+      "step": 22173
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.886107395375502e-06,
+      "loss": 0.3939,
+      "step": 22174
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8860118478884006e-06,
+      "loss": 0.6742,
+      "step": 22175
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8859163004012997e-06,
+      "loss": 0.8641,
+      "step": 22176
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8858207529141987e-06,
+      "loss": 0.665,
+      "step": 22177
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8857252054270973e-06,
+      "loss": 0.9832,
+      "step": 22178
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8856296579399964e-06,
+      "loss": 1.3178,
+      "step": 22179
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8855341104528954e-06,
+      "loss": 0.742,
+      "step": 22180
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8854385629657944e-06,
+      "loss": 1.1718,
+      "step": 22181
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.885343015478693e-06,
+      "loss": 0.7146,
+      "step": 22182
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.885247467991592e-06,
+      "loss": 0.9711,
+      "step": 22183
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8851519205044907e-06,
+      "loss": 0.6048,
+      "step": 22184
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8850563730173897e-06,
+      "loss": 0.7331,
+      "step": 22185
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8849608255302884e-06,
+      "loss": 0.9251,
+      "step": 22186
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884865278043188e-06,
+      "loss": 0.8471,
+      "step": 22187
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884769730556087e-06,
+      "loss": 1.19,
+      "step": 22188
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8846741830689855e-06,
+      "loss": 0.8798,
+      "step": 22189
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8845786355818845e-06,
+      "loss": 0.5534,
+      "step": 22190
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884483088094783e-06,
+      "loss": 1.1285,
+      "step": 22191
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884387540607682e-06,
+      "loss": 0.7449,
+      "step": 22192
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8842919931205808e-06,
+      "loss": 0.7039,
+      "step": 22193
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8841964456334802e-06,
+      "loss": 0.5874,
+      "step": 22194
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8841008981463793e-06,
+      "loss": 1.2338,
+      "step": 22195
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.884005350659278e-06,
+      "loss": 0.732,
+      "step": 22196
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.883909803172177e-06,
+      "loss": 1.1408,
+      "step": 22197
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8838142556850756e-06,
+      "loss": 0.6545,
+      "step": 22198
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8837187081979746e-06,
+      "loss": 0.998,
+      "step": 22199
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.883623160710873e-06,
+      "loss": 0.5764,
+      "step": 22200
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8835276132237727e-06,
+      "loss": 0.6565,
+      "step": 22201
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8834320657366717e-06,
+      "loss": 0.6362,
+      "step": 22202
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8833365182495703e-06,
+      "loss": 0.6845,
+      "step": 22203
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8832409707624694e-06,
+      "loss": 0.3723,
+      "step": 22204
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.883145423275368e-06,
+      "loss": 0.8719,
+      "step": 22205
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.883049875788267e-06,
+      "loss": 0.5519,
+      "step": 22206
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8829543283011656e-06,
+      "loss": 0.7169,
+      "step": 22207
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8828587808140647e-06,
+      "loss": 0.6128,
+      "step": 22208
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.882763233326964e-06,
+      "loss": 0.8735,
+      "step": 22209
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8826676858398627e-06,
+      "loss": 0.6288,
+      "step": 22210
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8825721383527618e-06,
+      "loss": 0.8224,
+      "step": 22211
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8824765908656604e-06,
+      "loss": 1.4216,
+      "step": 22212
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8823810433785594e-06,
+      "loss": 0.8961,
+      "step": 22213
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.882285495891458e-06,
+      "loss": 0.8575,
+      "step": 22214
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.882189948404357e-06,
+      "loss": 0.8053,
+      "step": 22215
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8820944009172566e-06,
+      "loss": 0.5339,
+      "step": 22216
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881998853430155e-06,
+      "loss": 0.6332,
+      "step": 22217
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881903305943054e-06,
+      "loss": 0.3894,
+      "step": 22218
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881807758455953e-06,
+      "loss": 0.9745,
+      "step": 22219
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881712210968852e-06,
+      "loss": 1.0188,
+      "step": 22220
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8816166634817505e-06,
+      "loss": 1.2278,
+      "step": 22221
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8815211159946495e-06,
+      "loss": 0.4561,
+      "step": 22222
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8814255685075486e-06,
+      "loss": 0.5445,
+      "step": 22223
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8813300210204476e-06,
+      "loss": 0.7207,
+      "step": 22224
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881234473533346e-06,
+      "loss": 0.6269,
+      "step": 22225
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8811389260462452e-06,
+      "loss": 0.7897,
+      "step": 22226
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.881043378559144e-06,
+      "loss": 0.8802,
+      "step": 22227
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.880947831072043e-06,
+      "loss": 0.7817,
+      "step": 22228
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.880852283584942e-06,
+      "loss": 0.6473,
+      "step": 22229
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.880756736097841e-06,
+      "loss": 0.6145,
+      "step": 22230
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.88066118861074e-06,
+      "loss": 0.5932,
+      "step": 22231
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8805656411236386e-06,
+      "loss": 0.9194,
+      "step": 22232
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8804700936365377e-06,
+      "loss": 0.8123,
+      "step": 22233
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8803745461494363e-06,
+      "loss": 0.4183,
+      "step": 22234
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8802789986623353e-06,
+      "loss": 0.9794,
+      "step": 22235
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.880183451175234e-06,
+      "loss": 0.9237,
+      "step": 22236
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8800879036881334e-06,
+      "loss": 0.7053,
+      "step": 22237
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8799923562010324e-06,
+      "loss": 0.7936,
+      "step": 22238
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.879896808713931e-06,
+      "loss": 0.5683,
+      "step": 22239
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.87980126122683e-06,
+      "loss": 0.8012,
+      "step": 22240
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8797057137397287e-06,
+      "loss": 1.2061,
+      "step": 22241
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8796101662526277e-06,
+      "loss": 1.3373,
+      "step": 22242
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8795146187655264e-06,
+      "loss": 0.8974,
+      "step": 22243
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8794190712784254e-06,
+      "loss": 0.9315,
+      "step": 22244
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.879323523791325e-06,
+      "loss": 0.5848,
+      "step": 22245
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8792279763042235e-06,
+      "loss": 1.2109,
+      "step": 22246
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8791324288171225e-06,
+      "loss": 0.8628,
+      "step": 22247
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.879036881330021e-06,
+      "loss": 0.9221,
+      "step": 22248
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.87894133384292e-06,
+      "loss": 0.667,
+      "step": 22249
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8788457863558188e-06,
+      "loss": 0.6041,
+      "step": 22250
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.878750238868718e-06,
+      "loss": 0.7613,
+      "step": 22251
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8786546913816173e-06,
+      "loss": 1.0811,
+      "step": 22252
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.878559143894516e-06,
+      "loss": 0.8567,
+      "step": 22253
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.878463596407415e-06,
+      "loss": 0.5718,
+      "step": 22254
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8783680489203136e-06,
+      "loss": 0.8024,
+      "step": 22255
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8782725014332126e-06,
+      "loss": 0.6911,
+      "step": 22256
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.878176953946111e-06,
+      "loss": 0.5468,
+      "step": 22257
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8780814064590102e-06,
+      "loss": 0.9409,
+      "step": 22258
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8779858589719097e-06,
+      "loss": 1.2602,
+      "step": 22259
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8778903114848083e-06,
+      "loss": 0.7495,
+      "step": 22260
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8777947639977074e-06,
+      "loss": 0.7667,
+      "step": 22261
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.877699216510606e-06,
+      "loss": 0.7704,
+      "step": 22262
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.877603669023505e-06,
+      "loss": 0.6051,
+      "step": 22263
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8775081215364036e-06,
+      "loss": 0.9818,
+      "step": 22264
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8774125740493027e-06,
+      "loss": 1.2312,
+      "step": 22265
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8773170265622017e-06,
+      "loss": 0.6703,
+      "step": 22266
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8772214790751007e-06,
+      "loss": 0.6474,
+      "step": 22267
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8771259315879998e-06,
+      "loss": 0.8377,
+      "step": 22268
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8770303841008984e-06,
+      "loss": 0.7686,
+      "step": 22269
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8769348366137974e-06,
+      "loss": 0.9533,
+      "step": 22270
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876839289126696e-06,
+      "loss": 0.8835,
+      "step": 22271
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876743741639595e-06,
+      "loss": 0.6371,
+      "step": 22272
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876648194152494e-06,
+      "loss": 0.444,
+      "step": 22273
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876552646665393e-06,
+      "loss": 0.9664,
+      "step": 22274
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8764570991782918e-06,
+      "loss": 0.5127,
+      "step": 22275
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876361551691191e-06,
+      "loss": 0.6333,
+      "step": 22276
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8762660042040894e-06,
+      "loss": 0.8426,
+      "step": 22277
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8761704567169885e-06,
+      "loss": 0.5518,
+      "step": 22278
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8760749092298875e-06,
+      "loss": 0.9436,
+      "step": 22279
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.875979361742786e-06,
+      "loss": 0.8427,
+      "step": 22280
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8758838142556856e-06,
+      "loss": 0.6919,
+      "step": 22281
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.875788266768584e-06,
+      "loss": 1.1826,
+      "step": 22282
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8756927192814832e-06,
+      "loss": 1.1079,
+      "step": 22283
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.875597171794382e-06,
+      "loss": 1.1626,
+      "step": 22284
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.875501624307281e-06,
+      "loss": 0.7169,
+      "step": 22285
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8754060768201795e-06,
+      "loss": 0.7361,
+      "step": 22286
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8753105293330786e-06,
+      "loss": 1.3694,
+      "step": 22287
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.875214981845978e-06,
+      "loss": 0.6991,
+      "step": 22288
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8751194343588766e-06,
+      "loss": 0.5577,
+      "step": 22289
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8750238868717757e-06,
+      "loss": 0.8576,
+      "step": 22290
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8749283393846743e-06,
+      "loss": 0.3732,
+      "step": 22291
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8748327918975733e-06,
+      "loss": 0.6921,
+      "step": 22292
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.874737244410472e-06,
+      "loss": 0.9614,
+      "step": 22293
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.874641696923371e-06,
+      "loss": 0.9466,
+      "step": 22294
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8745461494362704e-06,
+      "loss": 0.7567,
+      "step": 22295
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.874450601949169e-06,
+      "loss": 1.2466,
+      "step": 22296
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.874355054462068e-06,
+      "loss": 0.6808,
+      "step": 22297
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8742595069749667e-06,
+      "loss": 0.5951,
+      "step": 22298
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8741639594878657e-06,
+      "loss": 0.6964,
+      "step": 22299
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8740684120007644e-06,
+      "loss": 0.5031,
+      "step": 22300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8739728645136634e-06,
+      "loss": 0.9731,
+      "step": 22301
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873877317026563e-06,
+      "loss": 0.6291,
+      "step": 22302
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8737817695394615e-06,
+      "loss": 0.9388,
+      "step": 22303
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8736862220523605e-06,
+      "loss": 0.584,
+      "step": 22304
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873590674565259e-06,
+      "loss": 0.5973,
+      "step": 22305
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873495127078158e-06,
+      "loss": 0.7851,
+      "step": 22306
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8733995795910568e-06,
+      "loss": 0.6387,
+      "step": 22307
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873304032103956e-06,
+      "loss": 0.8123,
+      "step": 22308
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8732084846168553e-06,
+      "loss": 0.8996,
+      "step": 22309
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873112937129754e-06,
+      "loss": 0.5814,
+      "step": 22310
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.873017389642653e-06,
+      "loss": 0.9466,
+      "step": 22311
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8729218421555516e-06,
+      "loss": 0.4297,
+      "step": 22312
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8728262946684506e-06,
+      "loss": 1.211,
+      "step": 22313
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.872730747181349e-06,
+      "loss": 0.7058,
+      "step": 22314
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8726351996942482e-06,
+      "loss": 0.7545,
+      "step": 22315
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.872539652207147e-06,
+      "loss": 1.0298,
+      "step": 22316
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8724441047200463e-06,
+      "loss": 0.7192,
+      "step": 22317
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.872348557232945e-06,
+      "loss": 0.9407,
+      "step": 22318
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.872253009745844e-06,
+      "loss": 0.5583,
+      "step": 22319
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.872157462258743e-06,
+      "loss": 0.9603,
+      "step": 22320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8720619147716416e-06,
+      "loss": 0.838,
+      "step": 22321
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8719663672845407e-06,
+      "loss": 0.8096,
+      "step": 22322
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8718708197974393e-06,
+      "loss": 1.1825,
+      "step": 22323
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8717752723103387e-06,
+      "loss": 0.9806,
+      "step": 22324
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8716797248232374e-06,
+      "loss": 0.7986,
+      "step": 22325
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8715841773361364e-06,
+      "loss": 0.8041,
+      "step": 22326
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.871488629849035e-06,
+      "loss": 0.7353,
+      "step": 22327
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.871393082361934e-06,
+      "loss": 0.9671,
+      "step": 22328
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8712975348748327e-06,
+      "loss": 0.5986,
+      "step": 22329
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8712019873877317e-06,
+      "loss": 0.8188,
+      "step": 22330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.871106439900631e-06,
+      "loss": 0.5578,
+      "step": 22331
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8710108924135298e-06,
+      "loss": 0.9577,
+      "step": 22332
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.870915344926429e-06,
+      "loss": 1.137,
+      "step": 22333
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8708197974393274e-06,
+      "loss": 0.7402,
+      "step": 22334
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8707242499522265e-06,
+      "loss": 0.5042,
+      "step": 22335
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.870628702465125e-06,
+      "loss": 1.0265,
+      "step": 22336
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.870533154978024e-06,
+      "loss": 0.3263,
+      "step": 22337
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8704376074909236e-06,
+      "loss": 1.0739,
+      "step": 22338
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.870342060003822e-06,
+      "loss": 0.7787,
+      "step": 22339
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8702465125167212e-06,
+      "loss": 0.5589,
+      "step": 22340
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.87015096502962e-06,
+      "loss": 0.952,
+      "step": 22341
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.870055417542519e-06,
+      "loss": 0.7717,
+      "step": 22342
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8699598700554175e-06,
+      "loss": 0.9098,
+      "step": 22343
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8698643225683165e-06,
+      "loss": 0.6584,
+      "step": 22344
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.869768775081216e-06,
+      "loss": 0.8427,
+      "step": 22345
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8696732275941146e-06,
+      "loss": 0.7113,
+      "step": 22346
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8695776801070137e-06,
+      "loss": 1.1605,
+      "step": 22347
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8694821326199123e-06,
+      "loss": 0.7734,
+      "step": 22348
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8693865851328113e-06,
+      "loss": 0.7431,
+      "step": 22349
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.86929103764571e-06,
+      "loss": 0.5782,
+      "step": 22350
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.869195490158609e-06,
+      "loss": 0.5918,
+      "step": 22351
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8690999426715076e-06,
+      "loss": 0.5795,
+      "step": 22352
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.869004395184407e-06,
+      "loss": 0.6881,
+      "step": 22353
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.868908847697306e-06,
+      "loss": 1.0976,
+      "step": 22354
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8688133002102047e-06,
+      "loss": 0.5574,
+      "step": 22355
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8687177527231037e-06,
+      "loss": 0.6105,
+      "step": 22356
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8686222052360024e-06,
+      "loss": 0.4121,
+      "step": 22357
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8685266577489014e-06,
+      "loss": 0.8751,
+      "step": 22358
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8684311102618e-06,
+      "loss": 0.5094,
+      "step": 22359
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8683355627746995e-06,
+      "loss": 1.0232,
+      "step": 22360
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8682400152875985e-06,
+      "loss": 1.3503,
+      "step": 22361
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.868144467800497e-06,
+      "loss": 0.54,
+      "step": 22362
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.868048920313396e-06,
+      "loss": 1.0375,
+      "step": 22363
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8679533728262948e-06,
+      "loss": 1.01,
+      "step": 22364
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.867857825339194e-06,
+      "loss": 0.3393,
+      "step": 22365
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8677622778520924e-06,
+      "loss": 0.9407,
+      "step": 22366
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.867666730364992e-06,
+      "loss": 0.6907,
+      "step": 22367
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8675711828778905e-06,
+      "loss": 0.707,
+      "step": 22368
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8674756353907895e-06,
+      "loss": 0.305,
+      "step": 22369
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8673800879036886e-06,
+      "loss": 0.8028,
+      "step": 22370
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.867284540416587e-06,
+      "loss": 1.0436,
+      "step": 22371
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8671889929294862e-06,
+      "loss": 0.5446,
+      "step": 22372
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.867093445442385e-06,
+      "loss": 0.7914,
+      "step": 22373
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8669978979552843e-06,
+      "loss": 0.8108,
+      "step": 22374
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.866902350468183e-06,
+      "loss": 0.8025,
+      "step": 22375
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.866806802981082e-06,
+      "loss": 0.7668,
+      "step": 22376
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8667112554939806e-06,
+      "loss": 0.7033,
+      "step": 22377
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8666157080068796e-06,
+      "loss": 0.8677,
+      "step": 22378
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8665201605197782e-06,
+      "loss": 0.7267,
+      "step": 22379
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8664246130326773e-06,
+      "loss": 1.3406,
+      "step": 22380
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8663290655455767e-06,
+      "loss": 1.0415,
+      "step": 22381
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8662335180584754e-06,
+      "loss": 0.5754,
+      "step": 22382
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8661379705713744e-06,
+      "loss": 0.9733,
+      "step": 22383
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.866042423084273e-06,
+      "loss": 0.8208,
+      "step": 22384
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.865946875597172e-06,
+      "loss": 0.9292,
+      "step": 22385
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8658513281100707e-06,
+      "loss": 0.9182,
+      "step": 22386
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8657557806229697e-06,
+      "loss": 0.8836,
+      "step": 22387
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8656602331358683e-06,
+      "loss": 1.1243,
+      "step": 22388
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8655646856487678e-06,
+      "loss": 0.9385,
+      "step": 22389
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.865469138161667e-06,
+      "loss": 1.1419,
+      "step": 22390
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8653735906745654e-06,
+      "loss": 0.5464,
+      "step": 22391
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8652780431874645e-06,
+      "loss": 0.5598,
+      "step": 22392
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.865182495700363e-06,
+      "loss": 1.2647,
+      "step": 22393
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.865086948213262e-06,
+      "loss": 0.2785,
+      "step": 22394
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8649914007261607e-06,
+      "loss": 0.9715,
+      "step": 22395
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.86489585323906e-06,
+      "loss": 0.7312,
+      "step": 22396
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8648003057519592e-06,
+      "loss": 1.1353,
+      "step": 22397
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864704758264858e-06,
+      "loss": 0.7235,
+      "step": 22398
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864609210777757e-06,
+      "loss": 0.7421,
+      "step": 22399
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8645136632906555e-06,
+      "loss": 0.7362,
+      "step": 22400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8644181158035545e-06,
+      "loss": 1.2289,
+      "step": 22401
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.864322568316453e-06,
+      "loss": 0.7281,
+      "step": 22402
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8642270208293526e-06,
+      "loss": 0.6788,
+      "step": 22403
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8641314733422517e-06,
+      "loss": 0.7251,
+      "step": 22404
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8640359258551503e-06,
+      "loss": 1.0086,
+      "step": 22405
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8639403783680493e-06,
+      "loss": 0.7236,
+      "step": 22406
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.863844830880948e-06,
+      "loss": 0.5418,
+      "step": 22407
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.863749283393847e-06,
+      "loss": 0.8318,
+      "step": 22408
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8636537359067456e-06,
+      "loss": 1.0862,
+      "step": 22409
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.863558188419645e-06,
+      "loss": 0.6396,
+      "step": 22410
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.863462640932544e-06,
+      "loss": 0.9983,
+      "step": 22411
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8633670934454427e-06,
+      "loss": 1.0417,
+      "step": 22412
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8632715459583417e-06,
+      "loss": 0.7185,
+      "step": 22413
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8631759984712404e-06,
+      "loss": 1.0366,
+      "step": 22414
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8630804509841394e-06,
+      "loss": 0.5665,
+      "step": 22415
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.862984903497038e-06,
+      "loss": 0.5133,
+      "step": 22416
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8628893560099375e-06,
+      "loss": 0.7942,
+      "step": 22417
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.862793808522836e-06,
+      "loss": 0.9402,
+      "step": 22418
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.862698261035735e-06,
+      "loss": 1.1958,
+      "step": 22419
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8626027135486337e-06,
+      "loss": 0.7539,
+      "step": 22420
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8625071660615328e-06,
+      "loss": 0.5441,
+      "step": 22421
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.862411618574432e-06,
+      "loss": 0.5801,
+      "step": 22422
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8623160710873304e-06,
+      "loss": 0.7236,
+      "step": 22423
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8622205236002295e-06,
+      "loss": 0.5119,
+      "step": 22424
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8621249761131285e-06,
+      "loss": 0.7151,
+      "step": 22425
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8620294286260275e-06,
+      "loss": 0.9064,
+      "step": 22426
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861933881138926e-06,
+      "loss": 0.808,
+      "step": 22427
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861838333651825e-06,
+      "loss": 0.8453,
+      "step": 22428
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861742786164724e-06,
+      "loss": 0.7307,
+      "step": 22429
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861647238677623e-06,
+      "loss": 1.0646,
+      "step": 22430
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8615516911905215e-06,
+      "loss": 0.6985,
+      "step": 22431
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861456143703421e-06,
+      "loss": 0.7296,
+      "step": 22432
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.86136059621632e-06,
+      "loss": 1.0255,
+      "step": 22433
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8612650487292186e-06,
+      "loss": 0.3976,
+      "step": 22434
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8611695012421176e-06,
+      "loss": 0.5173,
+      "step": 22435
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8610739537550162e-06,
+      "loss": 0.6861,
+      "step": 22436
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8609784062679153e-06,
+      "loss": 0.6335,
+      "step": 22437
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.860882858780814e-06,
+      "loss": 0.9122,
+      "step": 22438
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8607873112937134e-06,
+      "loss": 1.3566,
+      "step": 22439
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8606917638066124e-06,
+      "loss": 0.7153,
+      "step": 22440
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.860596216319511e-06,
+      "loss": 0.9328,
+      "step": 22441
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.86050066883241e-06,
+      "loss": 0.4088,
+      "step": 22442
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8604051213453087e-06,
+      "loss": 0.7482,
+      "step": 22443
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8603095738582077e-06,
+      "loss": 1.2823,
+      "step": 22444
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8602140263711063e-06,
+      "loss": 0.7789,
+      "step": 22445
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8601184788840058e-06,
+      "loss": 0.7521,
+      "step": 22446
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.860022931396905e-06,
+      "loss": 0.8451,
+      "step": 22447
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8599273839098034e-06,
+      "loss": 0.9059,
+      "step": 22448
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8598318364227025e-06,
+      "loss": 0.9023,
+      "step": 22449
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859736288935601e-06,
+      "loss": 0.9269,
+      "step": 22450
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8596407414485e-06,
+      "loss": 0.8779,
+      "step": 22451
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8595451939613987e-06,
+      "loss": 0.5581,
+      "step": 22452
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859449646474298e-06,
+      "loss": 0.9449,
+      "step": 22453
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8593540989871972e-06,
+      "loss": 1.0889,
+      "step": 22454
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859258551500096e-06,
+      "loss": 0.7566,
+      "step": 22455
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859163004012995e-06,
+      "loss": 0.5717,
+      "step": 22456
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8590674565258935e-06,
+      "loss": 0.6126,
+      "step": 22457
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8589719090387925e-06,
+      "loss": 0.4596,
+      "step": 22458
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.858876361551691e-06,
+      "loss": 0.8703,
+      "step": 22459
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.85878081406459e-06,
+      "loss": 1.0377,
+      "step": 22460
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8586852665774897e-06,
+      "loss": 0.9635,
+      "step": 22461
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8585897190903883e-06,
+      "loss": 0.9053,
+      "step": 22462
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8584941716032873e-06,
+      "loss": 0.6903,
+      "step": 22463
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.858398624116186e-06,
+      "loss": 1.0371,
+      "step": 22464
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.858303076629085e-06,
+      "loss": 1.1681,
+      "step": 22465
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8582075291419836e-06,
+      "loss": 0.7754,
+      "step": 22466
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8581119816548826e-06,
+      "loss": 0.5596,
+      "step": 22467
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8580164341677817e-06,
+      "loss": 0.7814,
+      "step": 22468
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8579208866806807e-06,
+      "loss": 0.5644,
+      "step": 22469
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8578253391935793e-06,
+      "loss": 1.0324,
+      "step": 22470
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8577297917064784e-06,
+      "loss": 1.2917,
+      "step": 22471
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8576342442193774e-06,
+      "loss": 0.9357,
+      "step": 22472
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.857538696732276e-06,
+      "loss": 1.2356,
+      "step": 22473
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.857443149245175e-06,
+      "loss": 0.7423,
+      "step": 22474
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.857347601758074e-06,
+      "loss": 0.5984,
+      "step": 22475
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.857252054270973e-06,
+      "loss": 0.845,
+      "step": 22476
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8571565067838717e-06,
+      "loss": 0.8936,
+      "step": 22477
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8570609592967708e-06,
+      "loss": 0.8189,
+      "step": 22478
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8569654118096694e-06,
+      "loss": 0.8839,
+      "step": 22479
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8568698643225684e-06,
+      "loss": 1.1319,
+      "step": 22480
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856774316835467e-06,
+      "loss": 0.605,
+      "step": 22481
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8566787693483665e-06,
+      "loss": 0.5228,
+      "step": 22482
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8565832218612655e-06,
+      "loss": 0.592,
+      "step": 22483
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856487674374164e-06,
+      "loss": 0.9587,
+      "step": 22484
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856392126887063e-06,
+      "loss": 1.0664,
+      "step": 22485
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856296579399962e-06,
+      "loss": 1.0068,
+      "step": 22486
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856201031912861e-06,
+      "loss": 0.4427,
+      "step": 22487
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8561054844257595e-06,
+      "loss": 0.6841,
+      "step": 22488
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.856009936938659e-06,
+      "loss": 0.662,
+      "step": 22489
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855914389451558e-06,
+      "loss": 0.8979,
+      "step": 22490
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8558188419644566e-06,
+      "loss": 1.5714,
+      "step": 22491
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8557232944773556e-06,
+      "loss": 0.6751,
+      "step": 22492
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8556277469902542e-06,
+      "loss": 0.7675,
+      "step": 22493
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8555321995031533e-06,
+      "loss": 0.577,
+      "step": 22494
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855436652016052e-06,
+      "loss": 0.6437,
+      "step": 22495
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855341104528951e-06,
+      "loss": 0.8209,
+      "step": 22496
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8552455570418504e-06,
+      "loss": 0.5211,
+      "step": 22497
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.855150009554749e-06,
+      "loss": 0.5922,
+      "step": 22498
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.855054462067648e-06,
+      "loss": 1.0759,
+      "step": 22499
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8549589145805467e-06,
+      "loss": 0.6689,
+      "step": 22500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8548633670934457e-06,
+      "loss": 1.0471,
+      "step": 22501
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8547678196063443e-06,
+      "loss": 0.9783,
+      "step": 22502
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8546722721192434e-06,
+      "loss": 0.755,
+      "step": 22503
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.854576724632143e-06,
+      "loss": 0.5493,
+      "step": 22504
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8544811771450414e-06,
+      "loss": 1.1515,
+      "step": 22505
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8543856296579405e-06,
+      "loss": 0.451,
+      "step": 22506
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.854290082170839e-06,
+      "loss": 1.4075,
+      "step": 22507
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.854194534683738e-06,
+      "loss": 0.8235,
+      "step": 22508
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8540989871966367e-06,
+      "loss": 0.7026,
+      "step": 22509
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8540034397095358e-06,
+      "loss": 0.5471,
+      "step": 22510
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8539078922224352e-06,
+      "loss": 0.7229,
+      "step": 22511
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.853812344735334e-06,
+      "loss": 0.8994,
+      "step": 22512
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.853716797248233e-06,
+      "loss": 1.0805,
+      "step": 22513
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8536212497611315e-06,
+      "loss": 0.6818,
+      "step": 22514
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8535257022740305e-06,
+      "loss": 0.934,
+      "step": 22515
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.853430154786929e-06,
+      "loss": 0.6128,
+      "step": 22516
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.853334607299828e-06,
+      "loss": 1.0415,
+      "step": 22517
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8532390598127272e-06,
+      "loss": 0.6148,
+      "step": 22518
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8531435123256263e-06,
+      "loss": 0.592,
+      "step": 22519
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.853047964838525e-06,
+      "loss": 0.7067,
+      "step": 22520
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.852952417351424e-06,
+      "loss": 1.1651,
+      "step": 22521
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8528568698643225e-06,
+      "loss": 0.4354,
+      "step": 22522
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8527613223772216e-06,
+      "loss": 0.6291,
+      "step": 22523
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8526657748901206e-06,
+      "loss": 0.7979,
+      "step": 22524
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8525702274030197e-06,
+      "loss": 0.7366,
+      "step": 22525
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8524746799159187e-06,
+      "loss": 1.6497,
+      "step": 22526
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8523791324288173e-06,
+      "loss": 0.9108,
+      "step": 22527
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8522835849417164e-06,
+      "loss": 0.6124,
+      "step": 22528
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.852188037454615e-06,
+      "loss": 1.162,
+      "step": 22529
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.852092489967514e-06,
+      "loss": 0.7006,
+      "step": 22530
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8519969424804126e-06,
+      "loss": 0.892,
+      "step": 22531
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8519013949933117e-06,
+      "loss": 0.6967,
+      "step": 22532
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.851805847506211e-06,
+      "loss": 1.109,
+      "step": 22533
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8517103000191097e-06,
+      "loss": 0.8099,
+      "step": 22534
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8516147525320088e-06,
+      "loss": 1.0985,
+      "step": 22535
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8515192050449074e-06,
+      "loss": 0.8952,
+      "step": 22536
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8514236575578064e-06,
+      "loss": 1.3518,
+      "step": 22537
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.851328110070705e-06,
+      "loss": 0.6455,
+      "step": 22538
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.851232562583604e-06,
+      "loss": 0.8987,
+      "step": 22539
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8511370150965035e-06,
+      "loss": 0.7252,
+      "step": 22540
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.851041467609402e-06,
+      "loss": 0.7498,
+      "step": 22541
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.850945920122301e-06,
+      "loss": 1.113,
+      "step": 22542
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8508503726352e-06,
+      "loss": 0.7428,
+      "step": 22543
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.850754825148099e-06,
+      "loss": 0.8549,
+      "step": 22544
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8506592776609975e-06,
+      "loss": 0.7731,
+      "step": 22545
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8505637301738965e-06,
+      "loss": 0.9441,
+      "step": 22546
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.850468182686796e-06,
+      "loss": 0.8754,
+      "step": 22547
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8503726351996946e-06,
+      "loss": 1.1114,
+      "step": 22548
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8502770877125936e-06,
+      "loss": 0.7958,
+      "step": 22549
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8501815402254922e-06,
+      "loss": 0.5376,
+      "step": 22550
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8500859927383913e-06,
+      "loss": 0.806,
+      "step": 22551
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.84999044525129e-06,
+      "loss": 0.7299,
+      "step": 22552
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.849894897764189e-06,
+      "loss": 0.9277,
+      "step": 22553
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8497993502770884e-06,
+      "loss": 0.7979,
+      "step": 22554
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.849703802789987e-06,
+      "loss": 0.6885,
+      "step": 22555
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.849608255302886e-06,
+      "loss": 1.0975,
+      "step": 22556
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8495127078157847e-06,
+      "loss": 1.0923,
+      "step": 22557
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8494171603286837e-06,
+      "loss": 0.5532,
+      "step": 22558
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8493216128415823e-06,
+      "loss": 0.7052,
+      "step": 22559
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8492260653544814e-06,
+      "loss": 0.6236,
+      "step": 22560
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8491305178673804e-06,
+      "loss": 0.8276,
+      "step": 22561
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8490349703802794e-06,
+      "loss": 0.7247,
+      "step": 22562
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8489394228931785e-06,
+      "loss": 1.0389,
+      "step": 22563
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.848843875406077e-06,
+      "loss": 0.7527,
+      "step": 22564
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.848748327918976e-06,
+      "loss": 0.6568,
+      "step": 22565
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8486527804318747e-06,
+      "loss": 0.6913,
+      "step": 22566
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8485572329447738e-06,
+      "loss": 1.0648,
+      "step": 22567
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8484616854576724e-06,
+      "loss": 0.519,
+      "step": 22568
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.848366137970572e-06,
+      "loss": 0.8471,
+      "step": 22569
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8482705904834705e-06,
+      "loss": 0.9875,
+      "step": 22570
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8481750429963695e-06,
+      "loss": 1.2572,
+      "step": 22571
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.848079495509268e-06,
+      "loss": 0.9071,
+      "step": 22572
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847983948022167e-06,
+      "loss": 0.642,
+      "step": 22573
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847888400535066e-06,
+      "loss": 0.991,
+      "step": 22574
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847792853047965e-06,
+      "loss": 0.6374,
+      "step": 22575
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8476973055608643e-06,
+      "loss": 0.549,
+      "step": 22576
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847601758073763e-06,
+      "loss": 1.1105,
+      "step": 22577
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847506210586662e-06,
+      "loss": 0.3983,
+      "step": 22578
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8474106630995605e-06,
+      "loss": 0.6143,
+      "step": 22579
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8473151156124596e-06,
+      "loss": 0.4906,
+      "step": 22580
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.847219568125358e-06,
+      "loss": 0.97,
+      "step": 22581
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8471240206382572e-06,
+      "loss": 1.2103,
+      "step": 22582
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8470284731511567e-06,
+      "loss": 0.8748,
+      "step": 22583
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8469329256640553e-06,
+      "loss": 0.6643,
+      "step": 22584
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8468373781769544e-06,
+      "loss": 0.8614,
+      "step": 22585
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.846741830689853e-06,
+      "loss": 1.2341,
+      "step": 22586
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.846646283202752e-06,
+      "loss": 0.9958,
+      "step": 22587
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8465507357156506e-06,
+      "loss": 0.8646,
+      "step": 22588
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8464551882285497e-06,
+      "loss": 1.08,
+      "step": 22589
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.846359640741449e-06,
+      "loss": 0.947,
+      "step": 22590
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8462640932543477e-06,
+      "loss": 0.637,
+      "step": 22591
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8461685457672468e-06,
+      "loss": 0.826,
+      "step": 22592
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8460729982801454e-06,
+      "loss": 1.0959,
+      "step": 22593
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8459774507930444e-06,
+      "loss": 1.166,
+      "step": 22594
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845881903305943e-06,
+      "loss": 1.1642,
+      "step": 22595
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845786355818842e-06,
+      "loss": 0.6252,
+      "step": 22596
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8456908083317415e-06,
+      "loss": 0.9126,
+      "step": 22597
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.84559526084464e-06,
+      "loss": 1.1469,
+      "step": 22598
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845499713357539e-06,
+      "loss": 0.7504,
+      "step": 22599
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845404165870438e-06,
+      "loss": 0.9295,
+      "step": 22600
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.845308618383337e-06,
+      "loss": 0.546,
+      "step": 22601
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8452130708962355e-06,
+      "loss": 0.7478,
+      "step": 22602
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8451175234091345e-06,
+      "loss": 0.7977,
+      "step": 22603
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.845021975922033e-06,
+      "loss": 0.6084,
+      "step": 22604
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8449264284349326e-06,
+      "loss": 0.4893,
+      "step": 22605
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8448308809478316e-06,
+      "loss": 0.7034,
+      "step": 22606
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8447353334607302e-06,
+      "loss": 0.8204,
+      "step": 22607
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8446397859736293e-06,
+      "loss": 0.9995,
+      "step": 22608
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.844544238486528e-06,
+      "loss": 0.7781,
+      "step": 22609
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.844448690999427e-06,
+      "loss": 0.844,
+      "step": 22610
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8443531435123255e-06,
+      "loss": 0.9416,
+      "step": 22611
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.844257596025225e-06,
+      "loss": 0.8165,
+      "step": 22612
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8441620485381236e-06,
+      "loss": 1.1293,
+      "step": 22613
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8440665010510227e-06,
+      "loss": 0.5739,
+      "step": 22614
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8439709535639217e-06,
+      "loss": 0.7457,
+      "step": 22615
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8438754060768203e-06,
+      "loss": 0.6479,
+      "step": 22616
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8437798585897194e-06,
+      "loss": 0.4999,
+      "step": 22617
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.843684311102618e-06,
+      "loss": 1.0179,
+      "step": 22618
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8435887636155174e-06,
+      "loss": 1.1133,
+      "step": 22619
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.843493216128416e-06,
+      "loss": 0.6575,
+      "step": 22620
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.843397668641315e-06,
+      "loss": 0.9569,
+      "step": 22621
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8433021211542137e-06,
+      "loss": 1.0014,
+      "step": 22622
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8432065736671127e-06,
+      "loss": 0.5325,
+      "step": 22623
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8431110261800114e-06,
+      "loss": 1.0679,
+      "step": 22624
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8430154786929104e-06,
+      "loss": 0.7728,
+      "step": 22625
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.84291993120581e-06,
+      "loss": 0.8833,
+      "step": 22626
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8428243837187085e-06,
+      "loss": 0.3711,
+      "step": 22627
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8427288362316075e-06,
+      "loss": 0.8129,
+      "step": 22628
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.842633288744506e-06,
+      "loss": 1.0555,
+      "step": 22629
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.842537741257405e-06,
+      "loss": 0.8226,
+      "step": 22630
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8424421937703038e-06,
+      "loss": 0.9192,
+      "step": 22631
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.842346646283203e-06,
+      "loss": 0.6698,
+      "step": 22632
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8422510987961023e-06,
+      "loss": 0.8305,
+      "step": 22633
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.842155551309001e-06,
+      "loss": 0.5324,
+      "step": 22634
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8420600038219e-06,
+      "loss": 0.5404,
+      "step": 22635
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8419644563347985e-06,
+      "loss": 0.9021,
+      "step": 22636
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8418689088476976e-06,
+      "loss": 0.8688,
+      "step": 22637
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.841773361360596e-06,
+      "loss": 0.7644,
+      "step": 22638
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8416778138734952e-06,
+      "loss": 0.6072,
+      "step": 22639
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.841582266386394e-06,
+      "loss": 0.6527,
+      "step": 22640
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8414867188992933e-06,
+      "loss": 0.6528,
+      "step": 22641
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8413911714121924e-06,
+      "loss": 1.0481,
+      "step": 22642
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.841295623925091e-06,
+      "loss": 0.8114,
+      "step": 22643
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.84120007643799e-06,
+      "loss": 1.4333,
+      "step": 22644
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8411045289508886e-06,
+      "loss": 0.9425,
+      "step": 22645
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8410089814637877e-06,
+      "loss": 0.6472,
+      "step": 22646
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8409134339766863e-06,
+      "loss": 0.5718,
+      "step": 22647
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8408178864895857e-06,
+      "loss": 0.7817,
+      "step": 22648
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8407223390024848e-06,
+      "loss": 0.7233,
+      "step": 22649
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8406267915153834e-06,
+      "loss": 1.0816,
+      "step": 22650
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8405312440282824e-06,
+      "loss": 0.6526,
+      "step": 22651
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.840435696541181e-06,
+      "loss": 1.1508,
+      "step": 22652
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.84034014905408e-06,
+      "loss": 0.6178,
+      "step": 22653
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8402446015669787e-06,
+      "loss": 0.6673,
+      "step": 22654
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.840149054079878e-06,
+      "loss": 0.6337,
+      "step": 22655
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.840053506592777e-06,
+      "loss": 1.0616,
+      "step": 22656
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839957959105676e-06,
+      "loss": 0.7069,
+      "step": 22657
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839862411618575e-06,
+      "loss": 0.427,
+      "step": 22658
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8397668641314735e-06,
+      "loss": 0.6634,
+      "step": 22659
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8396713166443725e-06,
+      "loss": 0.8014,
+      "step": 22660
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839575769157271e-06,
+      "loss": 0.6868,
+      "step": 22661
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8394802216701706e-06,
+      "loss": 1.0006,
+      "step": 22662
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839384674183069e-06,
+      "loss": 0.9809,
+      "step": 22663
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8392891266959682e-06,
+      "loss": 0.7593,
+      "step": 22664
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8391935792088673e-06,
+      "loss": 0.9757,
+      "step": 22665
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839098031721766e-06,
+      "loss": 1.6505,
+      "step": 22666
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.839002484234665e-06,
+      "loss": 0.5716,
+      "step": 22667
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8389069367475635e-06,
+      "loss": 0.9033,
+      "step": 22668
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.838811389260463e-06,
+      "loss": 0.6688,
+      "step": 22669
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8387158417733616e-06,
+      "loss": 0.4513,
+      "step": 22670
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8386202942862607e-06,
+      "loss": 0.5553,
+      "step": 22671
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8385247467991593e-06,
+      "loss": 0.7511,
+      "step": 22672
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8384291993120583e-06,
+      "loss": 1.1045,
+      "step": 22673
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.838333651824957e-06,
+      "loss": 0.8868,
+      "step": 22674
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.838238104337856e-06,
+      "loss": 0.9103,
+      "step": 22675
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.838142556850755e-06,
+      "loss": 1.268,
+      "step": 22676
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.838047009363654e-06,
+      "loss": 0.5921,
+      "step": 22677
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.837951461876553e-06,
+      "loss": 0.8657,
+      "step": 22678
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8378559143894517e-06,
+      "loss": 0.8104,
+      "step": 22679
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8377603669023507e-06,
+      "loss": 0.8854,
+      "step": 22680
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8376648194152494e-06,
+      "loss": 0.8781,
+      "step": 22681
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8375692719281484e-06,
+      "loss": 0.7965,
+      "step": 22682
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.837473724441047e-06,
+      "loss": 0.7747,
+      "step": 22683
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8373781769539465e-06,
+      "loss": 0.9078,
+      "step": 22684
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8372826294668455e-06,
+      "loss": 1.1864,
+      "step": 22685
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.837187081979744e-06,
+      "loss": 0.9124,
+      "step": 22686
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.837091534492643e-06,
+      "loss": 1.0041,
+      "step": 22687
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8369959870055418e-06,
+      "loss": 0.8772,
+      "step": 22688
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.836900439518441e-06,
+      "loss": 0.932,
+      "step": 22689
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8368048920313394e-06,
+      "loss": 0.7054,
+      "step": 22690
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.836709344544239e-06,
+      "loss": 0.6778,
+      "step": 22691
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.836613797057138e-06,
+      "loss": 1.0735,
+      "step": 22692
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8365182495700365e-06,
+      "loss": 0.6526,
+      "step": 22693
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8364227020829356e-06,
+      "loss": 0.5513,
+      "step": 22694
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.836327154595834e-06,
+      "loss": 0.9175,
+      "step": 22695
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8362316071087332e-06,
+      "loss": 0.6955,
+      "step": 22696
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.836136059621632e-06,
+      "loss": 0.965,
+      "step": 22697
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8360405121345313e-06,
+      "loss": 0.7497,
+      "step": 22698
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8359449646474304e-06,
+      "loss": 0.7937,
+      "step": 22699
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.835849417160329e-06,
+      "loss": 0.3683,
+      "step": 22700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.835753869673228e-06,
+      "loss": 0.9493,
+      "step": 22701
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8356583221861266e-06,
+      "loss": 0.5784,
+      "step": 22702
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8355627746990257e-06,
+      "loss": 0.991,
+      "step": 22703
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8354672272119243e-06,
+      "loss": 0.4421,
+      "step": 22704
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8353716797248237e-06,
+      "loss": 1.2902,
+      "step": 22705
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8352761322377228e-06,
+      "loss": 0.7793,
+      "step": 22706
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8351805847506214e-06,
+      "loss": 0.996,
+      "step": 22707
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8350850372635204e-06,
+      "loss": 0.9516,
+      "step": 22708
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834989489776419e-06,
+      "loss": 0.6824,
+      "step": 22709
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834893942289318e-06,
+      "loss": 0.8009,
+      "step": 22710
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8347983948022167e-06,
+      "loss": 0.8344,
+      "step": 22711
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8347028473151157e-06,
+      "loss": 0.6394,
+      "step": 22712
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8346072998280148e-06,
+      "loss": 1.1036,
+      "step": 22713
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834511752340914e-06,
+      "loss": 1.0028,
+      "step": 22714
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8344162048538124e-06,
+      "loss": 0.62,
+      "step": 22715
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8343206573667115e-06,
+      "loss": 0.3694,
+      "step": 22716
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8342251098796105e-06,
+      "loss": 0.6792,
+      "step": 22717
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834129562392509e-06,
+      "loss": 0.7037,
+      "step": 22718
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.834034014905408e-06,
+      "loss": 0.5224,
+      "step": 22719
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.833938467418307e-06,
+      "loss": 0.5016,
+      "step": 22720
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8338429199312062e-06,
+      "loss": 0.8216,
+      "step": 22721
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.833747372444105e-06,
+      "loss": 0.8339,
+      "step": 22722
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.833651824957004e-06,
+      "loss": 0.8986,
+      "step": 22723
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8335562774699025e-06,
+      "loss": 1.0411,
+      "step": 22724
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8334607299828015e-06,
+      "loss": 0.6484,
+      "step": 22725
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8333651824957e-06,
+      "loss": 0.5051,
+      "step": 22726
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8332696350085996e-06,
+      "loss": 0.44,
+      "step": 22727
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8331740875214987e-06,
+      "loss": 0.6729,
+      "step": 22728
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8330785400343973e-06,
+      "loss": 0.6137,
+      "step": 22729
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8329829925472963e-06,
+      "loss": 0.8551,
+      "step": 22730
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.832887445060195e-06,
+      "loss": 0.7202,
+      "step": 22731
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.832791897573094e-06,
+      "loss": 0.7557,
+      "step": 22732
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8326963500859926e-06,
+      "loss": 0.6802,
+      "step": 22733
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.832600802598892e-06,
+      "loss": 0.7505,
+      "step": 22734
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.832505255111791e-06,
+      "loss": 0.9215,
+      "step": 22735
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8324097076246897e-06,
+      "loss": 0.939,
+      "step": 22736
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8323141601375887e-06,
+      "loss": 0.7851,
+      "step": 22737
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8322186126504873e-06,
+      "loss": 0.8978,
+      "step": 22738
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8321230651633864e-06,
+      "loss": 0.5206,
+      "step": 22739
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.832027517676285e-06,
+      "loss": 0.7109,
+      "step": 22740
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8319319701891845e-06,
+      "loss": 0.7443,
+      "step": 22741
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8318364227020835e-06,
+      "loss": 0.4025,
+      "step": 22742
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.831740875214982e-06,
+      "loss": 0.6444,
+      "step": 22743
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.831645327727881e-06,
+      "loss": 0.7188,
+      "step": 22744
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8315497802407798e-06,
+      "loss": 0.6891,
+      "step": 22745
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.831454232753679e-06,
+      "loss": 0.8756,
+      "step": 22746
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8313586852665774e-06,
+      "loss": 0.3614,
+      "step": 22747
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8312631377794765e-06,
+      "loss": 0.5719,
+      "step": 22748
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.831167590292376e-06,
+      "loss": 0.5748,
+      "step": 22749
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8310720428052745e-06,
+      "loss": 0.9905,
+      "step": 22750
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8309764953181736e-06,
+      "loss": 0.7156,
+      "step": 22751
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.830880947831072e-06,
+      "loss": 0.8631,
+      "step": 22752
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8307854003439712e-06,
+      "loss": 0.8199,
+      "step": 22753
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.83068985285687e-06,
+      "loss": 0.4139,
+      "step": 22754
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.830594305369769e-06,
+      "loss": 0.9515,
+      "step": 22755
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8304987578826683e-06,
+      "loss": 0.6233,
+      "step": 22756
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.830403210395567e-06,
+      "loss": 0.5261,
+      "step": 22757
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.830307662908466e-06,
+      "loss": 0.9926,
+      "step": 22758
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8302121154213646e-06,
+      "loss": 1.2006,
+      "step": 22759
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8301165679342637e-06,
+      "loss": 1.2077,
+      "step": 22760
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8300210204471623e-06,
+      "loss": 1.3356,
+      "step": 22761
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8299254729600613e-06,
+      "loss": 1.1161,
+      "step": 22762
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8298299254729603e-06,
+      "loss": 0.6181,
+      "step": 22763
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8297343779858594e-06,
+      "loss": 0.6297,
+      "step": 22764
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829638830498758e-06,
+      "loss": 0.887,
+      "step": 22765
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829543283011657e-06,
+      "loss": 0.812,
+      "step": 22766
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829447735524556e-06,
+      "loss": 1.0253,
+      "step": 22767
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8293521880374547e-06,
+      "loss": 0.8742,
+      "step": 22768
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8292566405503537e-06,
+      "loss": 0.9252,
+      "step": 22769
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8291610930632528e-06,
+      "loss": 0.8816,
+      "step": 22770
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.829065545576152e-06,
+      "loss": 0.8399,
+      "step": 22771
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8289699980890504e-06,
+      "loss": 0.6001,
+      "step": 22772
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8288744506019495e-06,
+      "loss": 0.3819,
+      "step": 22773
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.828778903114848e-06,
+      "loss": 1.0301,
+      "step": 22774
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.828683355627747e-06,
+      "loss": 0.6593,
+      "step": 22775
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8285878081406457e-06,
+      "loss": 0.6899,
+      "step": 22776
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.828492260653545e-06,
+      "loss": 0.8687,
+      "step": 22777
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8283967131664442e-06,
+      "loss": 0.6521,
+      "step": 22778
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.828301165679343e-06,
+      "loss": 0.8234,
+      "step": 22779
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.828205618192242e-06,
+      "loss": 1.0404,
+      "step": 22780
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8281100707051405e-06,
+      "loss": 1.0134,
+      "step": 22781
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8280145232180395e-06,
+      "loss": 0.7003,
+      "step": 22782
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.827918975730938e-06,
+      "loss": 1.1954,
+      "step": 22783
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.827823428243837e-06,
+      "loss": 0.8697,
+      "step": 22784
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8277278807567367e-06,
+      "loss": 1.0138,
+      "step": 22785
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8276323332696353e-06,
+      "loss": 0.5019,
+      "step": 22786
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8275367857825343e-06,
+      "loss": 0.5366,
+      "step": 22787
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.827441238295433e-06,
+      "loss": 1.0232,
+      "step": 22788
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.827345690808332e-06,
+      "loss": 0.5965,
+      "step": 22789
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8272501433212306e-06,
+      "loss": 0.9261,
+      "step": 22790
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8271545958341296e-06,
+      "loss": 0.8668,
+      "step": 22791
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.827059048347029e-06,
+      "loss": 0.9882,
+      "step": 22792
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8269635008599277e-06,
+      "loss": 0.653,
+      "step": 22793
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8268679533728267e-06,
+      "loss": 1.0154,
+      "step": 22794
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8267724058857253e-06,
+      "loss": 1.1614,
+      "step": 22795
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8266768583986244e-06,
+      "loss": 0.7418,
+      "step": 22796
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.826581310911523e-06,
+      "loss": 0.7934,
+      "step": 22797
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.826485763424422e-06,
+      "loss": 0.8895,
+      "step": 22798
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8263902159373215e-06,
+      "loss": 1.168,
+      "step": 22799
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.82629466845022e-06,
+      "loss": 0.5065,
+      "step": 22800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.826199120963119e-06,
+      "loss": 0.6051,
+      "step": 22801
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8261035734760178e-06,
+      "loss": 1.0161,
+      "step": 22802
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.826008025988917e-06,
+      "loss": 0.5244,
+      "step": 22803
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8259124785018154e-06,
+      "loss": 1.1568,
+      "step": 22804
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8258169310147145e-06,
+      "loss": 0.738,
+      "step": 22805
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.825721383527614e-06,
+      "loss": 0.7785,
+      "step": 22806
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8256258360405125e-06,
+      "loss": 1.0475,
+      "step": 22807
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8255302885534116e-06,
+      "loss": 1.2673,
+      "step": 22808
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.82543474106631e-06,
+      "loss": 0.8817,
+      "step": 22809
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8253391935792092e-06,
+      "loss": 1.0522,
+      "step": 22810
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.825243646092108e-06,
+      "loss": 0.9189,
+      "step": 22811
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.825148098605007e-06,
+      "loss": 0.5762,
+      "step": 22812
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.825052551117906e-06,
+      "loss": 0.8784,
+      "step": 22813
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.824957003630805e-06,
+      "loss": 1.1042,
+      "step": 22814
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8248614561437036e-06,
+      "loss": 0.9285,
+      "step": 22815
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8247659086566026e-06,
+      "loss": 0.8726,
+      "step": 22816
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8246703611695012e-06,
+      "loss": 0.7694,
+      "step": 22817
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8245748136824003e-06,
+      "loss": 1.2109,
+      "step": 22818
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8244792661952993e-06,
+      "loss": 0.6151,
+      "step": 22819
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.824383718708198e-06,
+      "loss": 0.702,
+      "step": 22820
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8242881712210974e-06,
+      "loss": 0.6741,
+      "step": 22821
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.824192623733996e-06,
+      "loss": 0.9902,
+      "step": 22822
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.824097076246895e-06,
+      "loss": 0.618,
+      "step": 22823
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8240015287597937e-06,
+      "loss": 1.1218,
+      "step": 22824
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8239059812726927e-06,
+      "loss": 0.8398,
+      "step": 22825
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8238104337855913e-06,
+      "loss": 0.8194,
+      "step": 22826
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8237148862984903e-06,
+      "loss": 1.1454,
+      "step": 22827
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.82361933881139e-06,
+      "loss": 1.0442,
+      "step": 22828
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8235237913242884e-06,
+      "loss": 1.077,
+      "step": 22829
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8234282438371875e-06,
+      "loss": 1.0665,
+      "step": 22830
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.823332696350086e-06,
+      "loss": 0.582,
+      "step": 22831
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.823237148862985e-06,
+      "loss": 0.6813,
+      "step": 22832
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8231416013758837e-06,
+      "loss": 0.9349,
+      "step": 22833
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8230460538887828e-06,
+      "loss": 0.7886,
+      "step": 22834
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8229505064016822e-06,
+      "loss": 0.8447,
+      "step": 22835
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.822854958914581e-06,
+      "loss": 0.7945,
+      "step": 22836
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.82275941142748e-06,
+      "loss": 0.9654,
+      "step": 22837
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8226638639403785e-06,
+      "loss": 1.1517,
+      "step": 22838
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8225683164532775e-06,
+      "loss": 0.923,
+      "step": 22839
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.822472768966176e-06,
+      "loss": 0.8041,
+      "step": 22840
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.822377221479075e-06,
+      "loss": 0.9618,
+      "step": 22841
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8222816739919747e-06,
+      "loss": 1.1081,
+      "step": 22842
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8221861265048733e-06,
+      "loss": 0.9031,
+      "step": 22843
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8220905790177723e-06,
+      "loss": 0.8855,
+      "step": 22844
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.821995031530671e-06,
+      "loss": 0.5251,
+      "step": 22845
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.82189948404357e-06,
+      "loss": 0.8027,
+      "step": 22846
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8218039365564686e-06,
+      "loss": 0.5033,
+      "step": 22847
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8217083890693676e-06,
+      "loss": 0.7325,
+      "step": 22848
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.821612841582267e-06,
+      "loss": 0.7798,
+      "step": 22849
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8215172940951657e-06,
+      "loss": 0.5537,
+      "step": 22850
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8214217466080647e-06,
+      "loss": 0.7141,
+      "step": 22851
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8213261991209633e-06,
+      "loss": 0.52,
+      "step": 22852
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8212306516338624e-06,
+      "loss": 0.7209,
+      "step": 22853
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.821135104146761e-06,
+      "loss": 0.934,
+      "step": 22854
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.82103955665966e-06,
+      "loss": 0.9224,
+      "step": 22855
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8209440091725587e-06,
+      "loss": 1.1032,
+      "step": 22856
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.820848461685458e-06,
+      "loss": 0.6855,
+      "step": 22857
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.820752914198357e-06,
+      "loss": 0.7263,
+      "step": 22858
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8206573667112558e-06,
+      "loss": 1.0272,
+      "step": 22859
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.820561819224155e-06,
+      "loss": 0.5164,
+      "step": 22860
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8204662717370534e-06,
+      "loss": 0.943,
+      "step": 22861
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8203707242499525e-06,
+      "loss": 0.478,
+      "step": 22862
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.820275176762851e-06,
+      "loss": 0.5052,
+      "step": 22863
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8201796292757505e-06,
+      "loss": 0.4994,
+      "step": 22864
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.820084081788649e-06,
+      "loss": 0.859,
+      "step": 22865
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819988534301548e-06,
+      "loss": 0.8515,
+      "step": 22866
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819892986814447e-06,
+      "loss": 0.5807,
+      "step": 22867
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819797439327346e-06,
+      "loss": 0.6958,
+      "step": 22868
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819701891840245e-06,
+      "loss": 0.9549,
+      "step": 22869
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8196063443531435e-06,
+      "loss": 0.9856,
+      "step": 22870
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819510796866043e-06,
+      "loss": 0.6955,
+      "step": 22871
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8194152493789416e-06,
+      "loss": 0.5393,
+      "step": 22872
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8193197018918406e-06,
+      "loss": 0.9707,
+      "step": 22873
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8192241544047392e-06,
+      "loss": 1.1882,
+      "step": 22874
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8191286069176383e-06,
+      "loss": 0.5893,
+      "step": 22875
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.819033059430537e-06,
+      "loss": 0.7231,
+      "step": 22876
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.818937511943436e-06,
+      "loss": 1.0206,
+      "step": 22877
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8188419644563354e-06,
+      "loss": 0.7534,
+      "step": 22878
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.818746416969234e-06,
+      "loss": 0.7317,
+      "step": 22879
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.818650869482133e-06,
+      "loss": 0.9843,
+      "step": 22880
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8185553219950317e-06,
+      "loss": 0.6153,
+      "step": 22881
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8184597745079307e-06,
+      "loss": 0.9821,
+      "step": 22882
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8183642270208293e-06,
+      "loss": 0.8294,
+      "step": 22883
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8182686795337283e-06,
+      "loss": 1.1654,
+      "step": 22884
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.818173132046628e-06,
+      "loss": 0.7555,
+      "step": 22885
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8180775845595264e-06,
+      "loss": 0.7947,
+      "step": 22886
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8179820370724255e-06,
+      "loss": 0.6101,
+      "step": 22887
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817886489585324e-06,
+      "loss": 1.2586,
+      "step": 22888
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817790942098223e-06,
+      "loss": 1.035,
+      "step": 22889
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8176953946111217e-06,
+      "loss": 0.5417,
+      "step": 22890
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8175998471240208e-06,
+      "loss": 0.8109,
+      "step": 22891
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8175042996369194e-06,
+      "loss": 0.881,
+      "step": 22892
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817408752149819e-06,
+      "loss": 0.7563,
+      "step": 22893
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817313204662718e-06,
+      "loss": 1.1378,
+      "step": 22894
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8172176571756165e-06,
+      "loss": 1.2533,
+      "step": 22895
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8171221096885155e-06,
+      "loss": 0.8823,
+      "step": 22896
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817026562201414e-06,
+      "loss": 0.5968,
+      "step": 22897
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816931014714313e-06,
+      "loss": 1.2179,
+      "step": 22898
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816835467227212e-06,
+      "loss": 0.5256,
+      "step": 22899
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8167399197401113e-06,
+      "loss": 0.9322,
+      "step": 22900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8166443722530103e-06,
+      "loss": 0.5607,
+      "step": 22901
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816548824765909e-06,
+      "loss": 0.4739,
+      "step": 22902
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816453277278808e-06,
+      "loss": 0.6816,
+      "step": 22903
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8163577297917066e-06,
+      "loss": 0.7343,
+      "step": 22904
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8162621823046056e-06,
+      "loss": 0.8067,
+      "step": 22905
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8161666348175042e-06,
+      "loss": 0.55,
+      "step": 22906
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8160710873304037e-06,
+      "loss": 1.005,
+      "step": 22907
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8159755398433023e-06,
+      "loss": 0.8056,
+      "step": 22908
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8158799923562013e-06,
+      "loss": 0.4819,
+      "step": 22909
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8157844448691004e-06,
+      "loss": 1.199,
+      "step": 22910
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.815688897381999e-06,
+      "loss": 0.5067,
+      "step": 22911
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.815593349894898e-06,
+      "loss": 0.7752,
+      "step": 22912
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8154978024077967e-06,
+      "loss": 0.7425,
+      "step": 22913
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.815402254920696e-06,
+      "loss": 1.0679,
+      "step": 22914
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8153067074335947e-06,
+      "loss": 1.1107,
+      "step": 22915
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8152111599464938e-06,
+      "loss": 1.0234,
+      "step": 22916
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8151156124593924e-06,
+      "loss": 0.741,
+      "step": 22917
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8150200649722914e-06,
+      "loss": 1.086,
+      "step": 22918
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.81492451748519e-06,
+      "loss": 1.1555,
+      "step": 22919
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814828969998089e-06,
+      "loss": 0.4197,
+      "step": 22920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8147334225109885e-06,
+      "loss": 1.1752,
+      "step": 22921
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814637875023887e-06,
+      "loss": 0.7796,
+      "step": 22922
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814542327536786e-06,
+      "loss": 0.4526,
+      "step": 22923
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814446780049685e-06,
+      "loss": 0.6849,
+      "step": 22924
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.814351232562584e-06,
+      "loss": 1.0462,
+      "step": 22925
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8142556850754825e-06,
+      "loss": 0.936,
+      "step": 22926
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8141601375883815e-06,
+      "loss": 0.9088,
+      "step": 22927
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.81406459010128e-06,
+      "loss": 0.5945,
+      "step": 22928
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8139690426141796e-06,
+      "loss": 0.4414,
+      "step": 22929
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8138734951270786e-06,
+      "loss": 0.7875,
+      "step": 22930
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8137779476399772e-06,
+      "loss": 0.5211,
+      "step": 22931
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8136824001528763e-06,
+      "loss": 0.3472,
+      "step": 22932
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.813586852665775e-06,
+      "loss": 0.5255,
+      "step": 22933
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.813491305178674e-06,
+      "loss": 0.9256,
+      "step": 22934
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8133957576915725e-06,
+      "loss": 1.0195,
+      "step": 22935
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.813300210204472e-06,
+      "loss": 1.114,
+      "step": 22936
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.813204662717371e-06,
+      "loss": 0.6746,
+      "step": 22937
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8131091152302697e-06,
+      "loss": 0.6221,
+      "step": 22938
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8130135677431687e-06,
+      "loss": 0.4333,
+      "step": 22939
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8129180202560673e-06,
+      "loss": 0.966,
+      "step": 22940
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8128224727689663e-06,
+      "loss": 0.7567,
+      "step": 22941
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.812726925281865e-06,
+      "loss": 1.555,
+      "step": 22942
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8126313777947644e-06,
+      "loss": 0.831,
+      "step": 22943
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8125358303076635e-06,
+      "loss": 0.8623,
+      "step": 22944
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.812440282820562e-06,
+      "loss": 1.0071,
+      "step": 22945
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.812344735333461e-06,
+      "loss": 1.0589,
+      "step": 22946
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8122491878463597e-06,
+      "loss": 0.8616,
+      "step": 22947
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8121536403592588e-06,
+      "loss": 1.0031,
+      "step": 22948
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8120580928721574e-06,
+      "loss": 0.8692,
+      "step": 22949
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811962545385057e-06,
+      "loss": 0.9908,
+      "step": 22950
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811866997897956e-06,
+      "loss": 0.8378,
+      "step": 22951
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8117714504108545e-06,
+      "loss": 0.7644,
+      "step": 22952
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8116759029237535e-06,
+      "loss": 0.5999,
+      "step": 22953
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811580355436652e-06,
+      "loss": 0.5477,
+      "step": 22954
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811484807949551e-06,
+      "loss": 0.4449,
+      "step": 22955
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.81138926046245e-06,
+      "loss": 0.8542,
+      "step": 22956
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8112937129753493e-06,
+      "loss": 0.6532,
+      "step": 22957
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811198165488248e-06,
+      "loss": 0.6006,
+      "step": 22958
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811102618001147e-06,
+      "loss": 0.8902,
+      "step": 22959
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811007070514046e-06,
+      "loss": 0.941,
+      "step": 22960
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8109115230269446e-06,
+      "loss": 0.8019,
+      "step": 22961
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8108159755398436e-06,
+      "loss": 0.8964,
+      "step": 22962
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8107204280527422e-06,
+      "loss": 1.115,
+      "step": 22963
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8106248805656413e-06,
+      "loss": 0.8469,
+      "step": 22964
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8105293330785403e-06,
+      "loss": 0.7585,
+      "step": 22965
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8104337855914393e-06,
+      "loss": 0.8977,
+      "step": 22966
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.810338238104338e-06,
+      "loss": 0.983,
+      "step": 22967
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.810242690617237e-06,
+      "loss": 1.015,
+      "step": 22968
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8101471431301356e-06,
+      "loss": 0.7854,
+      "step": 22969
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8100515956430347e-06,
+      "loss": 0.4071,
+      "step": 22970
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8099560481559337e-06,
+      "loss": 0.3712,
+      "step": 22971
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8098605006688327e-06,
+      "loss": 0.9128,
+      "step": 22972
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8097649531817318e-06,
+      "loss": 0.9535,
+      "step": 22973
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8096694056946304e-06,
+      "loss": 1.3826,
+      "step": 22974
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8095738582075294e-06,
+      "loss": 0.9847,
+      "step": 22975
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809478310720428e-06,
+      "loss": 0.6706,
+      "step": 22976
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809382763233327e-06,
+      "loss": 0.5987,
+      "step": 22977
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8092872157462257e-06,
+      "loss": 0.7469,
+      "step": 22978
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809191668259125e-06,
+      "loss": 0.6431,
+      "step": 22979
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809096120772024e-06,
+      "loss": 0.9331,
+      "step": 22980
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.809000573284923e-06,
+      "loss": 0.6306,
+      "step": 22981
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.808905025797822e-06,
+      "loss": 0.6017,
+      "step": 22982
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8088094783107205e-06,
+      "loss": 0.9235,
+      "step": 22983
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8087139308236195e-06,
+      "loss": 0.3681,
+      "step": 22984
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.808618383336518e-06,
+      "loss": 0.6038,
+      "step": 22985
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8085228358494176e-06,
+      "loss": 0.6777,
+      "step": 22986
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8084272883623166e-06,
+      "loss": 1.0894,
+      "step": 22987
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8083317408752152e-06,
+      "loss": 0.6159,
+      "step": 22988
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8082361933881143e-06,
+      "loss": 1.0367,
+      "step": 22989
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.808140645901013e-06,
+      "loss": 0.6613,
+      "step": 22990
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.808045098413912e-06,
+      "loss": 1.0418,
+      "step": 22991
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8079495509268105e-06,
+      "loss": 1.2777,
+      "step": 22992
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.80785400343971e-06,
+      "loss": 0.6723,
+      "step": 22993
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.807758455952609e-06,
+      "loss": 1.0265,
+      "step": 22994
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8076629084655077e-06,
+      "loss": 1.1434,
+      "step": 22995
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8075673609784067e-06,
+      "loss": 1.0343,
+      "step": 22996
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8074718134913053e-06,
+      "loss": 0.8359,
+      "step": 22997
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8073762660042043e-06,
+      "loss": 0.6169,
+      "step": 22998
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.807280718517103e-06,
+      "loss": 1.0351,
+      "step": 22999
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.807185171030002e-06,
+      "loss": 0.9907,
+      "step": 23000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8070896235429015e-06,
+      "loss": 0.7164,
+      "step": 23001
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8069940760558e-06,
+      "loss": 0.6685,
+      "step": 23002
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806898528568699e-06,
+      "loss": 1.2275,
+      "step": 23003
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8068029810815977e-06,
+      "loss": 0.9569,
+      "step": 23004
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8067074335944968e-06,
+      "loss": 1.0617,
+      "step": 23005
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8066118861073954e-06,
+      "loss": 0.8678,
+      "step": 23006
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8065163386202944e-06,
+      "loss": 0.931,
+      "step": 23007
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8064207911331935e-06,
+      "loss": 0.829,
+      "step": 23008
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8063252436460925e-06,
+      "loss": 0.9725,
+      "step": 23009
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806229696158991e-06,
+      "loss": 0.703,
+      "step": 23010
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.80613414867189e-06,
+      "loss": 0.7064,
+      "step": 23011
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806038601184789e-06,
+      "loss": 0.9734,
+      "step": 23012
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805943053697688e-06,
+      "loss": 0.6952,
+      "step": 23013
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805847506210587e-06,
+      "loss": 1.0234,
+      "step": 23014
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805751958723486e-06,
+      "loss": 0.7991,
+      "step": 23015
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805656411236385e-06,
+      "loss": 0.8019,
+      "step": 23016
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8055608637492835e-06,
+      "loss": 1.1802,
+      "step": 23017
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8054653162621826e-06,
+      "loss": 0.8602,
+      "step": 23018
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805369768775081e-06,
+      "loss": 0.8329,
+      "step": 23019
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8052742212879802e-06,
+      "loss": 0.7623,
+      "step": 23020
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.805178673800879e-06,
+      "loss": 0.8012,
+      "step": 23021
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8050831263137783e-06,
+      "loss": 1.1096,
+      "step": 23022
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8049875788266773e-06,
+      "loss": 0.7637,
+      "step": 23023
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.804892031339576e-06,
+      "loss": 1.5276,
+      "step": 23024
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.804796483852475e-06,
+      "loss": 0.869,
+      "step": 23025
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8047009363653736e-06,
+      "loss": 0.6069,
+      "step": 23026
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8046053888782727e-06,
+      "loss": 0.5944,
+      "step": 23027
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8045098413911713e-06,
+      "loss": 0.7008,
+      "step": 23028
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8044142939040707e-06,
+      "loss": 0.8592,
+      "step": 23029
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8043187464169698e-06,
+      "loss": 0.9812,
+      "step": 23030
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8042231989298684e-06,
+      "loss": 0.8285,
+      "step": 23031
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8041276514427674e-06,
+      "loss": 1.0474,
+      "step": 23032
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.804032103955666e-06,
+      "loss": 0.8511,
+      "step": 23033
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.803936556468565e-06,
+      "loss": 0.7121,
+      "step": 23034
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8038410089814637e-06,
+      "loss": 1.0062,
+      "step": 23035
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8037454614943627e-06,
+      "loss": 0.8506,
+      "step": 23036
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.803649914007262e-06,
+      "loss": 1.0116,
+      "step": 23037
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.803554366520161e-06,
+      "loss": 0.7327,
+      "step": 23038
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.80345881903306e-06,
+      "loss": 0.7425,
+      "step": 23039
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8033632715459585e-06,
+      "loss": 1.0525,
+      "step": 23040
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8032677240588575e-06,
+      "loss": 0.9772,
+      "step": 23041
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.803172176571756e-06,
+      "loss": 0.9752,
+      "step": 23042
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.803076629084655e-06,
+      "loss": 0.4365,
+      "step": 23043
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8029810815975546e-06,
+      "loss": 0.8017,
+      "step": 23044
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8028855341104532e-06,
+      "loss": 1.1602,
+      "step": 23045
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8027899866233523e-06,
+      "loss": 0.9203,
+      "step": 23046
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.802694439136251e-06,
+      "loss": 0.5556,
+      "step": 23047
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.80259889164915e-06,
+      "loss": 0.8883,
+      "step": 23048
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8025033441620485e-06,
+      "loss": 0.7611,
+      "step": 23049
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8024077966749476e-06,
+      "loss": 0.7337,
+      "step": 23050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.802312249187847e-06,
+      "loss": 1.1088,
+      "step": 23051
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8022167017007457e-06,
+      "loss": 0.9116,
+      "step": 23052
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8021211542136447e-06,
+      "loss": 1.4273,
+      "step": 23053
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8020256067265433e-06,
+      "loss": 0.8055,
+      "step": 23054
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8019300592394423e-06,
+      "loss": 0.6767,
+      "step": 23055
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.801834511752341e-06,
+      "loss": 0.4389,
+      "step": 23056
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.80173896426524e-06,
+      "loss": 0.8218,
+      "step": 23057
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.801643416778139e-06,
+      "loss": 0.9092,
+      "step": 23058
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.801547869291038e-06,
+      "loss": 0.6457,
+      "step": 23059
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8014523218039367e-06,
+      "loss": 0.7565,
+      "step": 23060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8013567743168357e-06,
+      "loss": 0.5814,
+      "step": 23061
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8012612268297348e-06,
+      "loss": 1.4455,
+      "step": 23062
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8011656793426334e-06,
+      "loss": 0.9143,
+      "step": 23063
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8010701318555324e-06,
+      "loss": 0.8706,
+      "step": 23064
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8009745843684315e-06,
+      "loss": 0.5925,
+      "step": 23065
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8008790368813305e-06,
+      "loss": 0.9361,
+      "step": 23066
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800783489394229e-06,
+      "loss": 0.8492,
+      "step": 23067
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800687941907128e-06,
+      "loss": 1.1021,
+      "step": 23068
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8005923944200268e-06,
+      "loss": 0.8145,
+      "step": 23069
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800496846932926e-06,
+      "loss": 0.4325,
+      "step": 23070
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8004012994458244e-06,
+      "loss": 0.9046,
+      "step": 23071
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8003057519587235e-06,
+      "loss": 0.8729,
+      "step": 23072
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800210204471623e-06,
+      "loss": 1.1801,
+      "step": 23073
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8001146569845215e-06,
+      "loss": 1.1893,
+      "step": 23074
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.8000191094974206e-06,
+      "loss": 0.7797,
+      "step": 23075
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799923562010319e-06,
+      "loss": 0.5723,
+      "step": 23076
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7998280145232182e-06,
+      "loss": 0.8964,
+      "step": 23077
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799732467036117e-06,
+      "loss": 1.1382,
+      "step": 23078
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799636919549016e-06,
+      "loss": 1.2634,
+      "step": 23079
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7995413720619153e-06,
+      "loss": 0.7637,
+      "step": 23080
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799445824574814e-06,
+      "loss": 0.7656,
+      "step": 23081
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.799350277087713e-06,
+      "loss": 0.7435,
+      "step": 23082
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7992547296006116e-06,
+      "loss": 1.0413,
+      "step": 23083
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7991591821135107e-06,
+      "loss": 0.5738,
+      "step": 23084
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7990636346264093e-06,
+      "loss": 0.4337,
+      "step": 23085
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7989680871393083e-06,
+      "loss": 0.4382,
+      "step": 23086
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7988725396522078e-06,
+      "loss": 0.7777,
+      "step": 23087
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7987769921651064e-06,
+      "loss": 0.6951,
+      "step": 23088
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7986814446780054e-06,
+      "loss": 0.7953,
+      "step": 23089
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.798585897190904e-06,
+      "loss": 0.8594,
+      "step": 23090
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.798490349703803e-06,
+      "loss": 0.7571,
+      "step": 23091
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7983948022167017e-06,
+      "loss": 0.7789,
+      "step": 23092
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7982992547296007e-06,
+      "loss": 0.9961,
+      "step": 23093
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7982037072425e-06,
+      "loss": 1.1844,
+      "step": 23094
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.798108159755399e-06,
+      "loss": 0.6892,
+      "step": 23095
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.798012612268298e-06,
+      "loss": 0.7258,
+      "step": 23096
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7979170647811965e-06,
+      "loss": 0.8417,
+      "step": 23097
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7978215172940955e-06,
+      "loss": 0.9614,
+      "step": 23098
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797725969806994e-06,
+      "loss": 0.6897,
+      "step": 23099
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797630422319893e-06,
+      "loss": 1.0666,
+      "step": 23100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7975348748327926e-06,
+      "loss": 0.7236,
+      "step": 23101
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7974393273456912e-06,
+      "loss": 0.8973,
+      "step": 23102
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7973437798585903e-06,
+      "loss": 1.1047,
+      "step": 23103
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797248232371489e-06,
+      "loss": 0.6616,
+      "step": 23104
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797152684884388e-06,
+      "loss": 0.5609,
+      "step": 23105
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7970571373972865e-06,
+      "loss": 0.9353,
+      "step": 23106
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7969615899101856e-06,
+      "loss": 0.5898,
+      "step": 23107
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.796866042423084e-06,
+      "loss": 0.772,
+      "step": 23108
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7967704949359837e-06,
+      "loss": 0.7696,
+      "step": 23109
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7966749474488823e-06,
+      "loss": 0.6285,
+      "step": 23110
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7965793999617813e-06,
+      "loss": 0.4661,
+      "step": 23111
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.79648385247468e-06,
+      "loss": 0.8905,
+      "step": 23112
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.796388304987579e-06,
+      "loss": 0.6028,
+      "step": 23113
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.796292757500478e-06,
+      "loss": 1.0904,
+      "step": 23114
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7961972100133766e-06,
+      "loss": 0.8922,
+      "step": 23115
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.796101662526276e-06,
+      "loss": 0.8498,
+      "step": 23116
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7960061150391747e-06,
+      "loss": 0.7861,
+      "step": 23117
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7959105675520737e-06,
+      "loss": 0.7835,
+      "step": 23118
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7958150200649723e-06,
+      "loss": 0.5772,
+      "step": 23119
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7957194725778714e-06,
+      "loss": 1.1111,
+      "step": 23120
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.79562392509077e-06,
+      "loss": 1.345,
+      "step": 23121
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.795528377603669e-06,
+      "loss": 0.8865,
+      "step": 23122
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7954328301165685e-06,
+      "loss": 0.5013,
+      "step": 23123
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.795337282629467e-06,
+      "loss": 0.5414,
+      "step": 23124
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.795241735142366e-06,
+      "loss": 0.6843,
+      "step": 23125
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7951461876552648e-06,
+      "loss": 1.0034,
+      "step": 23126
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.795050640168164e-06,
+      "loss": 0.8743,
+      "step": 23127
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7949550926810624e-06,
+      "loss": 0.7785,
+      "step": 23128
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7948595451939615e-06,
+      "loss": 0.7057,
+      "step": 23129
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.794763997706861e-06,
+      "loss": 1.2692,
+      "step": 23130
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7946684502197595e-06,
+      "loss": 0.9125,
+      "step": 23131
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7945729027326586e-06,
+      "loss": 0.7584,
+      "step": 23132
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.794477355245557e-06,
+      "loss": 1.039,
+      "step": 23133
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7943818077584562e-06,
+      "loss": 0.608,
+      "step": 23134
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.794286260271355e-06,
+      "loss": 0.7685,
+      "step": 23135
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.794190712784254e-06,
+      "loss": 0.5975,
+      "step": 23136
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7940951652971533e-06,
+      "loss": 0.8926,
+      "step": 23137
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.793999617810052e-06,
+      "loss": 0.7227,
+      "step": 23138
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.793904070322951e-06,
+      "loss": 0.9362,
+      "step": 23139
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7938085228358496e-06,
+      "loss": 0.5114,
+      "step": 23140
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7937129753487486e-06,
+      "loss": 1.0771,
+      "step": 23141
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7936174278616473e-06,
+      "loss": 0.9294,
+      "step": 23142
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7935218803745463e-06,
+      "loss": 1.1417,
+      "step": 23143
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.793426332887445e-06,
+      "loss": 1.0336,
+      "step": 23144
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7933307854003444e-06,
+      "loss": 1.1652,
+      "step": 23145
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7932352379132434e-06,
+      "loss": 0.8722,
+      "step": 23146
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.793139690426142e-06,
+      "loss": 0.6083,
+      "step": 23147
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.793044142939041e-06,
+      "loss": 0.5834,
+      "step": 23148
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7929485954519397e-06,
+      "loss": 0.7296,
+      "step": 23149
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7928530479648387e-06,
+      "loss": 0.7396,
+      "step": 23150
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7927575004777373e-06,
+      "loss": 0.985,
+      "step": 23151
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.792661952990637e-06,
+      "loss": 0.9252,
+      "step": 23152
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.792566405503536e-06,
+      "loss": 1.3524,
+      "step": 23153
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7924708580164345e-06,
+      "loss": 0.5939,
+      "step": 23154
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7923753105293335e-06,
+      "loss": 0.837,
+      "step": 23155
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.792279763042232e-06,
+      "loss": 0.7761,
+      "step": 23156
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.792184215555131e-06,
+      "loss": 0.6643,
+      "step": 23157
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7920886680680298e-06,
+      "loss": 0.6382,
+      "step": 23158
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7919931205809292e-06,
+      "loss": 0.8838,
+      "step": 23159
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.791897573093828e-06,
+      "loss": 0.8479,
+      "step": 23160
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.791802025606727e-06,
+      "loss": 1.0641,
+      "step": 23161
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7917064781196255e-06,
+      "loss": 0.6331,
+      "step": 23162
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7916109306325245e-06,
+      "loss": 0.5154,
+      "step": 23163
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7915153831454236e-06,
+      "loss": 0.802,
+      "step": 23164
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.791419835658322e-06,
+      "loss": 0.6843,
+      "step": 23165
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7913242881712216e-06,
+      "loss": 0.614,
+      "step": 23166
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7912287406841203e-06,
+      "loss": 0.7139,
+      "step": 23167
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7911331931970193e-06,
+      "loss": 0.8165,
+      "step": 23168
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.791037645709918e-06,
+      "loss": 0.9647,
+      "step": 23169
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.790942098222817e-06,
+      "loss": 0.6311,
+      "step": 23170
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7908465507357156e-06,
+      "loss": 0.6923,
+      "step": 23171
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7907510032486146e-06,
+      "loss": 0.5776,
+      "step": 23172
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.790655455761514e-06,
+      "loss": 0.3365,
+      "step": 23173
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7905599082744127e-06,
+      "loss": 0.8574,
+      "step": 23174
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7904643607873117e-06,
+      "loss": 0.7899,
+      "step": 23175
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7903688133002103e-06,
+      "loss": 0.6772,
+      "step": 23176
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7902732658131094e-06,
+      "loss": 0.9939,
+      "step": 23177
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.790177718326008e-06,
+      "loss": 1.1244,
+      "step": 23178
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.790082170838907e-06,
+      "loss": 0.8749,
+      "step": 23179
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7899866233518056e-06,
+      "loss": 0.7023,
+      "step": 23180
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789891075864705e-06,
+      "loss": 0.8118,
+      "step": 23181
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789795528377604e-06,
+      "loss": 0.7917,
+      "step": 23182
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7896999808905028e-06,
+      "loss": 0.7325,
+      "step": 23183
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789604433403402e-06,
+      "loss": 0.9382,
+      "step": 23184
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7895088859163004e-06,
+      "loss": 0.4977,
+      "step": 23185
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7894133384291995e-06,
+      "loss": 1.1422,
+      "step": 23186
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789317790942098e-06,
+      "loss": 0.6757,
+      "step": 23187
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7892222434549975e-06,
+      "loss": 0.7817,
+      "step": 23188
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7891266959678966e-06,
+      "loss": 0.7442,
+      "step": 23189
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.789031148480795e-06,
+      "loss": 0.8623,
+      "step": 23190
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7889356009936942e-06,
+      "loss": 0.3846,
+      "step": 23191
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.788840053506593e-06,
+      "loss": 0.5984,
+      "step": 23192
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.788744506019492e-06,
+      "loss": 0.3396,
+      "step": 23193
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7886489585323905e-06,
+      "loss": 0.836,
+      "step": 23194
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.78855341104529e-06,
+      "loss": 0.5624,
+      "step": 23195
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.788457863558189e-06,
+      "loss": 0.7255,
+      "step": 23196
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7883623160710876e-06,
+      "loss": 0.6779,
+      "step": 23197
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7882667685839866e-06,
+      "loss": 0.9977,
+      "step": 23198
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7881712210968853e-06,
+      "loss": 1.1153,
+      "step": 23199
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7880756736097843e-06,
+      "loss": 0.8988,
+      "step": 23200
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.787980126122683e-06,
+      "loss": 1.0099,
+      "step": 23201
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7878845786355824e-06,
+      "loss": 0.7831,
+      "step": 23202
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7877890311484814e-06,
+      "loss": 0.7457,
+      "step": 23203
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.78769348366138e-06,
+      "loss": 1.212,
+      "step": 23204
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.787597936174279e-06,
+      "loss": 0.9534,
+      "step": 23205
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7875023886871777e-06,
+      "loss": 0.6586,
+      "step": 23206
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7874068412000767e-06,
+      "loss": 0.6232,
+      "step": 23207
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7873112937129753e-06,
+      "loss": 0.8353,
+      "step": 23208
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.787215746225875e-06,
+      "loss": 0.7324,
+      "step": 23209
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7871201987387734e-06,
+      "loss": 0.8151,
+      "step": 23210
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7870246512516725e-06,
+      "loss": 0.7305,
+      "step": 23211
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786929103764571e-06,
+      "loss": 0.9084,
+      "step": 23212
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.78683355627747e-06,
+      "loss": 0.5474,
+      "step": 23213
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7867380087903687e-06,
+      "loss": 1.5168,
+      "step": 23214
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7866424613032678e-06,
+      "loss": 0.9518,
+      "step": 23215
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786546913816167e-06,
+      "loss": 0.786,
+      "step": 23216
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786451366329066e-06,
+      "loss": 0.8903,
+      "step": 23217
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786355818841965e-06,
+      "loss": 0.9967,
+      "step": 23218
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7862602713548635e-06,
+      "loss": 1.1303,
+      "step": 23219
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7861647238677625e-06,
+      "loss": 0.8414,
+      "step": 23220
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786069176380661e-06,
+      "loss": 0.5277,
+      "step": 23221
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.78597362889356e-06,
+      "loss": 0.644,
+      "step": 23222
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.785878081406459e-06,
+      "loss": 0.9481,
+      "step": 23223
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7857825339193583e-06,
+      "loss": 0.6837,
+      "step": 23224
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7856869864322573e-06,
+      "loss": 0.691,
+      "step": 23225
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.785591438945156e-06,
+      "loss": 1.0291,
+      "step": 23226
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.785495891458055e-06,
+      "loss": 0.7744,
+      "step": 23227
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7854003439709536e-06,
+      "loss": 0.9693,
+      "step": 23228
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7853047964838526e-06,
+      "loss": 1.0951,
+      "step": 23229
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7852092489967512e-06,
+      "loss": 0.8226,
+      "step": 23230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7851137015096507e-06,
+      "loss": 0.6787,
+      "step": 23231
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7850181540225497e-06,
+      "loss": 0.7428,
+      "step": 23232
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7849226065354483e-06,
+      "loss": 0.5174,
+      "step": 23233
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7848270590483474e-06,
+      "loss": 1.0401,
+      "step": 23234
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.784731511561246e-06,
+      "loss": 0.6396,
+      "step": 23235
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.784635964074145e-06,
+      "loss": 0.9664,
+      "step": 23236
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7845404165870436e-06,
+      "loss": 0.6658,
+      "step": 23237
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.784444869099943e-06,
+      "loss": 0.5101,
+      "step": 23238
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.784349321612842e-06,
+      "loss": 0.6534,
+      "step": 23239
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7842537741257408e-06,
+      "loss": 0.6603,
+      "step": 23240
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.78415822663864e-06,
+      "loss": 0.7684,
+      "step": 23241
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7840626791515384e-06,
+      "loss": 1.056,
+      "step": 23242
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7839671316644375e-06,
+      "loss": 0.8017,
+      "step": 23243
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.783871584177336e-06,
+      "loss": 0.5564,
+      "step": 23244
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7837760366902355e-06,
+      "loss": 1.0491,
+      "step": 23245
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7836804892031346e-06,
+      "loss": 0.7761,
+      "step": 23246
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.783584941716033e-06,
+      "loss": 0.8306,
+      "step": 23247
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7834893942289322e-06,
+      "loss": 0.5555,
+      "step": 23248
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.783393846741831e-06,
+      "loss": 0.6837,
+      "step": 23249
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.78329829925473e-06,
+      "loss": 0.6676,
+      "step": 23250
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7832027517676285e-06,
+      "loss": 0.5333,
+      "step": 23251
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7831072042805275e-06,
+      "loss": 0.5211,
+      "step": 23252
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7830116567934266e-06,
+      "loss": 0.9331,
+      "step": 23253
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7829161093063256e-06,
+      "loss": 0.6505,
+      "step": 23254
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7828205618192246e-06,
+      "loss": 0.4486,
+      "step": 23255
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7827250143321233e-06,
+      "loss": 0.8591,
+      "step": 23256
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7826294668450223e-06,
+      "loss": 0.5623,
+      "step": 23257
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.782533919357921e-06,
+      "loss": 0.6374,
+      "step": 23258
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.78243837187082e-06,
+      "loss": 0.8649,
+      "step": 23259
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.782342824383719e-06,
+      "loss": 0.9826,
+      "step": 23260
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.782247276896618e-06,
+      "loss": 0.8114,
+      "step": 23261
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7821517294095166e-06,
+      "loss": 0.733,
+      "step": 23262
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7820561819224157e-06,
+      "loss": 0.8804,
+      "step": 23263
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7819606344353143e-06,
+      "loss": 0.9557,
+      "step": 23264
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7818650869482133e-06,
+      "loss": 0.7712,
+      "step": 23265
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7817695394611124e-06,
+      "loss": 1.0112,
+      "step": 23266
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7816739919740114e-06,
+      "loss": 0.5708,
+      "step": 23267
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7815784444869105e-06,
+      "loss": 0.7848,
+      "step": 23268
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.781482896999809e-06,
+      "loss": 0.4903,
+      "step": 23269
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.781387349512708e-06,
+      "loss": 0.6433,
+      "step": 23270
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7812918020256067e-06,
+      "loss": 0.4942,
+      "step": 23271
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7811962545385058e-06,
+      "loss": 0.9221,
+      "step": 23272
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7811007070514044e-06,
+      "loss": 0.784,
+      "step": 23273
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.781005159564304e-06,
+      "loss": 0.6524,
+      "step": 23274
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.780909612077203e-06,
+      "loss": 1.0722,
+      "step": 23275
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7808140645901015e-06,
+      "loss": 0.9283,
+      "step": 23276
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7807185171030005e-06,
+      "loss": 0.4426,
+      "step": 23277
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.780622969615899e-06,
+      "loss": 0.7645,
+      "step": 23278
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.780527422128798e-06,
+      "loss": 0.6255,
+      "step": 23279
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.780431874641697e-06,
+      "loss": 0.5409,
+      "step": 23280
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7803363271545963e-06,
+      "loss": 0.8032,
+      "step": 23281
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7802407796674953e-06,
+      "loss": 0.6628,
+      "step": 23282
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.780145232180394e-06,
+      "loss": 1.1013,
+      "step": 23283
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.780049684693293e-06,
+      "loss": 0.8134,
+      "step": 23284
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7799541372061916e-06,
+      "loss": 1.2609,
+      "step": 23285
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7798585897190906e-06,
+      "loss": 0.4302,
+      "step": 23286
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7797630422319892e-06,
+      "loss": 0.8836,
+      "step": 23287
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7796674947448883e-06,
+      "loss": 0.6537,
+      "step": 23288
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7795719472577877e-06,
+      "loss": 1.0315,
+      "step": 23289
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7794763997706863e-06,
+      "loss": 1.2268,
+      "step": 23290
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7793808522835854e-06,
+      "loss": 1.0389,
+      "step": 23291
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.779285304796484e-06,
+      "loss": 0.7442,
+      "step": 23292
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.779189757309383e-06,
+      "loss": 0.7787,
+      "step": 23293
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7790942098222816e-06,
+      "loss": 0.7205,
+      "step": 23294
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7789986623351807e-06,
+      "loss": 0.7589,
+      "step": 23295
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.77890311484808e-06,
+      "loss": 0.8674,
+      "step": 23296
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7788075673609788e-06,
+      "loss": 0.6941,
+      "step": 23297
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.778712019873878e-06,
+      "loss": 0.7491,
+      "step": 23298
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7786164723867764e-06,
+      "loss": 1.032,
+      "step": 23299
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7785209248996755e-06,
+      "loss": 0.7994,
+      "step": 23300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.778425377412574e-06,
+      "loss": 1.0942,
+      "step": 23301
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.778329829925473e-06,
+      "loss": 1.2162,
+      "step": 23302
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.778234282438372e-06,
+      "loss": 0.6239,
+      "step": 23303
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.778138734951271e-06,
+      "loss": 0.696,
+      "step": 23304
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.77804318746417e-06,
+      "loss": 0.8204,
+      "step": 23305
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.777947639977069e-06,
+      "loss": 0.7348,
+      "step": 23306
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.777852092489968e-06,
+      "loss": 0.6903,
+      "step": 23307
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7777565450028665e-06,
+      "loss": 0.8641,
+      "step": 23308
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7776609975157655e-06,
+      "loss": 0.8174,
+      "step": 23309
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7775654500286646e-06,
+      "loss": 0.6348,
+      "step": 23310
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7774699025415636e-06,
+      "loss": 0.8523,
+      "step": 23311
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7773743550544622e-06,
+      "loss": 0.5109,
+      "step": 23312
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7772788075673613e-06,
+      "loss": 0.8103,
+      "step": 23313
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.77718326008026e-06,
+      "loss": 0.7908,
+      "step": 23314
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.777087712593159e-06,
+      "loss": 1.0476,
+      "step": 23315
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7769921651060575e-06,
+      "loss": 0.4966,
+      "step": 23316
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.776896617618957e-06,
+      "loss": 0.717,
+      "step": 23317
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.776801070131856e-06,
+      "loss": 0.7323,
+      "step": 23318
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7767055226447546e-06,
+      "loss": 0.9649,
+      "step": 23319
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7766099751576537e-06,
+      "loss": 1.4345,
+      "step": 23320
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7765144276705523e-06,
+      "loss": 0.6006,
+      "step": 23321
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7764188801834513e-06,
+      "loss": 0.8631,
+      "step": 23322
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.77632333269635e-06,
+      "loss": 0.763,
+      "step": 23323
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.776227785209249e-06,
+      "loss": 0.5751,
+      "step": 23324
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7761322377221485e-06,
+      "loss": 1.0587,
+      "step": 23325
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.776036690235047e-06,
+      "loss": 0.8649,
+      "step": 23326
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.775941142747946e-06,
+      "loss": 1.048,
+      "step": 23327
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7758455952608447e-06,
+      "loss": 1.0579,
+      "step": 23328
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7757500477737438e-06,
+      "loss": 1.0134,
+      "step": 23329
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7756545002866424e-06,
+      "loss": 0.7174,
+      "step": 23330
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7755589527995414e-06,
+      "loss": 0.7393,
+      "step": 23331
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.775463405312441e-06,
+      "loss": 0.7695,
+      "step": 23332
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7753678578253395e-06,
+      "loss": 0.9253,
+      "step": 23333
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7752723103382385e-06,
+      "loss": 0.6447,
+      "step": 23334
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.775176762851137e-06,
+      "loss": 1.0237,
+      "step": 23335
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.775081215364036e-06,
+      "loss": 0.8925,
+      "step": 23336
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.774985667876935e-06,
+      "loss": 1.026,
+      "step": 23337
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.774890120389834e-06,
+      "loss": 0.682,
+      "step": 23338
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7747945729027333e-06,
+      "loss": 0.2937,
+      "step": 23339
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.774699025415632e-06,
+      "loss": 1.0606,
+      "step": 23340
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.774603477928531e-06,
+      "loss": 0.8059,
+      "step": 23341
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7745079304414296e-06,
+      "loss": 0.5604,
+      "step": 23342
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7744123829543286e-06,
+      "loss": 0.555,
+      "step": 23343
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7743168354672272e-06,
+      "loss": 0.5907,
+      "step": 23344
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7742212879801263e-06,
+      "loss": 0.5639,
+      "step": 23345
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7741257404930257e-06,
+      "loss": 0.7337,
+      "step": 23346
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7740301930059243e-06,
+      "loss": 0.879,
+      "step": 23347
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7739346455188234e-06,
+      "loss": 0.6887,
+      "step": 23348
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.773839098031722e-06,
+      "loss": 0.6923,
+      "step": 23349
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.773743550544621e-06,
+      "loss": 0.6038,
+      "step": 23350
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7736480030575196e-06,
+      "loss": 0.7828,
+      "step": 23351
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7735524555704187e-06,
+      "loss": 0.8287,
+      "step": 23352
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7734569080833177e-06,
+      "loss": 0.7642,
+      "step": 23353
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7733613605962168e-06,
+      "loss": 0.977,
+      "step": 23354
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7732658131091154e-06,
+      "loss": 1.0155,
+      "step": 23355
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7731702656220144e-06,
+      "loss": 0.7442,
+      "step": 23356
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7730747181349135e-06,
+      "loss": 0.4181,
+      "step": 23357
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772979170647812e-06,
+      "loss": 0.6086,
+      "step": 23358
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772883623160711e-06,
+      "loss": 1.1434,
+      "step": 23359
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7727880756736097e-06,
+      "loss": 0.5053,
+      "step": 23360
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772692528186509e-06,
+      "loss": 0.9943,
+      "step": 23361
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772596980699408e-06,
+      "loss": 1.1414,
+      "step": 23362
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772501433212307e-06,
+      "loss": 0.8181,
+      "step": 23363
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7724058857252055e-06,
+      "loss": 0.8635,
+      "step": 23364
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7723103382381045e-06,
+      "loss": 1.0882,
+      "step": 23365
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772214790751003e-06,
+      "loss": 0.7622,
+      "step": 23366
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.772119243263902e-06,
+      "loss": 1.1347,
+      "step": 23367
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7720236957768016e-06,
+      "loss": 1.0533,
+      "step": 23368
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7719281482897002e-06,
+      "loss": 1.1668,
+      "step": 23369
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7718326008025993e-06,
+      "loss": 1.1188,
+      "step": 23370
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.771737053315498e-06,
+      "loss": 1.5391,
+      "step": 23371
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.771641505828397e-06,
+      "loss": 0.7328,
+      "step": 23372
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7715459583412955e-06,
+      "loss": 0.7884,
+      "step": 23373
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7714504108541946e-06,
+      "loss": 0.5474,
+      "step": 23374
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.771354863367094e-06,
+      "loss": 0.8948,
+      "step": 23375
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7712593158799926e-06,
+      "loss": 0.8583,
+      "step": 23376
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7711637683928917e-06,
+      "loss": 0.7649,
+      "step": 23377
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7710682209057903e-06,
+      "loss": 0.8789,
+      "step": 23378
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7709726734186893e-06,
+      "loss": 0.8645,
+      "step": 23379
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.770877125931588e-06,
+      "loss": 0.7941,
+      "step": 23380
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.770781578444487e-06,
+      "loss": 0.7073,
+      "step": 23381
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7706860309573865e-06,
+      "loss": 0.7743,
+      "step": 23382
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.770590483470285e-06,
+      "loss": 0.8125,
+      "step": 23383
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.770494935983184e-06,
+      "loss": 0.5763,
+      "step": 23384
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7703993884960827e-06,
+      "loss": 0.6443,
+      "step": 23385
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7703038410089818e-06,
+      "loss": 0.5943,
+      "step": 23386
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7702082935218804e-06,
+      "loss": 0.8117,
+      "step": 23387
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7701127460347794e-06,
+      "loss": 0.8381,
+      "step": 23388
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.770017198547679e-06,
+      "loss": 0.7819,
+      "step": 23389
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7699216510605775e-06,
+      "loss": 0.719,
+      "step": 23390
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7698261035734765e-06,
+      "loss": 0.8541,
+      "step": 23391
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769730556086375e-06,
+      "loss": 0.7766,
+      "step": 23392
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769635008599274e-06,
+      "loss": 1.2591,
+      "step": 23393
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769539461112173e-06,
+      "loss": 0.8386,
+      "step": 23394
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769443913625072e-06,
+      "loss": 0.9127,
+      "step": 23395
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7693483661379705e-06,
+      "loss": 0.946,
+      "step": 23396
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.76925281865087e-06,
+      "loss": 0.9239,
+      "step": 23397
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.769157271163769e-06,
+      "loss": 0.8502,
+      "step": 23398
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7690617236766676e-06,
+      "loss": 0.7909,
+      "step": 23399
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7689661761895666e-06,
+      "loss": 0.7068,
+      "step": 23400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7688706287024652e-06,
+      "loss": 0.7953,
+      "step": 23401
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7687750812153643e-06,
+      "loss": 0.797,
+      "step": 23402
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.768679533728263e-06,
+      "loss": 0.6725,
+      "step": 23403
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7685839862411623e-06,
+      "loss": 0.6733,
+      "step": 23404
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.768488438754061e-06,
+      "loss": 0.9471,
+      "step": 23405
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.76839289126696e-06,
+      "loss": 1.0867,
+      "step": 23406
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7682973437798586e-06,
+      "loss": 1.0029,
+      "step": 23407
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7682017962927576e-06,
+      "loss": 0.8697,
+      "step": 23408
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7681062488056567e-06,
+      "loss": 0.6145,
+      "step": 23409
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7680107013185553e-06,
+      "loss": 0.803,
+      "step": 23410
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7679151538314548e-06,
+      "loss": 1.0078,
+      "step": 23411
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7678196063443534e-06,
+      "loss": 1.3294,
+      "step": 23412
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7677240588572524e-06,
+      "loss": 1.4248,
+      "step": 23413
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.767628511370151e-06,
+      "loss": 1.1334,
+      "step": 23414
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.76753296388305e-06,
+      "loss": 0.699,
+      "step": 23415
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7674374163959487e-06,
+      "loss": 0.5202,
+      "step": 23416
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7673418689088477e-06,
+      "loss": 0.7354,
+      "step": 23417
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.767246321421747e-06,
+      "loss": 1.0619,
+      "step": 23418
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.767150773934646e-06,
+      "loss": 0.7012,
+      "step": 23419
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.767055226447545e-06,
+      "loss": 0.7216,
+      "step": 23420
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7669596789604435e-06,
+      "loss": 0.7885,
+      "step": 23421
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7668641314733425e-06,
+      "loss": 1.1392,
+      "step": 23422
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.766768583986241e-06,
+      "loss": 0.6471,
+      "step": 23423
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.76667303649914e-06,
+      "loss": 0.9575,
+      "step": 23424
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7665774890120396e-06,
+      "loss": 1.3442,
+      "step": 23425
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7664819415249382e-06,
+      "loss": 0.6542,
+      "step": 23426
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7663863940378373e-06,
+      "loss": 1.2347,
+      "step": 23427
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.766290846550736e-06,
+      "loss": 0.9334,
+      "step": 23428
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.766195299063635e-06,
+      "loss": 0.8779,
+      "step": 23429
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7660997515765335e-06,
+      "loss": 0.873,
+      "step": 23430
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7660042040894326e-06,
+      "loss": 1.1977,
+      "step": 23431
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.765908656602331e-06,
+      "loss": 0.515,
+      "step": 23432
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7658131091152306e-06,
+      "loss": 0.8512,
+      "step": 23433
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7657175616281297e-06,
+      "loss": 0.9744,
+      "step": 23434
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7656220141410283e-06,
+      "loss": 0.7556,
+      "step": 23435
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7655264666539273e-06,
+      "loss": 1.038,
+      "step": 23436
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.765430919166826e-06,
+      "loss": 0.7304,
+      "step": 23437
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.765335371679725e-06,
+      "loss": 1.0137,
+      "step": 23438
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7652398241926236e-06,
+      "loss": 1.3769,
+      "step": 23439
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.765144276705523e-06,
+      "loss": 0.8194,
+      "step": 23440
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.765048729218422e-06,
+      "loss": 0.9873,
+      "step": 23441
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7649531817313207e-06,
+      "loss": 0.7314,
+      "step": 23442
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7648576342442198e-06,
+      "loss": 0.932,
+      "step": 23443
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7647620867571184e-06,
+      "loss": 0.5202,
+      "step": 23444
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7646665392700174e-06,
+      "loss": 0.6475,
+      "step": 23445
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.764570991782916e-06,
+      "loss": 0.8203,
+      "step": 23446
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7644754442958155e-06,
+      "loss": 0.6925,
+      "step": 23447
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7643798968087145e-06,
+      "loss": 0.4667,
+      "step": 23448
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.764284349321613e-06,
+      "loss": 0.608,
+      "step": 23449
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.764188801834512e-06,
+      "loss": 0.6493,
+      "step": 23450
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.764093254347411e-06,
+      "loss": 0.7973,
+      "step": 23451
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.76399770686031e-06,
+      "loss": 1.085,
+      "step": 23452
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7639021593732085e-06,
+      "loss": 0.8994,
+      "step": 23453
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.763806611886108e-06,
+      "loss": 0.9025,
+      "step": 23454
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7637110643990065e-06,
+      "loss": 1.181,
+      "step": 23455
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7636155169119056e-06,
+      "loss": 0.8958,
+      "step": 23456
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.763519969424804e-06,
+      "loss": 0.8994,
+      "step": 23457
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7634244219377032e-06,
+      "loss": 0.8516,
+      "step": 23458
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7633288744506023e-06,
+      "loss": 1.0137,
+      "step": 23459
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.763233326963501e-06,
+      "loss": 0.6816,
+      "step": 23460
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7631377794764003e-06,
+      "loss": 0.6235,
+      "step": 23461
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.763042231989299e-06,
+      "loss": 0.7662,
+      "step": 23462
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.762946684502198e-06,
+      "loss": 0.7021,
+      "step": 23463
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7628511370150966e-06,
+      "loss": 0.9599,
+      "step": 23464
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7627555895279956e-06,
+      "loss": 0.5313,
+      "step": 23465
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7626600420408943e-06,
+      "loss": 0.9686,
+      "step": 23466
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7625644945537933e-06,
+      "loss": 0.9453,
+      "step": 23467
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.762468947066692e-06,
+      "loss": 1.1794,
+      "step": 23468
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7623733995795914e-06,
+      "loss": 1.0127,
+      "step": 23469
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7622778520924904e-06,
+      "loss": 0.7336,
+      "step": 23470
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.762182304605389e-06,
+      "loss": 1.0378,
+      "step": 23471
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.762086757118288e-06,
+      "loss": 1.1078,
+      "step": 23472
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7619912096311867e-06,
+      "loss": 0.5607,
+      "step": 23473
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7618956621440857e-06,
+      "loss": 0.4881,
+      "step": 23474
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7618001146569843e-06,
+      "loss": 0.4566,
+      "step": 23475
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.761704567169884e-06,
+      "loss": 0.8781,
+      "step": 23476
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.761609019682783e-06,
+      "loss": 0.9366,
+      "step": 23477
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7615134721956815e-06,
+      "loss": 0.9819,
+      "step": 23478
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7614179247085805e-06,
+      "loss": 0.6046,
+      "step": 23479
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.761322377221479e-06,
+      "loss": 1.0136,
+      "step": 23480
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.761226829734378e-06,
+      "loss": 0.6596,
+      "step": 23481
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7611312822472768e-06,
+      "loss": 1.009,
+      "step": 23482
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7610357347601762e-06,
+      "loss": 0.8304,
+      "step": 23483
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7609401872730753e-06,
+      "loss": 1.2796,
+      "step": 23484
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.760844639785974e-06,
+      "loss": 1.0421,
+      "step": 23485
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.760749092298873e-06,
+      "loss": 0.6154,
+      "step": 23486
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7606535448117715e-06,
+      "loss": 0.5408,
+      "step": 23487
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7605579973246706e-06,
+      "loss": 0.7462,
+      "step": 23488
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.760462449837569e-06,
+      "loss": 0.7783,
+      "step": 23489
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7603669023504686e-06,
+      "loss": 0.8478,
+      "step": 23490
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7602713548633677e-06,
+      "loss": 0.5126,
+      "step": 23491
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7601758073762663e-06,
+      "loss": 0.5809,
+      "step": 23492
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7600802598891653e-06,
+      "loss": 0.839,
+      "step": 23493
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759984712402064e-06,
+      "loss": 0.7359,
+      "step": 23494
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759889164914963e-06,
+      "loss": 0.5016,
+      "step": 23495
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7597936174278616e-06,
+      "loss": 0.8182,
+      "step": 23496
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759698069940761e-06,
+      "loss": 0.6372,
+      "step": 23497
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.75960252245366e-06,
+      "loss": 0.5835,
+      "step": 23498
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7595069749665587e-06,
+      "loss": 0.8721,
+      "step": 23499
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7594114274794578e-06,
+      "loss": 0.6455,
+      "step": 23500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7593158799923564e-06,
+      "loss": 0.732,
+      "step": 23501
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7592203325052554e-06,
+      "loss": 0.7099,
+      "step": 23502
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759124785018154e-06,
+      "loss": 1.2073,
+      "step": 23503
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.759029237531053e-06,
+      "loss": 1.0053,
+      "step": 23504
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.758933690043952e-06,
+      "loss": 0.6696,
+      "step": 23505
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.758838142556851e-06,
+      "loss": 0.8751,
+      "step": 23506
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7587425950697498e-06,
+      "loss": 0.9759,
+      "step": 23507
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.758647047582649e-06,
+      "loss": 0.4106,
+      "step": 23508
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7585515000955474e-06,
+      "loss": 0.9626,
+      "step": 23509
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7584559526084464e-06,
+      "loss": 0.6623,
+      "step": 23510
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7583604051213455e-06,
+      "loss": 0.9396,
+      "step": 23511
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7582648576342445e-06,
+      "loss": 0.9327,
+      "step": 23512
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7581693101471436e-06,
+      "loss": 0.793,
+      "step": 23513
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.758073762660042e-06,
+      "loss": 0.7295,
+      "step": 23514
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7579782151729412e-06,
+      "loss": 1.3658,
+      "step": 23515
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.75788266768584e-06,
+      "loss": 0.5473,
+      "step": 23516
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.757787120198739e-06,
+      "loss": 0.8482,
+      "step": 23517
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7576915727116375e-06,
+      "loss": 0.7368,
+      "step": 23518
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.757596025224537e-06,
+      "loss": 0.933,
+      "step": 23519
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.757500477737436e-06,
+      "loss": 0.7131,
+      "step": 23520
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7574049302503346e-06,
+      "loss": 0.7526,
+      "step": 23521
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7573093827632336e-06,
+      "loss": 0.8535,
+      "step": 23522
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7572138352761323e-06,
+      "loss": 0.7618,
+      "step": 23523
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7571182877890313e-06,
+      "loss": 0.6279,
+      "step": 23524
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.75702274030193e-06,
+      "loss": 0.8126,
+      "step": 23525
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7569271928148294e-06,
+      "loss": 0.9083,
+      "step": 23526
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7568316453277284e-06,
+      "loss": 0.8672,
+      "step": 23527
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.756736097840627e-06,
+      "loss": 0.9368,
+      "step": 23528
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.756640550353526e-06,
+      "loss": 0.6733,
+      "step": 23529
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7565450028664247e-06,
+      "loss": 0.947,
+      "step": 23530
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7564494553793237e-06,
+      "loss": 1.0513,
+      "step": 23531
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7563539078922223e-06,
+      "loss": 0.7835,
+      "step": 23532
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.756258360405122e-06,
+      "loss": 0.8095,
+      "step": 23533
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.756162812918021e-06,
+      "loss": 1.2442,
+      "step": 23534
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7560672654309194e-06,
+      "loss": 0.8376,
+      "step": 23535
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7559717179438185e-06,
+      "loss": 0.7927,
+      "step": 23536
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755876170456717e-06,
+      "loss": 1.0473,
+      "step": 23537
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755780622969616e-06,
+      "loss": 0.8453,
+      "step": 23538
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7556850754825148e-06,
+      "loss": 0.4984,
+      "step": 23539
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755589527995414e-06,
+      "loss": 0.601,
+      "step": 23540
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7554939805083133e-06,
+      "loss": 0.5723,
+      "step": 23541
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755398433021212e-06,
+      "loss": 0.6426,
+      "step": 23542
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.755302885534111e-06,
+      "loss": 0.6969,
+      "step": 23543
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7552073380470095e-06,
+      "loss": 0.6994,
+      "step": 23544
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7551117905599086e-06,
+      "loss": 1.021,
+      "step": 23545
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.755016243072807e-06,
+      "loss": 0.8579,
+      "step": 23546
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7549206955857062e-06,
+      "loss": 0.8938,
+      "step": 23547
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7548251480986053e-06,
+      "loss": 0.4637,
+      "step": 23548
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7547296006115043e-06,
+      "loss": 0.885,
+      "step": 23549
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7546340531244033e-06,
+      "loss": 0.8221,
+      "step": 23550
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.754538505637302e-06,
+      "loss": 0.8079,
+      "step": 23551
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.754442958150201e-06,
+      "loss": 0.892,
+      "step": 23552
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7543474106630996e-06,
+      "loss": 1.0232,
+      "step": 23553
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7542518631759986e-06,
+      "loss": 0.9443,
+      "step": 23554
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7541563156888977e-06,
+      "loss": 0.7922,
+      "step": 23555
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7540607682017967e-06,
+      "loss": 0.8395,
+      "step": 23556
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7539652207146953e-06,
+      "loss": 0.8201,
+      "step": 23557
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7538696732275944e-06,
+      "loss": 0.6068,
+      "step": 23558
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.753774125740493e-06,
+      "loss": 0.4981,
+      "step": 23559
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.753678578253392e-06,
+      "loss": 1.0342,
+      "step": 23560
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.753583030766291e-06,
+      "loss": 0.9196,
+      "step": 23561
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.75348748327919e-06,
+      "loss": 0.622,
+      "step": 23562
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.753391935792089e-06,
+      "loss": 0.4777,
+      "step": 23563
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7532963883049878e-06,
+      "loss": 0.5788,
+      "step": 23564
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.753200840817887e-06,
+      "loss": 0.9119,
+      "step": 23565
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7531052933307854e-06,
+      "loss": 0.8294,
+      "step": 23566
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7530097458436844e-06,
+      "loss": 0.8172,
+      "step": 23567
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.752914198356583e-06,
+      "loss": 0.8279,
+      "step": 23568
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7528186508694825e-06,
+      "loss": 0.7481,
+      "step": 23569
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7527231033823816e-06,
+      "loss": 1.0896,
+      "step": 23570
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.75262755589528e-06,
+      "loss": 0.938,
+      "step": 23571
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7525320084081792e-06,
+      "loss": 0.7531,
+      "step": 23572
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.752436460921078e-06,
+      "loss": 0.5781,
+      "step": 23573
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.752340913433977e-06,
+      "loss": 0.8861,
+      "step": 23574
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7522453659468755e-06,
+      "loss": 0.642,
+      "step": 23575
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7521498184597745e-06,
+      "loss": 0.6459,
+      "step": 23576
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.752054270972674e-06,
+      "loss": 0.925,
+      "step": 23577
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7519587234855726e-06,
+      "loss": 0.5195,
+      "step": 23578
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7518631759984716e-06,
+      "loss": 0.8604,
+      "step": 23579
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7517676285113703e-06,
+      "loss": 0.5902,
+      "step": 23580
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7516720810242693e-06,
+      "loss": 0.9597,
+      "step": 23581
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751576533537168e-06,
+      "loss": 1.0673,
+      "step": 23582
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751480986050067e-06,
+      "loss": 0.6638,
+      "step": 23583
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7513854385629664e-06,
+      "loss": 0.8044,
+      "step": 23584
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751289891075865e-06,
+      "loss": 0.9429,
+      "step": 23585
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751194343588764e-06,
+      "loss": 0.8798,
+      "step": 23586
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7510987961016627e-06,
+      "loss": 0.8848,
+      "step": 23587
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7510032486145617e-06,
+      "loss": 0.5612,
+      "step": 23588
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7509077011274603e-06,
+      "loss": 0.5928,
+      "step": 23589
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7508121536403594e-06,
+      "loss": 0.9706,
+      "step": 23590
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750716606153259e-06,
+      "loss": 0.7472,
+      "step": 23591
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7506210586661574e-06,
+      "loss": 0.5311,
+      "step": 23592
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7505255111790565e-06,
+      "loss": 0.7653,
+      "step": 23593
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750429963691955e-06,
+      "loss": 0.5729,
+      "step": 23594
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750334416204854e-06,
+      "loss": 0.6967,
+      "step": 23595
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7502388687177528e-06,
+      "loss": 0.5615,
+      "step": 23596
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750143321230652e-06,
+      "loss": 0.9353,
+      "step": 23597
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.750047773743551e-06,
+      "loss": 0.7927,
+      "step": 23598
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74995222625645e-06,
+      "loss": 0.7819,
+      "step": 23599
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7498566787693485e-06,
+      "loss": 0.6647,
+      "step": 23600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7497611312822475e-06,
+      "loss": 0.6701,
+      "step": 23601
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7496655837951466e-06,
+      "loss": 0.8091,
+      "step": 23602
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.749570036308045e-06,
+      "loss": 0.8237,
+      "step": 23603
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7494744888209442e-06,
+      "loss": 0.6058,
+      "step": 23604
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7493789413338433e-06,
+      "loss": 0.8192,
+      "step": 23605
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7492833938467423e-06,
+      "loss": 0.3767,
+      "step": 23606
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.749187846359641e-06,
+      "loss": 0.8139,
+      "step": 23607
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74909229887254e-06,
+      "loss": 0.93,
+      "step": 23608
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7489967513854386e-06,
+      "loss": 0.5277,
+      "step": 23609
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7489012038983376e-06,
+      "loss": 0.8898,
+      "step": 23610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7488056564112362e-06,
+      "loss": 0.53,
+      "step": 23611
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7487101089241353e-06,
+      "loss": 0.9065,
+      "step": 23612
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7486145614370347e-06,
+      "loss": 1.0476,
+      "step": 23613
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7485190139499333e-06,
+      "loss": 0.6819,
+      "step": 23614
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7484234664628324e-06,
+      "loss": 0.8124,
+      "step": 23615
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.748327918975731e-06,
+      "loss": 0.7704,
+      "step": 23616
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74823237148863e-06,
+      "loss": 0.5412,
+      "step": 23617
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7481368240015286e-06,
+      "loss": 1.1331,
+      "step": 23618
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7480412765144277e-06,
+      "loss": 0.6824,
+      "step": 23619
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.747945729027327e-06,
+      "loss": 0.379,
+      "step": 23620
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7478501815402258e-06,
+      "loss": 1.373,
+      "step": 23621
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.747754634053125e-06,
+      "loss": 0.6299,
+      "step": 23622
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7476590865660234e-06,
+      "loss": 1.2559,
+      "step": 23623
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7475635390789224e-06,
+      "loss": 1.0022,
+      "step": 23624
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.747467991591821e-06,
+      "loss": 0.6361,
+      "step": 23625
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74737244410472e-06,
+      "loss": 0.8183,
+      "step": 23626
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7472768966176196e-06,
+      "loss": 0.8255,
+      "step": 23627
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.747181349130518e-06,
+      "loss": 0.738,
+      "step": 23628
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7470858016434172e-06,
+      "loss": 0.926,
+      "step": 23629
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746990254156316e-06,
+      "loss": 0.9934,
+      "step": 23630
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746894706669215e-06,
+      "loss": 1.5221,
+      "step": 23631
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7467991591821135e-06,
+      "loss": 0.6519,
+      "step": 23632
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7467036116950125e-06,
+      "loss": 0.4057,
+      "step": 23633
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746608064207912e-06,
+      "loss": 0.8128,
+      "step": 23634
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7465125167208106e-06,
+      "loss": 1.2666,
+      "step": 23635
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7464169692337096e-06,
+      "loss": 1.1579,
+      "step": 23636
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7463214217466083e-06,
+      "loss": 0.6602,
+      "step": 23637
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7462258742595073e-06,
+      "loss": 0.3633,
+      "step": 23638
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746130326772406e-06,
+      "loss": 0.6346,
+      "step": 23639
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.746034779285305e-06,
+      "loss": 0.6256,
+      "step": 23640
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7459392317982044e-06,
+      "loss": 0.6607,
+      "step": 23641
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.745843684311103e-06,
+      "loss": 1.0425,
+      "step": 23642
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.745748136824002e-06,
+      "loss": 0.649,
+      "step": 23643
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7456525893369007e-06,
+      "loss": 1.4153,
+      "step": 23644
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7455570418497997e-06,
+      "loss": 0.8499,
+      "step": 23645
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7454614943626983e-06,
+      "loss": 1.0975,
+      "step": 23646
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7453659468755974e-06,
+      "loss": 0.5591,
+      "step": 23647
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.745270399388496e-06,
+      "loss": 0.9026,
+      "step": 23648
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7451748519013954e-06,
+      "loss": 0.99,
+      "step": 23649
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.745079304414294e-06,
+      "loss": 0.8491,
+      "step": 23650
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744983756927193e-06,
+      "loss": 0.6317,
+      "step": 23651
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744888209440092e-06,
+      "loss": 0.5251,
+      "step": 23652
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7447926619529908e-06,
+      "loss": 1.1744,
+      "step": 23653
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.74469711446589e-06,
+      "loss": 0.7559,
+      "step": 23654
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7446015669787884e-06,
+      "loss": 0.6223,
+      "step": 23655
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744506019491688e-06,
+      "loss": 0.5275,
+      "step": 23656
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7444104720045865e-06,
+      "loss": 0.7268,
+      "step": 23657
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7443149245174855e-06,
+      "loss": 0.6761,
+      "step": 23658
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744219377030384e-06,
+      "loss": 0.479,
+      "step": 23659
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744123829543283e-06,
+      "loss": 0.9571,
+      "step": 23660
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.744028282056182e-06,
+      "loss": 0.681,
+      "step": 23661
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.743932734569081e-06,
+      "loss": 0.9998,
+      "step": 23662
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7438371870819803e-06,
+      "loss": 0.8426,
+      "step": 23663
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.743741639594879e-06,
+      "loss": 1.0095,
+      "step": 23664
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.743646092107778e-06,
+      "loss": 0.8624,
+      "step": 23665
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7435505446206766e-06,
+      "loss": 0.6088,
+      "step": 23666
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7434549971335756e-06,
+      "loss": 1.1181,
+      "step": 23667
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7433594496464742e-06,
+      "loss": 0.8758,
+      "step": 23668
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7432639021593733e-06,
+      "loss": 0.8904,
+      "step": 23669
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7431683546722727e-06,
+      "loss": 0.5058,
+      "step": 23670
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7430728071851713e-06,
+      "loss": 1.1928,
+      "step": 23671
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7429772596980704e-06,
+      "loss": 0.7605,
+      "step": 23672
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742881712210969e-06,
+      "loss": 0.7466,
+      "step": 23673
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742786164723868e-06,
+      "loss": 0.6174,
+      "step": 23674
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7426906172367666e-06,
+      "loss": 0.397,
+      "step": 23675
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7425950697496657e-06,
+      "loss": 0.8252,
+      "step": 23676
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742499522262565e-06,
+      "loss": 1.313,
+      "step": 23677
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7424039747754638e-06,
+      "loss": 0.5158,
+      "step": 23678
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742308427288363e-06,
+      "loss": 0.9575,
+      "step": 23679
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7422128798012614e-06,
+      "loss": 0.7864,
+      "step": 23680
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7421173323141604e-06,
+      "loss": 1.042,
+      "step": 23681
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742021784827059e-06,
+      "loss": 0.5986,
+      "step": 23682
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741926237339958e-06,
+      "loss": 0.8102,
+      "step": 23683
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7418306898528567e-06,
+      "loss": 0.718,
+      "step": 23684
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741735142365756e-06,
+      "loss": 1.2025,
+      "step": 23685
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7416395948786552e-06,
+      "loss": 1.4404,
+      "step": 23686
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741544047391554e-06,
+      "loss": 0.7934,
+      "step": 23687
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741448499904453e-06,
+      "loss": 0.8966,
+      "step": 23688
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7413529524173515e-06,
+      "loss": 0.6633,
+      "step": 23689
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7412574049302505e-06,
+      "loss": 0.9924,
+      "step": 23690
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741161857443149e-06,
+      "loss": 1.1452,
+      "step": 23691
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7410663099560486e-06,
+      "loss": 0.8536,
+      "step": 23692
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7409707624689476e-06,
+      "loss": 0.9222,
+      "step": 23693
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7408752149818463e-06,
+      "loss": 0.9231,
+      "step": 23694
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7407796674947453e-06,
+      "loss": 0.6427,
+      "step": 23695
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.740684120007644e-06,
+      "loss": 0.8632,
+      "step": 23696
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.740588572520543e-06,
+      "loss": 1.2456,
+      "step": 23697
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7404930250334416e-06,
+      "loss": 0.6727,
+      "step": 23698
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.740397477546341e-06,
+      "loss": 0.8475,
+      "step": 23699
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7403019300592396e-06,
+      "loss": 0.805,
+      "step": 23700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7402063825721387e-06,
+      "loss": 0.5319,
+      "step": 23701
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7401108350850373e-06,
+      "loss": 0.8933,
+      "step": 23702
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7400152875979363e-06,
+      "loss": 1.5481,
+      "step": 23703
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7399197401108354e-06,
+      "loss": 0.6914,
+      "step": 23704
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.739824192623734e-06,
+      "loss": 0.7879,
+      "step": 23705
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7397286451366334e-06,
+      "loss": 0.8435,
+      "step": 23706
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.739633097649532e-06,
+      "loss": 0.9154,
+      "step": 23707
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.739537550162431e-06,
+      "loss": 0.7157,
+      "step": 23708
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7394420026753297e-06,
+      "loss": 0.495,
+      "step": 23709
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7393464551882288e-06,
+      "loss": 1.0419,
+      "step": 23710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7392509077011274e-06,
+      "loss": 0.6763,
+      "step": 23711
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7391553602140264e-06,
+      "loss": 0.7296,
+      "step": 23712
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.739059812726926e-06,
+      "loss": 1.084,
+      "step": 23713
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7389642652398245e-06,
+      "loss": 1.186,
+      "step": 23714
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7388687177527235e-06,
+      "loss": 0.891,
+      "step": 23715
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.738773170265622e-06,
+      "loss": 0.9899,
+      "step": 23716
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.738677622778521e-06,
+      "loss": 0.7713,
+      "step": 23717
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.73858207529142e-06,
+      "loss": 1.1378,
+      "step": 23718
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.738486527804319e-06,
+      "loss": 0.6946,
+      "step": 23719
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7383909803172174e-06,
+      "loss": 0.9671,
+      "step": 23720
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.738295432830117e-06,
+      "loss": 1.0878,
+      "step": 23721
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.738199885343016e-06,
+      "loss": 0.3314,
+      "step": 23722
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7381043378559146e-06,
+      "loss": 0.5056,
+      "step": 23723
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7380087903688136e-06,
+      "loss": 0.8613,
+      "step": 23724
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7379132428817122e-06,
+      "loss": 0.8033,
+      "step": 23725
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7378176953946113e-06,
+      "loss": 0.3155,
+      "step": 23726
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.73772214790751e-06,
+      "loss": 1.2077,
+      "step": 23727
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7376266004204093e-06,
+      "loss": 0.6579,
+      "step": 23728
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7375310529333084e-06,
+      "loss": 0.5001,
+      "step": 23729
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.737435505446207e-06,
+      "loss": 0.6957,
+      "step": 23730
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.737339957959106e-06,
+      "loss": 1.1774,
+      "step": 23731
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7372444104720046e-06,
+      "loss": 0.6204,
+      "step": 23732
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7371488629849037e-06,
+      "loss": 0.7969,
+      "step": 23733
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7370533154978023e-06,
+      "loss": 0.9483,
+      "step": 23734
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7369577680107018e-06,
+      "loss": 0.8355,
+      "step": 23735
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736862220523601e-06,
+      "loss": 0.8767,
+      "step": 23736
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7367666730364994e-06,
+      "loss": 0.624,
+      "step": 23737
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7366711255493984e-06,
+      "loss": 1.3451,
+      "step": 23738
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736575578062297e-06,
+      "loss": 0.6549,
+      "step": 23739
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736480030575196e-06,
+      "loss": 1.1182,
+      "step": 23740
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7363844830880947e-06,
+      "loss": 0.7124,
+      "step": 23741
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736288935600994e-06,
+      "loss": 0.7782,
+      "step": 23742
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7361933881138932e-06,
+      "loss": 1.0241,
+      "step": 23743
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736097840626792e-06,
+      "loss": 0.5613,
+      "step": 23744
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736002293139691e-06,
+      "loss": 0.6765,
+      "step": 23745
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7359067456525895e-06,
+      "loss": 0.678,
+      "step": 23746
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7358111981654885e-06,
+      "loss": 1.0403,
+      "step": 23747
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.735715650678387e-06,
+      "loss": 0.687,
+      "step": 23748
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7356201031912866e-06,
+      "loss": 0.4093,
+      "step": 23749
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7355245557041852e-06,
+      "loss": 0.938,
+      "step": 23750
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7354290082170843e-06,
+      "loss": 0.7424,
+      "step": 23751
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.735333460729983e-06,
+      "loss": 1.616,
+      "step": 23752
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.735237913242882e-06,
+      "loss": 0.7243,
+      "step": 23753
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.735142365755781e-06,
+      "loss": 0.8279,
+      "step": 23754
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7350468182686796e-06,
+      "loss": 1.0384,
+      "step": 23755
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7349512707815786e-06,
+      "loss": 0.7468,
+      "step": 23756
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7348557232944776e-06,
+      "loss": 0.9184,
+      "step": 23757
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7347601758073767e-06,
+      "loss": 0.769,
+      "step": 23758
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7346646283202753e-06,
+      "loss": 0.917,
+      "step": 23759
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7345690808331743e-06,
+      "loss": 0.805,
+      "step": 23760
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.734473533346073e-06,
+      "loss": 0.5362,
+      "step": 23761
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.734377985858972e-06,
+      "loss": 0.8347,
+      "step": 23762
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7342824383718706e-06,
+      "loss": 0.5827,
+      "step": 23763
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.73418689088477e-06,
+      "loss": 0.8879,
+      "step": 23764
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.734091343397669e-06,
+      "loss": 0.9347,
+      "step": 23765
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7339957959105677e-06,
+      "loss": 0.5697,
+      "step": 23766
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7339002484234668e-06,
+      "loss": 0.7018,
+      "step": 23767
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7338047009363654e-06,
+      "loss": 0.9781,
+      "step": 23768
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7337091534492644e-06,
+      "loss": 0.602,
+      "step": 23769
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.733613605962163e-06,
+      "loss": 0.6309,
+      "step": 23770
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7335180584750625e-06,
+      "loss": 0.8713,
+      "step": 23771
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7334225109879615e-06,
+      "loss": 0.7591,
+      "step": 23772
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.73332696350086e-06,
+      "loss": 0.6575,
+      "step": 23773
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.733231416013759e-06,
+      "loss": 0.6814,
+      "step": 23774
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.733135868526658e-06,
+      "loss": 0.9733,
+      "step": 23775
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.733040321039557e-06,
+      "loss": 0.653,
+      "step": 23776
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7329447735524554e-06,
+      "loss": 0.8215,
+      "step": 23777
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.732849226065355e-06,
+      "loss": 0.9649,
+      "step": 23778
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.732753678578254e-06,
+      "loss": 0.5861,
+      "step": 23779
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7326581310911526e-06,
+      "loss": 0.7531,
+      "step": 23780
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7325625836040516e-06,
+      "loss": 1.0123,
+      "step": 23781
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7324670361169502e-06,
+      "loss": 0.6945,
+      "step": 23782
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7323714886298493e-06,
+      "loss": 0.8421,
+      "step": 23783
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.732275941142748e-06,
+      "loss": 0.657,
+      "step": 23784
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7321803936556473e-06,
+      "loss": 0.8379,
+      "step": 23785
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7320848461685464e-06,
+      "loss": 1.1282,
+      "step": 23786
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731989298681445e-06,
+      "loss": 0.7987,
+      "step": 23787
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731893751194344e-06,
+      "loss": 0.7983,
+      "step": 23788
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7317982037072426e-06,
+      "loss": 1.0145,
+      "step": 23789
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7317026562201417e-06,
+      "loss": 0.7181,
+      "step": 23790
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7316071087330403e-06,
+      "loss": 1.4748,
+      "step": 23791
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7315115612459393e-06,
+      "loss": 0.8202,
+      "step": 23792
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731416013758839e-06,
+      "loss": 0.8759,
+      "step": 23793
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7313204662717374e-06,
+      "loss": 1.0255,
+      "step": 23794
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7312249187846364e-06,
+      "loss": 0.7065,
+      "step": 23795
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731129371297535e-06,
+      "loss": 0.8859,
+      "step": 23796
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731033823810434e-06,
+      "loss": 0.6741,
+      "step": 23797
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7309382763233327e-06,
+      "loss": 0.7938,
+      "step": 23798
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7308427288362318e-06,
+      "loss": 0.5875,
+      "step": 23799
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.730747181349131e-06,
+      "loss": 1.0126,
+      "step": 23800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.73065163386203e-06,
+      "loss": 0.8533,
+      "step": 23801
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7305560863749284e-06,
+      "loss": 0.7019,
+      "step": 23802
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7304605388878275e-06,
+      "loss": 0.7742,
+      "step": 23803
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.730364991400726e-06,
+      "loss": 1.2086,
+      "step": 23804
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.730269443913625e-06,
+      "loss": 0.8373,
+      "step": 23805
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.730173896426524e-06,
+      "loss": 0.6774,
+      "step": 23806
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7300783489394232e-06,
+      "loss": 0.9113,
+      "step": 23807
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7299828014523223e-06,
+      "loss": 0.8552,
+      "step": 23808
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.729887253965221e-06,
+      "loss": 0.8211,
+      "step": 23809
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.72979170647812e-06,
+      "loss": 1.064,
+      "step": 23810
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7296961589910185e-06,
+      "loss": 0.7546,
+      "step": 23811
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7296006115039176e-06,
+      "loss": 1.2046,
+      "step": 23812
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.729505064016816e-06,
+      "loss": 0.6683,
+      "step": 23813
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7294095165297156e-06,
+      "loss": 0.7065,
+      "step": 23814
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7293139690426147e-06,
+      "loss": 1.0174,
+      "step": 23815
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7292184215555133e-06,
+      "loss": 0.8886,
+      "step": 23816
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7291228740684123e-06,
+      "loss": 0.6311,
+      "step": 23817
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.729027326581311e-06,
+      "loss": 1.1447,
+      "step": 23818
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.72893177909421e-06,
+      "loss": 1.0324,
+      "step": 23819
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7288362316071086e-06,
+      "loss": 0.6929,
+      "step": 23820
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.728740684120008e-06,
+      "loss": 0.8437,
+      "step": 23821
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.728645136632907e-06,
+      "loss": 0.7558,
+      "step": 23822
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7285495891458057e-06,
+      "loss": 0.6956,
+      "step": 23823
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7284540416587048e-06,
+      "loss": 0.9152,
+      "step": 23824
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7283584941716034e-06,
+      "loss": 0.8474,
+      "step": 23825
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7282629466845024e-06,
+      "loss": 1.2543,
+      "step": 23826
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.728167399197401e-06,
+      "loss": 0.7127,
+      "step": 23827
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7280718517103e-06,
+      "loss": 1.0941,
+      "step": 23828
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7279763042231995e-06,
+      "loss": 0.9026,
+      "step": 23829
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727880756736098e-06,
+      "loss": 0.521,
+      "step": 23830
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727785209248997e-06,
+      "loss": 0.9804,
+      "step": 23831
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727689661761896e-06,
+      "loss": 0.4537,
+      "step": 23832
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727594114274795e-06,
+      "loss": 1.074,
+      "step": 23833
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7274985667876934e-06,
+      "loss": 0.8683,
+      "step": 23834
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7274030193005925e-06,
+      "loss": 1.1928,
+      "step": 23835
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727307471813492e-06,
+      "loss": 0.7852,
+      "step": 23836
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7272119243263906e-06,
+      "loss": 0.8877,
+      "step": 23837
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7271163768392896e-06,
+      "loss": 0.881,
+      "step": 23838
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7270208293521882e-06,
+      "loss": 0.9157,
+      "step": 23839
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7269252818650873e-06,
+      "loss": 0.6851,
+      "step": 23840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.726829734377986e-06,
+      "loss": 0.8184,
+      "step": 23841
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.726734186890885e-06,
+      "loss": 0.5191,
+      "step": 23842
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.726638639403784e-06,
+      "loss": 1.0913,
+      "step": 23843
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.726543091916683e-06,
+      "loss": 1.4403,
+      "step": 23844
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.726447544429582e-06,
+      "loss": 1.2054,
+      "step": 23845
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7263519969424806e-06,
+      "loss": 1.1254,
+      "step": 23846
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7262564494553797e-06,
+      "loss": 0.8618,
+      "step": 23847
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7261609019682783e-06,
+      "loss": 1.0683,
+      "step": 23848
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7260653544811773e-06,
+      "loss": 0.7218,
+      "step": 23849
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7259698069940764e-06,
+      "loss": 0.9593,
+      "step": 23850
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7258742595069754e-06,
+      "loss": 1.1993,
+      "step": 23851
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725778712019874e-06,
+      "loss": 1.0241,
+      "step": 23852
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725683164532773e-06,
+      "loss": 0.6788,
+      "step": 23853
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7255876170456717e-06,
+      "loss": 0.6534,
+      "step": 23854
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7254920695585707e-06,
+      "loss": 0.8672,
+      "step": 23855
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7253965220714698e-06,
+      "loss": 0.4949,
+      "step": 23856
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725300974584369e-06,
+      "loss": 0.7836,
+      "step": 23857
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725205427097268e-06,
+      "loss": 0.6601,
+      "step": 23858
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7251098796101664e-06,
+      "loss": 0.6753,
+      "step": 23859
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7250143321230655e-06,
+      "loss": 1.1368,
+      "step": 23860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724918784635964e-06,
+      "loss": 0.7977,
+      "step": 23861
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724823237148863e-06,
+      "loss": 0.4786,
+      "step": 23862
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7247276896617618e-06,
+      "loss": 0.4831,
+      "step": 23863
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724632142174661e-06,
+      "loss": 0.8419,
+      "step": 23864
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7245365946875603e-06,
+      "loss": 0.9684,
+      "step": 23865
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724441047200459e-06,
+      "loss": 0.6457,
+      "step": 23866
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724345499713358e-06,
+      "loss": 0.944,
+      "step": 23867
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7242499522262565e-06,
+      "loss": 0.6229,
+      "step": 23868
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7241544047391556e-06,
+      "loss": 0.3581,
+      "step": 23869
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.724058857252054e-06,
+      "loss": 0.6471,
+      "step": 23870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.723963309764953e-06,
+      "loss": 0.805,
+      "step": 23871
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7238677622778527e-06,
+      "loss": 1.2686,
+      "step": 23872
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7237722147907513e-06,
+      "loss": 0.8376,
+      "step": 23873
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7236766673036503e-06,
+      "loss": 0.713,
+      "step": 23874
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.723581119816549e-06,
+      "loss": 0.612,
+      "step": 23875
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.723485572329448e-06,
+      "loss": 0.6482,
+      "step": 23876
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7233900248423466e-06,
+      "loss": 0.9953,
+      "step": 23877
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7232944773552456e-06,
+      "loss": 0.4934,
+      "step": 23878
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.723198929868145e-06,
+      "loss": 0.7884,
+      "step": 23879
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7231033823810437e-06,
+      "loss": 0.9814,
+      "step": 23880
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7230078348939428e-06,
+      "loss": 0.9974,
+      "step": 23881
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7229122874068414e-06,
+      "loss": 1.9408,
+      "step": 23882
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7228167399197404e-06,
+      "loss": 1.099,
+      "step": 23883
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722721192432639e-06,
+      "loss": 1.062,
+      "step": 23884
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722625644945538e-06,
+      "loss": 0.657,
+      "step": 23885
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7225300974584375e-06,
+      "loss": 1.0163,
+      "step": 23886
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722434549971336e-06,
+      "loss": 0.9467,
+      "step": 23887
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722339002484235e-06,
+      "loss": 0.5276,
+      "step": 23888
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722243454997134e-06,
+      "loss": 0.9191,
+      "step": 23889
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.722147907510033e-06,
+      "loss": 0.6626,
+      "step": 23890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7220523600229314e-06,
+      "loss": 0.8273,
+      "step": 23891
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7219568125358305e-06,
+      "loss": 0.5216,
+      "step": 23892
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7218612650487295e-06,
+      "loss": 0.816,
+      "step": 23893
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7217657175616286e-06,
+      "loss": 0.8933,
+      "step": 23894
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721670170074527e-06,
+      "loss": 0.7858,
+      "step": 23895
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721574622587426e-06,
+      "loss": 0.957,
+      "step": 23896
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7214790751003253e-06,
+      "loss": 0.9213,
+      "step": 23897
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721383527613224e-06,
+      "loss": 0.458,
+      "step": 23898
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721287980126123e-06,
+      "loss": 0.9101,
+      "step": 23899
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7211924326390215e-06,
+      "loss": 0.7453,
+      "step": 23900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721096885151921e-06,
+      "loss": 1.1054,
+      "step": 23901
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7210013376648196e-06,
+      "loss": 0.7916,
+      "step": 23902
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7209057901777186e-06,
+      "loss": 0.738,
+      "step": 23903
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7208102426906172e-06,
+      "loss": 0.5097,
+      "step": 23904
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7207146952035163e-06,
+      "loss": 0.6941,
+      "step": 23905
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.720619147716415e-06,
+      "loss": 0.5294,
+      "step": 23906
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.720523600229314e-06,
+      "loss": 1.1342,
+      "step": 23907
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7204280527422134e-06,
+      "loss": 0.4936,
+      "step": 23908
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.720332505255112e-06,
+      "loss": 0.8779,
+      "step": 23909
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.720236957768011e-06,
+      "loss": 1.0597,
+      "step": 23910
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7201414102809097e-06,
+      "loss": 0.5575,
+      "step": 23911
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7200458627938087e-06,
+      "loss": 0.974,
+      "step": 23912
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7199503153067073e-06,
+      "loss": 1.0467,
+      "step": 23913
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7198547678196064e-06,
+      "loss": 1.0717,
+      "step": 23914
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.719759220332506e-06,
+      "loss": 0.5976,
+      "step": 23915
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7196636728454044e-06,
+      "loss": 0.9302,
+      "step": 23916
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7195681253583035e-06,
+      "loss": 0.7425,
+      "step": 23917
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.719472577871202e-06,
+      "loss": 1.1396,
+      "step": 23918
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.719377030384101e-06,
+      "loss": 0.6552,
+      "step": 23919
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7192814828969997e-06,
+      "loss": 0.996,
+      "step": 23920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7191859354098988e-06,
+      "loss": 0.9907,
+      "step": 23921
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7190903879227982e-06,
+      "loss": 1.322,
+      "step": 23922
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718994840435697e-06,
+      "loss": 0.6059,
+      "step": 23923
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718899292948596e-06,
+      "loss": 0.681,
+      "step": 23924
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7188037454614945e-06,
+      "loss": 0.5648,
+      "step": 23925
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7187081979743936e-06,
+      "loss": 0.8689,
+      "step": 23926
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718612650487292e-06,
+      "loss": 0.4707,
+      "step": 23927
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718517103000191e-06,
+      "loss": 0.7272,
+      "step": 23928
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7184215555130907e-06,
+      "loss": 0.451,
+      "step": 23929
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7183260080259893e-06,
+      "loss": 0.6707,
+      "step": 23930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7182304605388883e-06,
+      "loss": 0.608,
+      "step": 23931
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718134913051787e-06,
+      "loss": 1.2291,
+      "step": 23932
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.718039365564686e-06,
+      "loss": 0.7271,
+      "step": 23933
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7179438180775846e-06,
+      "loss": 0.7311,
+      "step": 23934
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7178482705904836e-06,
+      "loss": 0.4912,
+      "step": 23935
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7177527231033822e-06,
+      "loss": 0.7539,
+      "step": 23936
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7176571756162817e-06,
+      "loss": 0.638,
+      "step": 23937
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7175616281291807e-06,
+      "loss": 1.2149,
+      "step": 23938
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7174660806420794e-06,
+      "loss": 0.8795,
+      "step": 23939
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7173705331549784e-06,
+      "loss": 0.6502,
+      "step": 23940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.717274985667877e-06,
+      "loss": 0.9042,
+      "step": 23941
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.717179438180776e-06,
+      "loss": 1.1672,
+      "step": 23942
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7170838906936747e-06,
+      "loss": 0.7737,
+      "step": 23943
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716988343206574e-06,
+      "loss": 0.7845,
+      "step": 23944
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7168927957194727e-06,
+      "loss": 0.7541,
+      "step": 23945
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7167972482323718e-06,
+      "loss": 0.7909,
+      "step": 23946
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716701700745271e-06,
+      "loss": 0.7719,
+      "step": 23947
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7166061532581694e-06,
+      "loss": 0.7719,
+      "step": 23948
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7165106057710685e-06,
+      "loss": 1.1215,
+      "step": 23949
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716415058283967e-06,
+      "loss": 0.8394,
+      "step": 23950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7163195107968666e-06,
+      "loss": 0.8327,
+      "step": 23951
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716223963309765e-06,
+      "loss": 0.5279,
+      "step": 23952
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716128415822664e-06,
+      "loss": 0.947,
+      "step": 23953
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.716032868335563e-06,
+      "loss": 0.7724,
+      "step": 23954
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.715937320848462e-06,
+      "loss": 0.6895,
+      "step": 23955
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7158417733613605e-06,
+      "loss": 0.5299,
+      "step": 23956
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7157462258742595e-06,
+      "loss": 0.6662,
+      "step": 23957
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.715650678387159e-06,
+      "loss": 0.4865,
+      "step": 23958
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7155551309000576e-06,
+      "loss": 0.8796,
+      "step": 23959
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7154595834129566e-06,
+      "loss": 0.6477,
+      "step": 23960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7153640359258552e-06,
+      "loss": 0.7439,
+      "step": 23961
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7152684884387543e-06,
+      "loss": 0.6357,
+      "step": 23962
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.715172940951653e-06,
+      "loss": 0.8921,
+      "step": 23963
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.715077393464552e-06,
+      "loss": 0.769,
+      "step": 23964
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7149818459774514e-06,
+      "loss": 0.9898,
+      "step": 23965
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.71488629849035e-06,
+      "loss": 0.8298,
+      "step": 23966
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.714790751003249e-06,
+      "loss": 0.7036,
+      "step": 23967
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7146952035161477e-06,
+      "loss": 1.1423,
+      "step": 23968
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7145996560290467e-06,
+      "loss": 0.7718,
+      "step": 23969
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7145041085419453e-06,
+      "loss": 0.6259,
+      "step": 23970
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7144085610548444e-06,
+      "loss": 0.7844,
+      "step": 23971
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.714313013567743e-06,
+      "loss": 0.6584,
+      "step": 23972
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7142174660806424e-06,
+      "loss": 0.5925,
+      "step": 23973
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7141219185935415e-06,
+      "loss": 0.711,
+      "step": 23974
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.71402637110644e-06,
+      "loss": 0.3841,
+      "step": 23975
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.713930823619339e-06,
+      "loss": 0.6499,
+      "step": 23976
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7138352761322377e-06,
+      "loss": 0.7673,
+      "step": 23977
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7137397286451368e-06,
+      "loss": 1.0019,
+      "step": 23978
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7136441811580354e-06,
+      "loss": 0.5148,
+      "step": 23979
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.713548633670935e-06,
+      "loss": 1.0396,
+      "step": 23980
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.713453086183834e-06,
+      "loss": 0.5304,
+      "step": 23981
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7133575386967325e-06,
+      "loss": 0.6415,
+      "step": 23982
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7132619912096316e-06,
+      "loss": 0.6859,
+      "step": 23983
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.71316644372253e-06,
+      "loss": 0.7786,
+      "step": 23984
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.713070896235429e-06,
+      "loss": 0.7204,
+      "step": 23985
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.712975348748328e-06,
+      "loss": 0.5844,
+      "step": 23986
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7128798012612273e-06,
+      "loss": 0.5374,
+      "step": 23987
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7127842537741263e-06,
+      "loss": 0.7487,
+      "step": 23988
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.712688706287025e-06,
+      "loss": 1.1972,
+      "step": 23989
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.712593158799924e-06,
+      "loss": 0.6827,
+      "step": 23990
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7124976113128226e-06,
+      "loss": 0.7627,
+      "step": 23991
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7124020638257216e-06,
+      "loss": 0.9426,
+      "step": 23992
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7123065163386202e-06,
+      "loss": 0.7627,
+      "step": 23993
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7122109688515197e-06,
+      "loss": 0.5013,
+      "step": 23994
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7121154213644183e-06,
+      "loss": 0.3987,
+      "step": 23995
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7120198738773174e-06,
+      "loss": 0.9637,
+      "step": 23996
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711924326390216e-06,
+      "loss": 0.8389,
+      "step": 23997
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711828778903115e-06,
+      "loss": 0.9017,
+      "step": 23998
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711733231416014e-06,
+      "loss": 0.7849,
+      "step": 23999
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7116376839289127e-06,
+      "loss": 0.7791,
+      "step": 24000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711542136441812e-06,
+      "loss": 0.871,
+      "step": 24001
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7114465889547107e-06,
+      "loss": 1.0029,
+      "step": 24002
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7113510414676098e-06,
+      "loss": 1.023,
+      "step": 24003
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7112554939805084e-06,
+      "loss": 0.8683,
+      "step": 24004
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7111599464934074e-06,
+      "loss": 0.5611,
+      "step": 24005
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.711064399006306e-06,
+      "loss": 0.5357,
+      "step": 24006
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.710968851519205e-06,
+      "loss": 1.1088,
+      "step": 24007
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7108733040321037e-06,
+      "loss": 0.6632,
+      "step": 24008
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.710777756545003e-06,
+      "loss": 0.5283,
+      "step": 24009
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.710682209057902e-06,
+      "loss": 0.7443,
+      "step": 24010
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.710586661570801e-06,
+      "loss": 1.2113,
+      "step": 24011
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7104911140837e-06,
+      "loss": 0.5469,
+      "step": 24012
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7103955665965985e-06,
+      "loss": 0.8354,
+      "step": 24013
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7103000191094975e-06,
+      "loss": 0.517,
+      "step": 24014
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.710204471622396e-06,
+      "loss": 1.4205,
+      "step": 24015
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7101089241352956e-06,
+      "loss": 0.6637,
+      "step": 24016
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7100133766481946e-06,
+      "loss": 0.7282,
+      "step": 24017
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7099178291610932e-06,
+      "loss": 0.475,
+      "step": 24018
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7098222816739923e-06,
+      "loss": 0.6633,
+      "step": 24019
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.709726734186891e-06,
+      "loss": 0.8646,
+      "step": 24020
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.70963118669979e-06,
+      "loss": 0.8413,
+      "step": 24021
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7095356392126886e-06,
+      "loss": 0.7348,
+      "step": 24022
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.709440091725588e-06,
+      "loss": 0.8471,
+      "step": 24023
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.709344544238487e-06,
+      "loss": 0.7704,
+      "step": 24024
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7092489967513857e-06,
+      "loss": 0.8951,
+      "step": 24025
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7091534492642847e-06,
+      "loss": 0.6056,
+      "step": 24026
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7090579017771833e-06,
+      "loss": 0.7408,
+      "step": 24027
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7089623542900824e-06,
+      "loss": 1.1855,
+      "step": 24028
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.708866806802981e-06,
+      "loss": 0.7276,
+      "step": 24029
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7087712593158804e-06,
+      "loss": 0.772,
+      "step": 24030
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7086757118287795e-06,
+      "loss": 1.4096,
+      "step": 24031
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.708580164341678e-06,
+      "loss": 1.2807,
+      "step": 24032
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.708484616854577e-06,
+      "loss": 0.7856,
+      "step": 24033
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7083890693674757e-06,
+      "loss": 0.6437,
+      "step": 24034
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7082935218803748e-06,
+      "loss": 0.775,
+      "step": 24035
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7081979743932734e-06,
+      "loss": 0.9213,
+      "step": 24036
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.708102426906173e-06,
+      "loss": 0.5844,
+      "step": 24037
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.708006879419072e-06,
+      "loss": 0.676,
+      "step": 24038
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7079113319319705e-06,
+      "loss": 0.5084,
+      "step": 24039
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7078157844448696e-06,
+      "loss": 0.7441,
+      "step": 24040
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707720236957768e-06,
+      "loss": 0.8013,
+      "step": 24041
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707624689470667e-06,
+      "loss": 0.7521,
+      "step": 24042
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707529141983566e-06,
+      "loss": 0.8448,
+      "step": 24043
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707433594496465e-06,
+      "loss": 0.9833,
+      "step": 24044
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707338047009364e-06,
+      "loss": 1.0524,
+      "step": 24045
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707242499522263e-06,
+      "loss": 0.3979,
+      "step": 24046
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7071469520351616e-06,
+      "loss": 0.832,
+      "step": 24047
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7070514045480606e-06,
+      "loss": 0.6698,
+      "step": 24048
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7069558570609596e-06,
+      "loss": 0.4652,
+      "step": 24049
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7068603095738582e-06,
+      "loss": 1.1066,
+      "step": 24050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7067647620867573e-06,
+      "loss": 0.7155,
+      "step": 24051
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7066692145996563e-06,
+      "loss": 0.8604,
+      "step": 24052
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7065736671125554e-06,
+      "loss": 0.453,
+      "step": 24053
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.706478119625454e-06,
+      "loss": 0.9,
+      "step": 24054
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.706382572138353e-06,
+      "loss": 0.5398,
+      "step": 24055
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7062870246512516e-06,
+      "loss": 0.9233,
+      "step": 24056
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7061914771641507e-06,
+      "loss": 1.0247,
+      "step": 24057
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7060959296770493e-06,
+      "loss": 1.1706,
+      "step": 24058
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7060003821899487e-06,
+      "loss": 1.1061,
+      "step": 24059
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7059048347028478e-06,
+      "loss": 0.9002,
+      "step": 24060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7058092872157464e-06,
+      "loss": 0.8541,
+      "step": 24061
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7057137397286454e-06,
+      "loss": 0.8648,
+      "step": 24062
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.705618192241544e-06,
+      "loss": 0.6054,
+      "step": 24063
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.705522644754443e-06,
+      "loss": 1.2315,
+      "step": 24064
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7054270972673417e-06,
+      "loss": 0.291,
+      "step": 24065
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.705331549780241e-06,
+      "loss": 0.94,
+      "step": 24066
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.70523600229314e-06,
+      "loss": 0.8674,
+      "step": 24067
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.705140454806039e-06,
+      "loss": 0.7846,
+      "step": 24068
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.705044907318938e-06,
+      "loss": 0.7762,
+      "step": 24069
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7049493598318365e-06,
+      "loss": 0.7603,
+      "step": 24070
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7048538123447355e-06,
+      "loss": 0.8979,
+      "step": 24071
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.704758264857634e-06,
+      "loss": 0.7021,
+      "step": 24072
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7046627173705336e-06,
+      "loss": 0.7257,
+      "step": 24073
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7045671698834326e-06,
+      "loss": 1.1774,
+      "step": 24074
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7044716223963312e-06,
+      "loss": 0.5698,
+      "step": 24075
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7043760749092303e-06,
+      "loss": 0.9221,
+      "step": 24076
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.704280527422129e-06,
+      "loss": 0.67,
+      "step": 24077
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.704184979935028e-06,
+      "loss": 0.6877,
+      "step": 24078
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7040894324479266e-06,
+      "loss": 0.8699,
+      "step": 24079
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7039938849608256e-06,
+      "loss": 0.6483,
+      "step": 24080
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.703898337473725e-06,
+      "loss": 0.9601,
+      "step": 24081
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7038027899866237e-06,
+      "loss": 0.9014,
+      "step": 24082
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7037072424995227e-06,
+      "loss": 1.1826,
+      "step": 24083
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7036116950124213e-06,
+      "loss": 0.8499,
+      "step": 24084
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7035161475253204e-06,
+      "loss": 0.9858,
+      "step": 24085
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.703420600038219e-06,
+      "loss": 0.7113,
+      "step": 24086
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.703325052551118e-06,
+      "loss": 0.911,
+      "step": 24087
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7032295050640175e-06,
+      "loss": 0.49,
+      "step": 24088
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.703133957576916e-06,
+      "loss": 0.6402,
+      "step": 24089
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.703038410089815e-06,
+      "loss": 0.5921,
+      "step": 24090
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7029428626027137e-06,
+      "loss": 1.0204,
+      "step": 24091
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7028473151156128e-06,
+      "loss": 0.4253,
+      "step": 24092
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7027517676285114e-06,
+      "loss": 1.4484,
+      "step": 24093
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7026562201414104e-06,
+      "loss": 0.9704,
+      "step": 24094
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7025606726543095e-06,
+      "loss": 0.799,
+      "step": 24095
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7024651251672085e-06,
+      "loss": 0.7361,
+      "step": 24096
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.702369577680107e-06,
+      "loss": 0.7471,
+      "step": 24097
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.702274030193006e-06,
+      "loss": 0.5792,
+      "step": 24098
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7021784827059048e-06,
+      "loss": 0.8087,
+      "step": 24099
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.702082935218804e-06,
+      "loss": 0.9708,
+      "step": 24100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701987387731703e-06,
+      "loss": 0.8178,
+      "step": 24101
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701891840244602e-06,
+      "loss": 0.6514,
+      "step": 24102
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701796292757501e-06,
+      "loss": 1.1729,
+      "step": 24103
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7017007452703996e-06,
+      "loss": 0.6729,
+      "step": 24104
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7016051977832986e-06,
+      "loss": 0.5855,
+      "step": 24105
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701509650296197e-06,
+      "loss": 0.9671,
+      "step": 24106
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7014141028090962e-06,
+      "loss": 0.7406,
+      "step": 24107
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701318555321995e-06,
+      "loss": 1.4696,
+      "step": 24108
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7012230078348943e-06,
+      "loss": 0.4576,
+      "step": 24109
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7011274603477934e-06,
+      "loss": 0.9088,
+      "step": 24110
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701031912860692e-06,
+      "loss": 0.9613,
+      "step": 24111
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.700936365373591e-06,
+      "loss": 0.8921,
+      "step": 24112
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7008408178864896e-06,
+      "loss": 0.6722,
+      "step": 24113
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7007452703993887e-06,
+      "loss": 0.8903,
+      "step": 24114
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7006497229122873e-06,
+      "loss": 0.6011,
+      "step": 24115
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7005541754251863e-06,
+      "loss": 0.6548,
+      "step": 24116
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7004586279380858e-06,
+      "loss": 0.8434,
+      "step": 24117
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7003630804509844e-06,
+      "loss": 0.9068,
+      "step": 24118
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7002675329638834e-06,
+      "loss": 0.5604,
+      "step": 24119
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.700171985476782e-06,
+      "loss": 1.1509,
+      "step": 24120
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.700076437989681e-06,
+      "loss": 0.8032,
+      "step": 24121
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6999808905025797e-06,
+      "loss": 0.4593,
+      "step": 24122
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6998853430154787e-06,
+      "loss": 0.68,
+      "step": 24123
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699789795528378e-06,
+      "loss": 1.0084,
+      "step": 24124
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699694248041277e-06,
+      "loss": 0.9917,
+      "step": 24125
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699598700554176e-06,
+      "loss": 0.285,
+      "step": 24126
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6995031530670745e-06,
+      "loss": 1.1745,
+      "step": 24127
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6994076055799735e-06,
+      "loss": 1.1918,
+      "step": 24128
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699312058092872e-06,
+      "loss": 0.9122,
+      "step": 24129
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.699216510605771e-06,
+      "loss": 0.6321,
+      "step": 24130
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6991209631186706e-06,
+      "loss": 1.4019,
+      "step": 24131
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6990254156315692e-06,
+      "loss": 0.5892,
+      "step": 24132
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6989298681444683e-06,
+      "loss": 0.8877,
+      "step": 24133
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.698834320657367e-06,
+      "loss": 0.9802,
+      "step": 24134
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.698738773170266e-06,
+      "loss": 0.9366,
+      "step": 24135
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6986432256831646e-06,
+      "loss": 1.0859,
+      "step": 24136
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6985476781960636e-06,
+      "loss": 0.8691,
+      "step": 24137
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6984521307089626e-06,
+      "loss": 0.7354,
+      "step": 24138
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6983565832218617e-06,
+      "loss": 0.6094,
+      "step": 24139
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6982610357347607e-06,
+      "loss": 0.7144,
+      "step": 24140
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6981654882476593e-06,
+      "loss": 0.8874,
+      "step": 24141
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6980699407605584e-06,
+      "loss": 1.3759,
+      "step": 24142
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697974393273457e-06,
+      "loss": 0.9128,
+      "step": 24143
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697878845786356e-06,
+      "loss": 0.8607,
+      "step": 24144
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697783298299255e-06,
+      "loss": 0.502,
+      "step": 24145
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697687750812154e-06,
+      "loss": 1.0417,
+      "step": 24146
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6975922033250527e-06,
+      "loss": 0.6948,
+      "step": 24147
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6974966558379517e-06,
+      "loss": 0.7581,
+      "step": 24148
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6974011083508504e-06,
+      "loss": 0.8202,
+      "step": 24149
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6973055608637494e-06,
+      "loss": 1.0059,
+      "step": 24150
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6972100133766484e-06,
+      "loss": 1.0561,
+      "step": 24151
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697114465889547e-06,
+      "loss": 0.8742,
+      "step": 24152
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6970189184024465e-06,
+      "loss": 0.749,
+      "step": 24153
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.696923370915345e-06,
+      "loss": 0.9332,
+      "step": 24154
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.696827823428244e-06,
+      "loss": 0.9055,
+      "step": 24155
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6967322759411428e-06,
+      "loss": 0.434,
+      "step": 24156
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.696636728454042e-06,
+      "loss": 0.8672,
+      "step": 24157
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6965411809669404e-06,
+      "loss": 1.0595,
+      "step": 24158
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6964456334798395e-06,
+      "loss": 0.3547,
+      "step": 24159
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.696350085992739e-06,
+      "loss": 1.1095,
+      "step": 24160
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6962545385056376e-06,
+      "loss": 0.8982,
+      "step": 24161
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6961589910185366e-06,
+      "loss": 1.058,
+      "step": 24162
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.696063443531435e-06,
+      "loss": 0.6289,
+      "step": 24163
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6959678960443342e-06,
+      "loss": 0.7624,
+      "step": 24164
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.695872348557233e-06,
+      "loss": 0.7895,
+      "step": 24165
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.695776801070132e-06,
+      "loss": 0.6728,
+      "step": 24166
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6956812535830314e-06,
+      "loss": 0.7725,
+      "step": 24167
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.69558570609593e-06,
+      "loss": 0.666,
+      "step": 24168
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.695490158608829e-06,
+      "loss": 0.6884,
+      "step": 24169
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6953946111217276e-06,
+      "loss": 0.6535,
+      "step": 24170
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6952990636346267e-06,
+      "loss": 0.5534,
+      "step": 24171
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6952035161475253e-06,
+      "loss": 1.0326,
+      "step": 24172
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6951079686604243e-06,
+      "loss": 0.776,
+      "step": 24173
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6950124211733238e-06,
+      "loss": 1.0957,
+      "step": 24174
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6949168736862224e-06,
+      "loss": 0.9354,
+      "step": 24175
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6948213261991214e-06,
+      "loss": 1.1536,
+      "step": 24176
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.69472577871202e-06,
+      "loss": 1.0525,
+      "step": 24177
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.694630231224919e-06,
+      "loss": 0.911,
+      "step": 24178
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6945346837378177e-06,
+      "loss": 0.5899,
+      "step": 24179
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6944391362507167e-06,
+      "loss": 0.4916,
+      "step": 24180
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.694343588763616e-06,
+      "loss": 1.3516,
+      "step": 24181
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.694248041276515e-06,
+      "loss": 0.8608,
+      "step": 24182
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.694152493789414e-06,
+      "loss": 0.868,
+      "step": 24183
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6940569463023125e-06,
+      "loss": 0.9943,
+      "step": 24184
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6939613988152115e-06,
+      "loss": 0.9535,
+      "step": 24185
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.69386585132811e-06,
+      "loss": 1.2055,
+      "step": 24186
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.693770303841009e-06,
+      "loss": 0.7646,
+      "step": 24187
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6936747563539078e-06,
+      "loss": 0.596,
+      "step": 24188
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6935792088668072e-06,
+      "loss": 0.6126,
+      "step": 24189
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6934836613797063e-06,
+      "loss": 0.7275,
+      "step": 24190
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.693388113892605e-06,
+      "loss": 0.8171,
+      "step": 24191
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.693292566405504e-06,
+      "loss": 0.8931,
+      "step": 24192
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6931970189184026e-06,
+      "loss": 0.8726,
+      "step": 24193
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6931014714313016e-06,
+      "loss": 0.794,
+      "step": 24194
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6930059239442e-06,
+      "loss": 1.0617,
+      "step": 24195
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6929103764570997e-06,
+      "loss": 0.8372,
+      "step": 24196
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6928148289699983e-06,
+      "loss": 1.0162,
+      "step": 24197
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6927192814828973e-06,
+      "loss": 0.663,
+      "step": 24198
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.692623733995796e-06,
+      "loss": 0.58,
+      "step": 24199
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.692528186508695e-06,
+      "loss": 0.7414,
+      "step": 24200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6924326390215936e-06,
+      "loss": 0.5128,
+      "step": 24201
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6923370915344926e-06,
+      "loss": 0.8628,
+      "step": 24202
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.692241544047392e-06,
+      "loss": 0.6839,
+      "step": 24203
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6921459965602907e-06,
+      "loss": 0.8681,
+      "step": 24204
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6920504490731897e-06,
+      "loss": 0.934,
+      "step": 24205
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6919549015860884e-06,
+      "loss": 0.9103,
+      "step": 24206
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6918593540989874e-06,
+      "loss": 0.7816,
+      "step": 24207
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.691763806611886e-06,
+      "loss": 0.4751,
+      "step": 24208
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.691668259124785e-06,
+      "loss": 0.5846,
+      "step": 24209
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6915727116376845e-06,
+      "loss": 0.9697,
+      "step": 24210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.691477164150583e-06,
+      "loss": 0.8968,
+      "step": 24211
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.691381616663482e-06,
+      "loss": 1.2373,
+      "step": 24212
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6912860691763808e-06,
+      "loss": 0.9423,
+      "step": 24213
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.69119052168928e-06,
+      "loss": 0.9746,
+      "step": 24214
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6910949742021784e-06,
+      "loss": 0.7498,
+      "step": 24215
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6909994267150775e-06,
+      "loss": 0.5012,
+      "step": 24216
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690903879227977e-06,
+      "loss": 0.8667,
+      "step": 24217
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6908083317408756e-06,
+      "loss": 0.9431,
+      "step": 24218
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6907127842537746e-06,
+      "loss": 1.1861,
+      "step": 24219
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690617236766673e-06,
+      "loss": 0.9918,
+      "step": 24220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6905216892795722e-06,
+      "loss": 0.6721,
+      "step": 24221
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690426141792471e-06,
+      "loss": 0.5914,
+      "step": 24222
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.69033059430537e-06,
+      "loss": 0.8578,
+      "step": 24223
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6902350468182685e-06,
+      "loss": 0.8149,
+      "step": 24224
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690139499331168e-06,
+      "loss": 0.8619,
+      "step": 24225
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.690043951844067e-06,
+      "loss": 1.1655,
+      "step": 24226
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6899484043569656e-06,
+      "loss": 1.0058,
+      "step": 24227
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6898528568698647e-06,
+      "loss": 0.6518,
+      "step": 24228
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6897573093827633e-06,
+      "loss": 0.862,
+      "step": 24229
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6896617618956623e-06,
+      "loss": 0.4487,
+      "step": 24230
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.689566214408561e-06,
+      "loss": 1.1182,
+      "step": 24231
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6894706669214604e-06,
+      "loss": 0.914,
+      "step": 24232
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6893751194343594e-06,
+      "loss": 0.8349,
+      "step": 24233
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.689279571947258e-06,
+      "loss": 1.131,
+      "step": 24234
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.689184024460157e-06,
+      "loss": 0.7167,
+      "step": 24235
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6890884769730557e-06,
+      "loss": 0.4097,
+      "step": 24236
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6889929294859547e-06,
+      "loss": 0.6115,
+      "step": 24237
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6888973819988534e-06,
+      "loss": 1.1941,
+      "step": 24238
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.688801834511753e-06,
+      "loss": 0.8466,
+      "step": 24239
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6887062870246514e-06,
+      "loss": 1.0268,
+      "step": 24240
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6886107395375505e-06,
+      "loss": 0.5046,
+      "step": 24241
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6885151920504495e-06,
+      "loss": 0.9126,
+      "step": 24242
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.688419644563348e-06,
+      "loss": 0.5834,
+      "step": 24243
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.688324097076247e-06,
+      "loss": 0.9642,
+      "step": 24244
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6882285495891458e-06,
+      "loss": 0.8405,
+      "step": 24245
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6881330021020452e-06,
+      "loss": 0.8176,
+      "step": 24246
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.688037454614944e-06,
+      "loss": 0.7948,
+      "step": 24247
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.687941907127843e-06,
+      "loss": 1.0935,
+      "step": 24248
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6878463596407415e-06,
+      "loss": 0.5691,
+      "step": 24249
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6877508121536406e-06,
+      "loss": 0.643,
+      "step": 24250
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.687655264666539e-06,
+      "loss": 0.7556,
+      "step": 24251
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.687559717179438e-06,
+      "loss": 1.4566,
+      "step": 24252
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6874641696923377e-06,
+      "loss": 0.9005,
+      "step": 24253
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6873686222052363e-06,
+      "loss": 1.3305,
+      "step": 24254
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6872730747181353e-06,
+      "loss": 0.9645,
+      "step": 24255
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.687177527231034e-06,
+      "loss": 0.9005,
+      "step": 24256
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.687081979743933e-06,
+      "loss": 1.0475,
+      "step": 24257
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6869864322568316e-06,
+      "loss": 1.0348,
+      "step": 24258
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6868908847697306e-06,
+      "loss": 0.3516,
+      "step": 24259
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6867953372826292e-06,
+      "loss": 1.3019,
+      "step": 24260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6866997897955287e-06,
+      "loss": 0.6789,
+      "step": 24261
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6866042423084277e-06,
+      "loss": 1.2496,
+      "step": 24262
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6865086948213264e-06,
+      "loss": 0.7273,
+      "step": 24263
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6864131473342254e-06,
+      "loss": 1.1011,
+      "step": 24264
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.686317599847124e-06,
+      "loss": 1.0463,
+      "step": 24265
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.686222052360023e-06,
+      "loss": 0.5539,
+      "step": 24266
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6861265048729217e-06,
+      "loss": 0.6064,
+      "step": 24267
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.686030957385821e-06,
+      "loss": 0.9265,
+      "step": 24268
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.68593540989872e-06,
+      "loss": 1.0738,
+      "step": 24269
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6858398624116188e-06,
+      "loss": 1.2876,
+      "step": 24270
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.685744314924518e-06,
+      "loss": 1.6575,
+      "step": 24271
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6856487674374164e-06,
+      "loss": 0.7266,
+      "step": 24272
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6855532199503155e-06,
+      "loss": 1.1578,
+      "step": 24273
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.685457672463214e-06,
+      "loss": 0.9316,
+      "step": 24274
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6853621249761136e-06,
+      "loss": 0.925,
+      "step": 24275
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6852665774890126e-06,
+      "loss": 0.7374,
+      "step": 24276
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.685171030001911e-06,
+      "loss": 0.5124,
+      "step": 24277
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6850754825148102e-06,
+      "loss": 0.4406,
+      "step": 24278
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684979935027709e-06,
+      "loss": 1.0198,
+      "step": 24279
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684884387540608e-06,
+      "loss": 0.7272,
+      "step": 24280
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6847888400535065e-06,
+      "loss": 0.9529,
+      "step": 24281
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684693292566406e-06,
+      "loss": 0.4021,
+      "step": 24282
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684597745079305e-06,
+      "loss": 0.61,
+      "step": 24283
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6845021975922036e-06,
+      "loss": 0.9078,
+      "step": 24284
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6844066501051027e-06,
+      "loss": 0.7192,
+      "step": 24285
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6843111026180013e-06,
+      "loss": 0.7598,
+      "step": 24286
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6842155551309003e-06,
+      "loss": 0.8453,
+      "step": 24287
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.684120007643799e-06,
+      "loss": 0.9514,
+      "step": 24288
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6840244601566984e-06,
+      "loss": 0.6509,
+      "step": 24289
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.683928912669597e-06,
+      "loss": 0.9328,
+      "step": 24290
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.683833365182496e-06,
+      "loss": 1.0031,
+      "step": 24291
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6837378176953947e-06,
+      "loss": 0.652,
+      "step": 24292
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6836422702082937e-06,
+      "loss": 1.0923,
+      "step": 24293
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6835467227211927e-06,
+      "loss": 0.5681,
+      "step": 24294
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6834511752340914e-06,
+      "loss": 0.5837,
+      "step": 24295
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6833556277469904e-06,
+      "loss": 0.5088,
+      "step": 24296
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6832600802598894e-06,
+      "loss": 0.8551,
+      "step": 24297
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6831645327727885e-06,
+      "loss": 1.1657,
+      "step": 24298
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.683068985285687e-06,
+      "loss": 1.086,
+      "step": 24299
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682973437798586e-06,
+      "loss": 0.7944,
+      "step": 24300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6828778903114847e-06,
+      "loss": 0.556,
+      "step": 24301
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6827823428243838e-06,
+      "loss": 0.8601,
+      "step": 24302
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6826867953372824e-06,
+      "loss": 0.8152,
+      "step": 24303
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682591247850182e-06,
+      "loss": 0.9539,
+      "step": 24304
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682495700363081e-06,
+      "loss": 1.0361,
+      "step": 24305
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6824001528759795e-06,
+      "loss": 0.5665,
+      "step": 24306
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6823046053888785e-06,
+      "loss": 0.7316,
+      "step": 24307
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682209057901777e-06,
+      "loss": 0.5812,
+      "step": 24308
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682113510414676e-06,
+      "loss": 0.7601,
+      "step": 24309
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.682017962927575e-06,
+      "loss": 0.7484,
+      "step": 24310
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6819224154404743e-06,
+      "loss": 0.569,
+      "step": 24311
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6818268679533733e-06,
+      "loss": 0.7669,
+      "step": 24312
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.681731320466272e-06,
+      "loss": 0.5598,
+      "step": 24313
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.681635772979171e-06,
+      "loss": 0.4392,
+      "step": 24314
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6815402254920696e-06,
+      "loss": 0.9255,
+      "step": 24315
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6814446780049686e-06,
+      "loss": 1.2461,
+      "step": 24316
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6813491305178672e-06,
+      "loss": 0.7235,
+      "step": 24317
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6812535830307667e-06,
+      "loss": 0.6402,
+      "step": 24318
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6811580355436657e-06,
+      "loss": 0.9437,
+      "step": 24319
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6810624880565644e-06,
+      "loss": 0.6135,
+      "step": 24320
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6809669405694634e-06,
+      "loss": 0.6112,
+      "step": 24321
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.680871393082362e-06,
+      "loss": 0.9389,
+      "step": 24322
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.680775845595261e-06,
+      "loss": 0.7933,
+      "step": 24323
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6806802981081597e-06,
+      "loss": 1.048,
+      "step": 24324
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.680584750621059e-06,
+      "loss": 0.6702,
+      "step": 24325
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.680489203133958e-06,
+      "loss": 0.8639,
+      "step": 24326
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6803936556468568e-06,
+      "loss": 0.6559,
+      "step": 24327
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.680298108159756e-06,
+      "loss": 0.7099,
+      "step": 24328
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6802025606726544e-06,
+      "loss": 0.6098,
+      "step": 24329
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6801070131855535e-06,
+      "loss": 0.7146,
+      "step": 24330
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.680011465698452e-06,
+      "loss": 1.1229,
+      "step": 24331
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.679915918211351e-06,
+      "loss": 0.862,
+      "step": 24332
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6798203707242506e-06,
+      "loss": 0.4749,
+      "step": 24333
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.679724823237149e-06,
+      "loss": 0.6725,
+      "step": 24334
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6796292757500482e-06,
+      "loss": 0.9764,
+      "step": 24335
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.679533728262947e-06,
+      "loss": 1.1919,
+      "step": 24336
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.679438180775846e-06,
+      "loss": 0.5355,
+      "step": 24337
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6793426332887445e-06,
+      "loss": 0.8253,
+      "step": 24338
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6792470858016435e-06,
+      "loss": 0.7391,
+      "step": 24339
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6791515383145426e-06,
+      "loss": 0.8433,
+      "step": 24340
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6790559908274416e-06,
+      "loss": 0.6507,
+      "step": 24341
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6789604433403402e-06,
+      "loss": 0.7968,
+      "step": 24342
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6788648958532393e-06,
+      "loss": 1.0623,
+      "step": 24343
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6787693483661383e-06,
+      "loss": 0.4034,
+      "step": 24344
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.678673800879037e-06,
+      "loss": 0.6944,
+      "step": 24345
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.678578253391936e-06,
+      "loss": 1.0229,
+      "step": 24346
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.678482705904835e-06,
+      "loss": 0.8034,
+      "step": 24347
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.678387158417734e-06,
+      "loss": 1.024,
+      "step": 24348
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6782916109306327e-06,
+      "loss": 0.6426,
+      "step": 24349
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6781960634435317e-06,
+      "loss": 0.8849,
+      "step": 24350
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6781005159564303e-06,
+      "loss": 1.2268,
+      "step": 24351
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6780049684693294e-06,
+      "loss": 0.7288,
+      "step": 24352
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.677909420982228e-06,
+      "loss": 0.7464,
+      "step": 24353
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6778138734951274e-06,
+      "loss": 0.7032,
+      "step": 24354
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6777183260080265e-06,
+      "loss": 0.9495,
+      "step": 24355
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.677622778520925e-06,
+      "loss": 0.9191,
+      "step": 24356
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.677527231033824e-06,
+      "loss": 1.0002,
+      "step": 24357
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6774316835467227e-06,
+      "loss": 0.9885,
+      "step": 24358
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6773361360596218e-06,
+      "loss": 0.8676,
+      "step": 24359
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6772405885725204e-06,
+      "loss": 0.7964,
+      "step": 24360
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.67714504108542e-06,
+      "loss": 0.7018,
+      "step": 24361
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.677049493598319e-06,
+      "loss": 0.4848,
+      "step": 24362
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6769539461112175e-06,
+      "loss": 0.4766,
+      "step": 24363
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6768583986241165e-06,
+      "loss": 0.8347,
+      "step": 24364
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676762851137015e-06,
+      "loss": 0.6397,
+      "step": 24365
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676667303649914e-06,
+      "loss": 1.0991,
+      "step": 24366
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676571756162813e-06,
+      "loss": 1.0928,
+      "step": 24367
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676476208675712e-06,
+      "loss": 0.5896,
+      "step": 24368
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6763806611886113e-06,
+      "loss": 0.8065,
+      "step": 24369
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.67628511370151e-06,
+      "loss": 1.0246,
+      "step": 24370
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676189566214409e-06,
+      "loss": 0.8239,
+      "step": 24371
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6760940187273076e-06,
+      "loss": 0.6743,
+      "step": 24372
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6759984712402066e-06,
+      "loss": 0.4643,
+      "step": 24373
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6759029237531052e-06,
+      "loss": 0.9031,
+      "step": 24374
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6758073762660043e-06,
+      "loss": 0.6479,
+      "step": 24375
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6757118287789037e-06,
+      "loss": 0.5352,
+      "step": 24376
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6756162812918024e-06,
+      "loss": 0.6682,
+      "step": 24377
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6755207338047014e-06,
+      "loss": 0.6897,
+      "step": 24378
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6754251863176e-06,
+      "loss": 0.6223,
+      "step": 24379
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.675329638830499e-06,
+      "loss": 0.6755,
+      "step": 24380
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6752340913433977e-06,
+      "loss": 0.5655,
+      "step": 24381
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6751385438562967e-06,
+      "loss": 0.9605,
+      "step": 24382
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.675042996369196e-06,
+      "loss": 0.8186,
+      "step": 24383
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6749474488820948e-06,
+      "loss": 0.9153,
+      "step": 24384
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674851901394994e-06,
+      "loss": 0.796,
+      "step": 24385
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6747563539078924e-06,
+      "loss": 0.9505,
+      "step": 24386
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6746608064207915e-06,
+      "loss": 0.9026,
+      "step": 24387
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.67456525893369e-06,
+      "loss": 0.6331,
+      "step": 24388
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674469711446589e-06,
+      "loss": 0.5461,
+      "step": 24389
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674374163959488e-06,
+      "loss": 1.1392,
+      "step": 24390
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674278616472387e-06,
+      "loss": 0.926,
+      "step": 24391
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674183068985286e-06,
+      "loss": 0.6772,
+      "step": 24392
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.674087521498185e-06,
+      "loss": 0.7811,
+      "step": 24393
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6739919740110835e-06,
+      "loss": 0.7783,
+      "step": 24394
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6738964265239825e-06,
+      "loss": 0.5626,
+      "step": 24395
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6738008790368815e-06,
+      "loss": 1.1701,
+      "step": 24396
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6737053315497806e-06,
+      "loss": 1.011,
+      "step": 24397
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6736097840626796e-06,
+      "loss": 0.5223,
+      "step": 24398
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6735142365755782e-06,
+      "loss": 0.9515,
+      "step": 24399
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6734186890884773e-06,
+      "loss": 1.0227,
+      "step": 24400
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.673323141601376e-06,
+      "loss": 0.8349,
+      "step": 24401
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.673227594114275e-06,
+      "loss": 0.7215,
+      "step": 24402
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6731320466271735e-06,
+      "loss": 0.5903,
+      "step": 24403
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6730364991400726e-06,
+      "loss": 0.5603,
+      "step": 24404
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.672940951652972e-06,
+      "loss": 1.3476,
+      "step": 24405
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6728454041658707e-06,
+      "loss": 0.6641,
+      "step": 24406
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6727498566787697e-06,
+      "loss": 0.876,
+      "step": 24407
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6726543091916683e-06,
+      "loss": 0.616,
+      "step": 24408
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6725587617045674e-06,
+      "loss": 0.3253,
+      "step": 24409
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.672463214217466e-06,
+      "loss": 0.4644,
+      "step": 24410
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.672367666730365e-06,
+      "loss": 1.1682,
+      "step": 24411
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6722721192432645e-06,
+      "loss": 0.7817,
+      "step": 24412
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.672176571756163e-06,
+      "loss": 0.7304,
+      "step": 24413
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.672081024269062e-06,
+      "loss": 0.4837,
+      "step": 24414
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6719854767819607e-06,
+      "loss": 0.6695,
+      "step": 24415
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6718899292948598e-06,
+      "loss": 0.8345,
+      "step": 24416
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6717943818077584e-06,
+      "loss": 0.7091,
+      "step": 24417
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6716988343206574e-06,
+      "loss": 1.1671,
+      "step": 24418
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671603286833557e-06,
+      "loss": 1.0113,
+      "step": 24419
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6715077393464555e-06,
+      "loss": 0.6835,
+      "step": 24420
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6714121918593545e-06,
+      "loss": 0.594,
+      "step": 24421
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671316644372253e-06,
+      "loss": 0.9322,
+      "step": 24422
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671221096885152e-06,
+      "loss": 0.7963,
+      "step": 24423
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671125549398051e-06,
+      "loss": 1.0344,
+      "step": 24424
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.67103000191095e-06,
+      "loss": 0.5913,
+      "step": 24425
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6709344544238493e-06,
+      "loss": 0.9321,
+      "step": 24426
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.670838906936748e-06,
+      "loss": 0.7021,
+      "step": 24427
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.670743359449647e-06,
+      "loss": 0.615,
+      "step": 24428
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6706478119625456e-06,
+      "loss": 0.3508,
+      "step": 24429
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6705522644754446e-06,
+      "loss": 0.6883,
+      "step": 24430
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6704567169883432e-06,
+      "loss": 0.7652,
+      "step": 24431
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6703611695012423e-06,
+      "loss": 0.5008,
+      "step": 24432
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6702656220141413e-06,
+      "loss": 0.7348,
+      "step": 24433
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6701700745270404e-06,
+      "loss": 0.6252,
+      "step": 24434
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6700745270399394e-06,
+      "loss": 0.8348,
+      "step": 24435
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669978979552838e-06,
+      "loss": 0.6161,
+      "step": 24436
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669883432065737e-06,
+      "loss": 0.8893,
+      "step": 24437
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6697878845786357e-06,
+      "loss": 0.687,
+      "step": 24438
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6696923370915347e-06,
+      "loss": 0.6243,
+      "step": 24439
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6695967896044333e-06,
+      "loss": 0.6418,
+      "step": 24440
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6695012421173328e-06,
+      "loss": 0.9245,
+      "step": 24441
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6694056946302314e-06,
+      "loss": 0.9551,
+      "step": 24442
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6693101471431304e-06,
+      "loss": 0.8518,
+      "step": 24443
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669214599656029e-06,
+      "loss": 1.0891,
+      "step": 24444
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669119052168928e-06,
+      "loss": 0.9294,
+      "step": 24445
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.669023504681827e-06,
+      "loss": 0.8929,
+      "step": 24446
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6689279571947257e-06,
+      "loss": 0.5183,
+      "step": 24447
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.668832409707625e-06,
+      "loss": 0.6608,
+      "step": 24448
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.668736862220524e-06,
+      "loss": 1.0123,
+      "step": 24449
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.668641314733423e-06,
+      "loss": 0.7805,
+      "step": 24450
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6685457672463215e-06,
+      "loss": 0.6629,
+      "step": 24451
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6684502197592205e-06,
+      "loss": 0.8957,
+      "step": 24452
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.668354672272119e-06,
+      "loss": 1.012,
+      "step": 24453
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.668259124785018e-06,
+      "loss": 0.7521,
+      "step": 24454
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6681635772979176e-06,
+      "loss": 0.8473,
+      "step": 24455
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6680680298108162e-06,
+      "loss": 0.9395,
+      "step": 24456
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6679724823237153e-06,
+      "loss": 0.6756,
+      "step": 24457
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.667876934836614e-06,
+      "loss": 0.6571,
+      "step": 24458
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.667781387349513e-06,
+      "loss": 0.837,
+      "step": 24459
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6676858398624115e-06,
+      "loss": 0.943,
+      "step": 24460
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6675902923753106e-06,
+      "loss": 0.9871,
+      "step": 24461
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.66749474488821e-06,
+      "loss": 0.4261,
+      "step": 24462
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6673991974011087e-06,
+      "loss": 1.0225,
+      "step": 24463
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6673036499140077e-06,
+      "loss": 0.6471,
+      "step": 24464
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6672081024269063e-06,
+      "loss": 0.7433,
+      "step": 24465
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6671125549398054e-06,
+      "loss": 0.711,
+      "step": 24466
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.667017007452704e-06,
+      "loss": 0.5467,
+      "step": 24467
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.666921459965603e-06,
+      "loss": 0.6379,
+      "step": 24468
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6668259124785025e-06,
+      "loss": 0.9894,
+      "step": 24469
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.666730364991401e-06,
+      "loss": 1.1421,
+      "step": 24470
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6666348175043e-06,
+      "loss": 1.1797,
+      "step": 24471
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6665392700171987e-06,
+      "loss": 0.6375,
+      "step": 24472
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6664437225300978e-06,
+      "loss": 0.5122,
+      "step": 24473
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6663481750429964e-06,
+      "loss": 0.6816,
+      "step": 24474
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6662526275558954e-06,
+      "loss": 0.4367,
+      "step": 24475
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.666157080068794e-06,
+      "loss": 1.2677,
+      "step": 24476
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6660615325816935e-06,
+      "loss": 0.6152,
+      "step": 24477
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6659659850945925e-06,
+      "loss": 0.5914,
+      "step": 24478
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665870437607491e-06,
+      "loss": 0.7953,
+      "step": 24479
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.66577489012039e-06,
+      "loss": 0.7287,
+      "step": 24480
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665679342633289e-06,
+      "loss": 0.8177,
+      "step": 24481
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665583795146188e-06,
+      "loss": 0.935,
+      "step": 24482
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6654882476590865e-06,
+      "loss": 0.7863,
+      "step": 24483
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665392700171986e-06,
+      "loss": 0.6352,
+      "step": 24484
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665297152684885e-06,
+      "loss": 0.6142,
+      "step": 24485
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6652016051977836e-06,
+      "loss": 0.4871,
+      "step": 24486
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6651060577106826e-06,
+      "loss": 1.2516,
+      "step": 24487
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6650105102235812e-06,
+      "loss": 0.7217,
+      "step": 24488
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6649149627364803e-06,
+      "loss": 0.9395,
+      "step": 24489
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.664819415249379e-06,
+      "loss": 0.9888,
+      "step": 24490
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6647238677622784e-06,
+      "loss": 0.7668,
+      "step": 24491
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.664628320275177e-06,
+      "loss": 0.7045,
+      "step": 24492
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.664532772788076e-06,
+      "loss": 0.6734,
+      "step": 24493
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6644372253009746e-06,
+      "loss": 1.3321,
+      "step": 24494
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6643416778138737e-06,
+      "loss": 0.6389,
+      "step": 24495
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6642461303267723e-06,
+      "loss": 0.9382,
+      "step": 24496
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6641505828396713e-06,
+      "loss": 0.9914,
+      "step": 24497
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6640550353525708e-06,
+      "loss": 1.0355,
+      "step": 24498
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6639594878654694e-06,
+      "loss": 0.9149,
+      "step": 24499
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6638639403783684e-06,
+      "loss": 1.1273,
+      "step": 24500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.663768392891267e-06,
+      "loss": 1.1152,
+      "step": 24501
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.663672845404166e-06,
+      "loss": 0.5884,
+      "step": 24502
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6635772979170647e-06,
+      "loss": 0.8501,
+      "step": 24503
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6634817504299637e-06,
+      "loss": 0.7273,
+      "step": 24504
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.663386202942863e-06,
+      "loss": 0.5381,
+      "step": 24505
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.663290655455762e-06,
+      "loss": 0.8199,
+      "step": 24506
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.663195107968661e-06,
+      "loss": 0.8835,
+      "step": 24507
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6630995604815595e-06,
+      "loss": 0.8921,
+      "step": 24508
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6630040129944585e-06,
+      "loss": 0.9384,
+      "step": 24509
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.662908465507357e-06,
+      "loss": 0.4143,
+      "step": 24510
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.662812918020256e-06,
+      "loss": 1.05,
+      "step": 24511
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6627173705331548e-06,
+      "loss": 0.8107,
+      "step": 24512
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6626218230460542e-06,
+      "loss": 0.5185,
+      "step": 24513
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6625262755589533e-06,
+      "loss": 1.0386,
+      "step": 24514
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.662430728071852e-06,
+      "loss": 0.8579,
+      "step": 24515
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.662335180584751e-06,
+      "loss": 1.4724,
+      "step": 24516
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6622396330976495e-06,
+      "loss": 0.8574,
+      "step": 24517
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6621440856105486e-06,
+      "loss": 0.5821,
+      "step": 24518
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.662048538123447e-06,
+      "loss": 0.9548,
+      "step": 24519
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6619529906363467e-06,
+      "loss": 1.0898,
+      "step": 24520
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6618574431492457e-06,
+      "loss": 0.7028,
+      "step": 24521
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6617618956621443e-06,
+      "loss": 0.8189,
+      "step": 24522
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6616663481750434e-06,
+      "loss": 0.6951,
+      "step": 24523
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661570800687942e-06,
+      "loss": 0.7283,
+      "step": 24524
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661475253200841e-06,
+      "loss": 0.921,
+      "step": 24525
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6613797057137396e-06,
+      "loss": 0.6703,
+      "step": 24526
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661284158226639e-06,
+      "loss": 1.2496,
+      "step": 24527
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661188610739538e-06,
+      "loss": 0.9381,
+      "step": 24528
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6610930632524367e-06,
+      "loss": 0.4902,
+      "step": 24529
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6609975157653358e-06,
+      "loss": 0.9397,
+      "step": 24530
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6609019682782344e-06,
+      "loss": 1.2253,
+      "step": 24531
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6608064207911334e-06,
+      "loss": 1.0238,
+      "step": 24532
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660710873304032e-06,
+      "loss": 0.93,
+      "step": 24533
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6606153258169315e-06,
+      "loss": 1.0708,
+      "step": 24534
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.66051977832983e-06,
+      "loss": 0.6484,
+      "step": 24535
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660424230842729e-06,
+      "loss": 0.7297,
+      "step": 24536
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660328683355628e-06,
+      "loss": 0.8504,
+      "step": 24537
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660233135868527e-06,
+      "loss": 0.7783,
+      "step": 24538
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660137588381426e-06,
+      "loss": 1.0118,
+      "step": 24539
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6600420408943245e-06,
+      "loss": 0.6155,
+      "step": 24540
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659946493407224e-06,
+      "loss": 0.8714,
+      "step": 24541
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6598509459201225e-06,
+      "loss": 0.8867,
+      "step": 24542
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6597553984330216e-06,
+      "loss": 1.1552,
+      "step": 24543
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.65965985094592e-06,
+      "loss": 0.8786,
+      "step": 24544
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6595643034588192e-06,
+      "loss": 0.5243,
+      "step": 24545
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659468755971718e-06,
+      "loss": 1.1234,
+      "step": 24546
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659373208484617e-06,
+      "loss": 0.789,
+      "step": 24547
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659277660997516e-06,
+      "loss": 0.9542,
+      "step": 24548
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659182113510415e-06,
+      "loss": 1.0027,
+      "step": 24549
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.659086566023314e-06,
+      "loss": 1.0855,
+      "step": 24550
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6589910185362126e-06,
+      "loss": 0.4778,
+      "step": 24551
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6588954710491117e-06,
+      "loss": 0.6376,
+      "step": 24552
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6587999235620103e-06,
+      "loss": 1.1977,
+      "step": 24553
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6587043760749093e-06,
+      "loss": 1.0666,
+      "step": 24554
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.658608828587808e-06,
+      "loss": 0.8579,
+      "step": 24555
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6585132811007074e-06,
+      "loss": 0.3806,
+      "step": 24556
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6584177336136064e-06,
+      "loss": 0.5416,
+      "step": 24557
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.658322186126505e-06,
+      "loss": 0.635,
+      "step": 24558
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.658226638639404e-06,
+      "loss": 0.79,
+      "step": 24559
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6581310911523027e-06,
+      "loss": 0.5764,
+      "step": 24560
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6580355436652017e-06,
+      "loss": 0.9058,
+      "step": 24561
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6579399961781004e-06,
+      "loss": 1.1543,
+      "step": 24562
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.657844448691e-06,
+      "loss": 0.8006,
+      "step": 24563
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.657748901203899e-06,
+      "loss": 1.1389,
+      "step": 24564
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6576533537167975e-06,
+      "loss": 0.867,
+      "step": 24565
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6575578062296965e-06,
+      "loss": 0.7937,
+      "step": 24566
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.657462258742595e-06,
+      "loss": 0.6851,
+      "step": 24567
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.657366711255494e-06,
+      "loss": 1.2247,
+      "step": 24568
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6572711637683928e-06,
+      "loss": 0.7099,
+      "step": 24569
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6571756162812922e-06,
+      "loss": 0.8053,
+      "step": 24570
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6570800687941913e-06,
+      "loss": 1.1006,
+      "step": 24571
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.65698452130709e-06,
+      "loss": 0.8419,
+      "step": 24572
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.656888973819989e-06,
+      "loss": 0.9562,
+      "step": 24573
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6567934263328875e-06,
+      "loss": 0.4688,
+      "step": 24574
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6566978788457866e-06,
+      "loss": 0.7443,
+      "step": 24575
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.656602331358685e-06,
+      "loss": 1.0863,
+      "step": 24576
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6565067838715847e-06,
+      "loss": 1.154,
+      "step": 24577
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6564112363844837e-06,
+      "loss": 1.0683,
+      "step": 24578
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6563156888973823e-06,
+      "loss": 0.5246,
+      "step": 24579
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6562201414102814e-06,
+      "loss": 0.7645,
+      "step": 24580
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.65612459392318e-06,
+      "loss": 0.5545,
+      "step": 24581
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.656029046436079e-06,
+      "loss": 0.4784,
+      "step": 24582
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6559334989489776e-06,
+      "loss": 0.5774,
+      "step": 24583
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6558379514618767e-06,
+      "loss": 0.7799,
+      "step": 24584
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6557424039747757e-06,
+      "loss": 0.4465,
+      "step": 24585
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6556468564876747e-06,
+      "loss": 0.9396,
+      "step": 24586
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6555513090005734e-06,
+      "loss": 0.6377,
+      "step": 24587
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6554557615134724e-06,
+      "loss": 0.692,
+      "step": 24588
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6553602140263714e-06,
+      "loss": 1.1905,
+      "step": 24589
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.65526466653927e-06,
+      "loss": 0.5911,
+      "step": 24590
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.655169119052169e-06,
+      "loss": 0.905,
+      "step": 24591
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.655073571565068e-06,
+      "loss": 0.7926,
+      "step": 24592
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654978024077967e-06,
+      "loss": 1.0822,
+      "step": 24593
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6548824765908658e-06,
+      "loss": 0.7756,
+      "step": 24594
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654786929103765e-06,
+      "loss": 0.7849,
+      "step": 24595
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6546913816166634e-06,
+      "loss": 1.0765,
+      "step": 24596
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6545958341295625e-06,
+      "loss": 0.972,
+      "step": 24597
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654500286642461e-06,
+      "loss": 0.5354,
+      "step": 24598
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6544047391553605e-06,
+      "loss": 0.9885,
+      "step": 24599
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6543091916682596e-06,
+      "loss": 0.7741,
+      "step": 24600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654213644181158e-06,
+      "loss": 1.0088,
+      "step": 24601
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6541180966940572e-06,
+      "loss": 0.7077,
+      "step": 24602
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654022549206956e-06,
+      "loss": 0.7393,
+      "step": 24603
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.653927001719855e-06,
+      "loss": 1.2045,
+      "step": 24604
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6538314542327535e-06,
+      "loss": 0.4737,
+      "step": 24605
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.653735906745653e-06,
+      "loss": 0.933,
+      "step": 24606
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.653640359258552e-06,
+      "loss": 0.657,
+      "step": 24607
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6535448117714506e-06,
+      "loss": 0.7347,
+      "step": 24608
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6534492642843497e-06,
+      "loss": 0.7484,
+      "step": 24609
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6533537167972483e-06,
+      "loss": 1.3876,
+      "step": 24610
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6532581693101473e-06,
+      "loss": 0.9813,
+      "step": 24611
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.653162621823046e-06,
+      "loss": 0.8074,
+      "step": 24612
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6530670743359454e-06,
+      "loss": 0.511,
+      "step": 24613
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6529715268488444e-06,
+      "loss": 0.6669,
+      "step": 24614
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.652875979361743e-06,
+      "loss": 0.9427,
+      "step": 24615
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.652780431874642e-06,
+      "loss": 1.077,
+      "step": 24616
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6526848843875407e-06,
+      "loss": 1.1581,
+      "step": 24617
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6525893369004397e-06,
+      "loss": 0.7049,
+      "step": 24618
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6524937894133384e-06,
+      "loss": 1.1103,
+      "step": 24619
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6523982419262374e-06,
+      "loss": 0.2889,
+      "step": 24620
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.652302694439137e-06,
+      "loss": 0.6365,
+      "step": 24621
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6522071469520355e-06,
+      "loss": 0.522,
+      "step": 24622
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6521115994649345e-06,
+      "loss": 1.3496,
+      "step": 24623
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.652016051977833e-06,
+      "loss": 0.7652,
+      "step": 24624
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651920504490732e-06,
+      "loss": 1.1774,
+      "step": 24625
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6518249570036308e-06,
+      "loss": 0.9981,
+      "step": 24626
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.65172940951653e-06,
+      "loss": 0.6129,
+      "step": 24627
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6516338620294293e-06,
+      "loss": 1.1328,
+      "step": 24628
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651538314542328e-06,
+      "loss": 0.5704,
+      "step": 24629
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651442767055227e-06,
+      "loss": 1.0818,
+      "step": 24630
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6513472195681255e-06,
+      "loss": 1.2962,
+      "step": 24631
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6512516720810246e-06,
+      "loss": 0.7895,
+      "step": 24632
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651156124593923e-06,
+      "loss": 1.0001,
+      "step": 24633
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6510605771068222e-06,
+      "loss": 1.2142,
+      "step": 24634
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6509650296197213e-06,
+      "loss": 0.8662,
+      "step": 24635
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6508694821326203e-06,
+      "loss": 0.9393,
+      "step": 24636
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650773934645519e-06,
+      "loss": 1.2433,
+      "step": 24637
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650678387158418e-06,
+      "loss": 0.9358,
+      "step": 24638
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650582839671317e-06,
+      "loss": 0.4734,
+      "step": 24639
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6504872921842156e-06,
+      "loss": 0.5882,
+      "step": 24640
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6503917446971147e-06,
+      "loss": 0.8096,
+      "step": 24641
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6502961972100137e-06,
+      "loss": 0.6701,
+      "step": 24642
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6502006497229127e-06,
+      "loss": 0.6175,
+      "step": 24643
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6501051022358114e-06,
+      "loss": 0.9814,
+      "step": 24644
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6500095547487104e-06,
+      "loss": 1.1537,
+      "step": 24645
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649914007261609e-06,
+      "loss": 0.7725,
+      "step": 24646
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649818459774508e-06,
+      "loss": 0.7033,
+      "step": 24647
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6497229122874067e-06,
+      "loss": 0.6263,
+      "step": 24648
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649627364800306e-06,
+      "loss": 0.919,
+      "step": 24649
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649531817313205e-06,
+      "loss": 1.1769,
+      "step": 24650
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6494362698261038e-06,
+      "loss": 1.1683,
+      "step": 24651
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649340722339003e-06,
+      "loss": 0.6142,
+      "step": 24652
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6492451748519014e-06,
+      "loss": 0.7132,
+      "step": 24653
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6491496273648005e-06,
+      "loss": 0.3914,
+      "step": 24654
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.649054079877699e-06,
+      "loss": 0.6235,
+      "step": 24655
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.648958532390598e-06,
+      "loss": 1.2634,
+      "step": 24656
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6488629849034976e-06,
+      "loss": 0.8103,
+      "step": 24657
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.648767437416396e-06,
+      "loss": 0.5809,
+      "step": 24658
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6486718899292952e-06,
+      "loss": 0.8811,
+      "step": 24659
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.648576342442194e-06,
+      "loss": 0.7496,
+      "step": 24660
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.648480794955093e-06,
+      "loss": 0.9954,
+      "step": 24661
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6483852474679915e-06,
+      "loss": 0.3421,
+      "step": 24662
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6482896999808905e-06,
+      "loss": 0.713,
+      "step": 24663
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.64819415249379e-06,
+      "loss": 0.9918,
+      "step": 24664
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6480986050066886e-06,
+      "loss": 1.095,
+      "step": 24665
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6480030575195877e-06,
+      "loss": 0.8448,
+      "step": 24666
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6479075100324863e-06,
+      "loss": 0.5401,
+      "step": 24667
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6478119625453853e-06,
+      "loss": 0.9257,
+      "step": 24668
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.647716415058284e-06,
+      "loss": 0.7761,
+      "step": 24669
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.647620867571183e-06,
+      "loss": 0.5231,
+      "step": 24670
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6475253200840824e-06,
+      "loss": 0.6042,
+      "step": 24671
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.647429772596981e-06,
+      "loss": 0.8248,
+      "step": 24672
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.64733422510988e-06,
+      "loss": 0.9313,
+      "step": 24673
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6472386776227787e-06,
+      "loss": 0.8399,
+      "step": 24674
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6471431301356777e-06,
+      "loss": 0.7208,
+      "step": 24675
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6470475826485763e-06,
+      "loss": 0.6601,
+      "step": 24676
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6469520351614754e-06,
+      "loss": 1.2046,
+      "step": 24677
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.646856487674375e-06,
+      "loss": 1.0939,
+      "step": 24678
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6467609401872735e-06,
+      "loss": 0.4066,
+      "step": 24679
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6466653927001725e-06,
+      "loss": 0.8793,
+      "step": 24680
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.646569845213071e-06,
+      "loss": 0.6751,
+      "step": 24681
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.64647429772597e-06,
+      "loss": 0.8886,
+      "step": 24682
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6463787502388688e-06,
+      "loss": 0.5442,
+      "step": 24683
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.646283202751768e-06,
+      "loss": 0.6058,
+      "step": 24684
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.646187655264667e-06,
+      "loss": 1.2684,
+      "step": 24685
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.646092107777566e-06,
+      "loss": 0.6636,
+      "step": 24686
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6459965602904645e-06,
+      "loss": 0.6629,
+      "step": 24687
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6459010128033635e-06,
+      "loss": 0.6903,
+      "step": 24688
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.645805465316262e-06,
+      "loss": 0.7829,
+      "step": 24689
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.645709917829161e-06,
+      "loss": 0.8565,
+      "step": 24690
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6456143703420602e-06,
+      "loss": 0.5178,
+      "step": 24691
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.645518822854959e-06,
+      "loss": 1.0984,
+      "step": 24692
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6454232753678583e-06,
+      "loss": 0.6776,
+      "step": 24693
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.645327727880757e-06,
+      "loss": 0.6197,
+      "step": 24694
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.645232180393656e-06,
+      "loss": 0.7232,
+      "step": 24695
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6451366329065546e-06,
+      "loss": 0.7024,
+      "step": 24696
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6450410854194536e-06,
+      "loss": 0.7785,
+      "step": 24697
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6449455379323522e-06,
+      "loss": 1.0465,
+      "step": 24698
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6448499904452513e-06,
+      "loss": 0.5676,
+      "step": 24699
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6447544429581507e-06,
+      "loss": 0.8636,
+      "step": 24700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6446588954710493e-06,
+      "loss": 0.6738,
+      "step": 24701
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6445633479839484e-06,
+      "loss": 0.8966,
+      "step": 24702
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.644467800496847e-06,
+      "loss": 0.5431,
+      "step": 24703
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.644372253009746e-06,
+      "loss": 1.1575,
+      "step": 24704
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6442767055226447e-06,
+      "loss": 0.7323,
+      "step": 24705
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6441811580355437e-06,
+      "loss": 0.8436,
+      "step": 24706
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.644085610548443e-06,
+      "loss": 0.8569,
+      "step": 24707
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6439900630613418e-06,
+      "loss": 0.8577,
+      "step": 24708
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643894515574241e-06,
+      "loss": 0.4882,
+      "step": 24709
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6437989680871394e-06,
+      "loss": 0.4557,
+      "step": 24710
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6437034206000385e-06,
+      "loss": 0.7534,
+      "step": 24711
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643607873112937e-06,
+      "loss": 0.7947,
+      "step": 24712
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643512325625836e-06,
+      "loss": 0.7929,
+      "step": 24713
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6434167781387356e-06,
+      "loss": 1.0946,
+      "step": 24714
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643321230651634e-06,
+      "loss": 0.8024,
+      "step": 24715
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6432256831645332e-06,
+      "loss": 1.4162,
+      "step": 24716
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643130135677432e-06,
+      "loss": 0.8244,
+      "step": 24717
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.643034588190331e-06,
+      "loss": 1.0996,
+      "step": 24718
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6429390407032295e-06,
+      "loss": 0.4861,
+      "step": 24719
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6428434932161285e-06,
+      "loss": 0.9354,
+      "step": 24720
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.642747945729028e-06,
+      "loss": 1.2068,
+      "step": 24721
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6426523982419266e-06,
+      "loss": 0.6905,
+      "step": 24722
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6425568507548257e-06,
+      "loss": 0.9334,
+      "step": 24723
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6424613032677243e-06,
+      "loss": 0.7983,
+      "step": 24724
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6423657557806233e-06,
+      "loss": 0.8887,
+      "step": 24725
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.642270208293522e-06,
+      "loss": 0.7505,
+      "step": 24726
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.642174660806421e-06,
+      "loss": 0.6543,
+      "step": 24727
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6420791133193196e-06,
+      "loss": 0.6574,
+      "step": 24728
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.641983565832219e-06,
+      "loss": 0.8164,
+      "step": 24729
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.641888018345118e-06,
+      "loss": 0.8333,
+      "step": 24730
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6417924708580167e-06,
+      "loss": 0.691,
+      "step": 24731
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6416969233709157e-06,
+      "loss": 0.782,
+      "step": 24732
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6416013758838143e-06,
+      "loss": 1.0095,
+      "step": 24733
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6415058283967134e-06,
+      "loss": 0.7048,
+      "step": 24734
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.641410280909612e-06,
+      "loss": 0.9315,
+      "step": 24735
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6413147334225115e-06,
+      "loss": 0.7991,
+      "step": 24736
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.64121918593541e-06,
+      "loss": 0.8594,
+      "step": 24737
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.641123638448309e-06,
+      "loss": 0.5798,
+      "step": 24738
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6410280909612077e-06,
+      "loss": 0.7325,
+      "step": 24739
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6409325434741068e-06,
+      "loss": 0.8079,
+      "step": 24740
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640836995987006e-06,
+      "loss": 0.8046,
+      "step": 24741
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6407414484999044e-06,
+      "loss": 1.1518,
+      "step": 24742
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640645901012804e-06,
+      "loss": 0.9341,
+      "step": 24743
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6405503535257025e-06,
+      "loss": 0.6745,
+      "step": 24744
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6404548060386015e-06,
+      "loss": 1.1868,
+      "step": 24745
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6403592585515e-06,
+      "loss": 0.8066,
+      "step": 24746
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640263711064399e-06,
+      "loss": 0.5396,
+      "step": 24747
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640168163577298e-06,
+      "loss": 0.7554,
+      "step": 24748
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640072616090197e-06,
+      "loss": 0.6203,
+      "step": 24749
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6399770686030963e-06,
+      "loss": 0.9865,
+      "step": 24750
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.639881521115995e-06,
+      "loss": 0.6841,
+      "step": 24751
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.639785973628894e-06,
+      "loss": 0.9284,
+      "step": 24752
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6396904261417926e-06,
+      "loss": 0.575,
+      "step": 24753
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6395948786546916e-06,
+      "loss": 0.8141,
+      "step": 24754
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6394993311675902e-06,
+      "loss": 0.7993,
+      "step": 24755
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6394037836804893e-06,
+      "loss": 0.4187,
+      "step": 24756
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6393082361933887e-06,
+      "loss": 1.1158,
+      "step": 24757
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6392126887062873e-06,
+      "loss": 0.6876,
+      "step": 24758
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6391171412191864e-06,
+      "loss": 0.4337,
+      "step": 24759
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.639021593732085e-06,
+      "loss": 0.6925,
+      "step": 24760
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.638926046244984e-06,
+      "loss": 0.8897,
+      "step": 24761
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6388304987578827e-06,
+      "loss": 0.5826,
+      "step": 24762
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6387349512707817e-06,
+      "loss": 0.8055,
+      "step": 24763
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6386394037836803e-06,
+      "loss": 0.6094,
+      "step": 24764
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6385438562965798e-06,
+      "loss": 0.8132,
+      "step": 24765
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.638448308809479e-06,
+      "loss": 1.194,
+      "step": 24766
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6383527613223774e-06,
+      "loss": 0.8105,
+      "step": 24767
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6382572138352765e-06,
+      "loss": 0.7459,
+      "step": 24768
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.638161666348175e-06,
+      "loss": 0.5882,
+      "step": 24769
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.638066118861074e-06,
+      "loss": 0.9088,
+      "step": 24770
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6379705713739727e-06,
+      "loss": 0.6792,
+      "step": 24771
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.637875023886872e-06,
+      "loss": 0.9554,
+      "step": 24772
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6377794763997712e-06,
+      "loss": 1.096,
+      "step": 24773
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.63768392891267e-06,
+      "loss": 0.7114,
+      "step": 24774
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.637588381425569e-06,
+      "loss": 0.6748,
+      "step": 24775
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6374928339384675e-06,
+      "loss": 0.7616,
+      "step": 24776
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6373972864513665e-06,
+      "loss": 0.5394,
+      "step": 24777
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.637301738964265e-06,
+      "loss": 0.5137,
+      "step": 24778
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6372061914771646e-06,
+      "loss": 0.9584,
+      "step": 24779
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6371106439900637e-06,
+      "loss": 1.1614,
+      "step": 24780
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6370150965029623e-06,
+      "loss": 0.9606,
+      "step": 24781
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6369195490158613e-06,
+      "loss": 0.8612,
+      "step": 24782
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.63682400152876e-06,
+      "loss": 1.1467,
+      "step": 24783
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.636728454041659e-06,
+      "loss": 0.5909,
+      "step": 24784
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6366329065545576e-06,
+      "loss": 1.364,
+      "step": 24785
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.636537359067457e-06,
+      "loss": 1.0384,
+      "step": 24786
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6364418115803557e-06,
+      "loss": 0.6409,
+      "step": 24787
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6363462640932547e-06,
+      "loss": 0.9783,
+      "step": 24788
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6362507166061533e-06,
+      "loss": 0.5138,
+      "step": 24789
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6361551691190523e-06,
+      "loss": 0.5712,
+      "step": 24790
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.636059621631951e-06,
+      "loss": 0.73,
+      "step": 24791
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.63596407414485e-06,
+      "loss": 0.586,
+      "step": 24792
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6358685266577495e-06,
+      "loss": 1.0522,
+      "step": 24793
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.635772979170648e-06,
+      "loss": 0.8558,
+      "step": 24794
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.635677431683547e-06,
+      "loss": 0.599,
+      "step": 24795
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6355818841964457e-06,
+      "loss": 0.916,
+      "step": 24796
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6354863367093448e-06,
+      "loss": 0.4302,
+      "step": 24797
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6353907892222434e-06,
+      "loss": 0.8517,
+      "step": 24798
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6352952417351424e-06,
+      "loss": 0.6759,
+      "step": 24799
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.635199694248041e-06,
+      "loss": 0.3205,
+      "step": 24800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6351041467609405e-06,
+      "loss": 0.3519,
+      "step": 24801
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6350085992738395e-06,
+      "loss": 0.8292,
+      "step": 24802
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634913051786738e-06,
+      "loss": 0.6292,
+      "step": 24803
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634817504299637e-06,
+      "loss": 0.9571,
+      "step": 24804
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634721956812536e-06,
+      "loss": 0.8702,
+      "step": 24805
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634626409325435e-06,
+      "loss": 0.6174,
+      "step": 24806
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6345308618383335e-06,
+      "loss": 1.0497,
+      "step": 24807
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634435314351233e-06,
+      "loss": 1.0173,
+      "step": 24808
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.634339766864132e-06,
+      "loss": 1.2328,
+      "step": 24809
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6342442193770306e-06,
+      "loss": 1.0297,
+      "step": 24810
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6341486718899296e-06,
+      "loss": 0.8391,
+      "step": 24811
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6340531244028282e-06,
+      "loss": 0.9962,
+      "step": 24812
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6339575769157273e-06,
+      "loss": 0.4638,
+      "step": 24813
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.633862029428626e-06,
+      "loss": 0.7632,
+      "step": 24814
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6337664819415253e-06,
+      "loss": 1.2536,
+      "step": 24815
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6336709344544244e-06,
+      "loss": 0.7868,
+      "step": 24816
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.633575386967323e-06,
+      "loss": 0.9922,
+      "step": 24817
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.633479839480222e-06,
+      "loss": 0.9958,
+      "step": 24818
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6333842919931207e-06,
+      "loss": 1.0127,
+      "step": 24819
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6332887445060197e-06,
+      "loss": 0.6392,
+      "step": 24820
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6331931970189183e-06,
+      "loss": 0.8341,
+      "step": 24821
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6330976495318178e-06,
+      "loss": 1.3489,
+      "step": 24822
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.633002102044717e-06,
+      "loss": 0.7228,
+      "step": 24823
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6329065545576154e-06,
+      "loss": 0.9886,
+      "step": 24824
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6328110070705145e-06,
+      "loss": 0.846,
+      "step": 24825
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632715459583413e-06,
+      "loss": 1.7045,
+      "step": 24826
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632619912096312e-06,
+      "loss": 0.7886,
+      "step": 24827
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6325243646092107e-06,
+      "loss": 1.2687,
+      "step": 24828
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.63242881712211e-06,
+      "loss": 0.9223,
+      "step": 24829
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632333269635009e-06,
+      "loss": 0.7346,
+      "step": 24830
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632237722147908e-06,
+      "loss": 0.6004,
+      "step": 24831
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632142174660807e-06,
+      "loss": 1.0038,
+      "step": 24832
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6320466271737055e-06,
+      "loss": 0.5584,
+      "step": 24833
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6319510796866045e-06,
+      "loss": 1.0274,
+      "step": 24834
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.631855532199503e-06,
+      "loss": 0.8451,
+      "step": 24835
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.631759984712402e-06,
+      "loss": 0.6732,
+      "step": 24836
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6316644372253012e-06,
+      "loss": 0.9667,
+      "step": 24837
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6315688897382003e-06,
+      "loss": 0.628,
+      "step": 24838
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.631473342251099e-06,
+      "loss": 0.7421,
+      "step": 24839
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.631377794763998e-06,
+      "loss": 0.8996,
+      "step": 24840
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6312822472768965e-06,
+      "loss": 0.7089,
+      "step": 24841
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6311866997897956e-06,
+      "loss": 0.7136,
+      "step": 24842
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6310911523026946e-06,
+      "loss": 0.9717,
+      "step": 24843
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6309956048155937e-06,
+      "loss": 0.4776,
+      "step": 24844
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6309000573284927e-06,
+      "loss": 0.9608,
+      "step": 24845
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6308045098413913e-06,
+      "loss": 1.1917,
+      "step": 24846
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6307089623542903e-06,
+      "loss": 1.4136,
+      "step": 24847
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.630613414867189e-06,
+      "loss": 0.5626,
+      "step": 24848
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.630517867380088e-06,
+      "loss": 0.4782,
+      "step": 24849
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6304223198929866e-06,
+      "loss": 0.9685,
+      "step": 24850
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.630326772405886e-06,
+      "loss": 0.619,
+      "step": 24851
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.630231224918785e-06,
+      "loss": 0.7598,
+      "step": 24852
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6301356774316837e-06,
+      "loss": 0.9527,
+      "step": 24853
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6300401299445828e-06,
+      "loss": 0.5802,
+      "step": 24854
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6299445824574814e-06,
+      "loss": 1.0409,
+      "step": 24855
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6298490349703804e-06,
+      "loss": 0.9238,
+      "step": 24856
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629753487483279e-06,
+      "loss": 0.926,
+      "step": 24857
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6296579399961785e-06,
+      "loss": 0.9039,
+      "step": 24858
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6295623925090775e-06,
+      "loss": 1.0301,
+      "step": 24859
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629466845021976e-06,
+      "loss": 0.6467,
+      "step": 24860
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629371297534875e-06,
+      "loss": 1.2052,
+      "step": 24861
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629275750047774e-06,
+      "loss": 0.5262,
+      "step": 24862
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.629180202560673e-06,
+      "loss": 0.6733,
+      "step": 24863
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6290846550735715e-06,
+      "loss": 1.1319,
+      "step": 24864
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.628989107586471e-06,
+      "loss": 0.742,
+      "step": 24865
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.62889356009937e-06,
+      "loss": 0.8239,
+      "step": 24866
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6287980126122686e-06,
+      "loss": 0.5168,
+      "step": 24867
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6287024651251676e-06,
+      "loss": 0.9089,
+      "step": 24868
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6286069176380662e-06,
+      "loss": 1.0202,
+      "step": 24869
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6285113701509653e-06,
+      "loss": 0.5519,
+      "step": 24870
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.628415822663864e-06,
+      "loss": 0.993,
+      "step": 24871
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.628320275176763e-06,
+      "loss": 0.6867,
+      "step": 24872
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6282247276896624e-06,
+      "loss": 0.8973,
+      "step": 24873
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.628129180202561e-06,
+      "loss": 0.9646,
+      "step": 24874
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.62803363271546e-06,
+      "loss": 0.8443,
+      "step": 24875
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6279380852283587e-06,
+      "loss": 1.0792,
+      "step": 24876
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6278425377412577e-06,
+      "loss": 0.8869,
+      "step": 24877
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6277469902541563e-06,
+      "loss": 0.9304,
+      "step": 24878
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6276514427670553e-06,
+      "loss": 0.4716,
+      "step": 24879
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6275558952799544e-06,
+      "loss": 0.6846,
+      "step": 24880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6274603477928534e-06,
+      "loss": 1.0468,
+      "step": 24881
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6273648003057525e-06,
+      "loss": 0.9362,
+      "step": 24882
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.627269252818651e-06,
+      "loss": 1.3021,
+      "step": 24883
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.62717370533155e-06,
+      "loss": 0.8323,
+      "step": 24884
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6270781578444487e-06,
+      "loss": 0.7871,
+      "step": 24885
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6269826103573478e-06,
+      "loss": 0.831,
+      "step": 24886
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.626887062870247e-06,
+      "loss": 0.9283,
+      "step": 24887
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.626791515383146e-06,
+      "loss": 0.5081,
+      "step": 24888
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6266959678960445e-06,
+      "loss": 0.5016,
+      "step": 24889
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6266004204089435e-06,
+      "loss": 0.8983,
+      "step": 24890
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.626504872921842e-06,
+      "loss": 0.6437,
+      "step": 24891
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.626409325434741e-06,
+      "loss": 0.5966,
+      "step": 24892
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6263137779476398e-06,
+      "loss": 0.7065,
+      "step": 24893
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6262182304605392e-06,
+      "loss": 0.8819,
+      "step": 24894
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6261226829734383e-06,
+      "loss": 0.7521,
+      "step": 24895
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.626027135486337e-06,
+      "loss": 0.8797,
+      "step": 24896
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.625931587999236e-06,
+      "loss": 0.8913,
+      "step": 24897
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6258360405121345e-06,
+      "loss": 0.5638,
+      "step": 24898
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6257404930250336e-06,
+      "loss": 1.256,
+      "step": 24899
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.625644945537932e-06,
+      "loss": 0.9501,
+      "step": 24900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6255493980508317e-06,
+      "loss": 1.2546,
+      "step": 24901
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6254538505637307e-06,
+      "loss": 0.6684,
+      "step": 24902
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6253583030766293e-06,
+      "loss": 0.7139,
+      "step": 24903
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6252627555895283e-06,
+      "loss": 1.1901,
+      "step": 24904
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.625167208102427e-06,
+      "loss": 0.6473,
+      "step": 24905
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.625071660615326e-06,
+      "loss": 1.1157,
+      "step": 24906
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6249761131282246e-06,
+      "loss": 0.8868,
+      "step": 24907
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6248805656411237e-06,
+      "loss": 0.795,
+      "step": 24908
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.624785018154023e-06,
+      "loss": 0.9085,
+      "step": 24909
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6246894706669217e-06,
+      "loss": 0.4387,
+      "step": 24910
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6245939231798208e-06,
+      "loss": 0.539,
+      "step": 24911
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6244983756927194e-06,
+      "loss": 0.7173,
+      "step": 24912
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6244028282056184e-06,
+      "loss": 0.6778,
+      "step": 24913
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.624307280718517e-06,
+      "loss": 0.8877,
+      "step": 24914
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.624211733231416e-06,
+      "loss": 1.126,
+      "step": 24915
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6241161857443155e-06,
+      "loss": 0.7131,
+      "step": 24916
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.624020638257214e-06,
+      "loss": 0.9329,
+      "step": 24917
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.623925090770113e-06,
+      "loss": 0.5941,
+      "step": 24918
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.623829543283012e-06,
+      "loss": 0.7137,
+      "step": 24919
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.623733995795911e-06,
+      "loss": 0.9954,
+      "step": 24920
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6236384483088095e-06,
+      "loss": 0.4502,
+      "step": 24921
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6235429008217085e-06,
+      "loss": 0.8034,
+      "step": 24922
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.623447353334608e-06,
+      "loss": 0.7252,
+      "step": 24923
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6233518058475066e-06,
+      "loss": 0.3462,
+      "step": 24924
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6232562583604056e-06,
+      "loss": 0.751,
+      "step": 24925
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6231607108733042e-06,
+      "loss": 1.1263,
+      "step": 24926
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6230651633862033e-06,
+      "loss": 0.3972,
+      "step": 24927
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.622969615899102e-06,
+      "loss": 1.2349,
+      "step": 24928
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.622874068412001e-06,
+      "loss": 0.8158,
+      "step": 24929
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6227785209249e-06,
+      "loss": 1.0578,
+      "step": 24930
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.622682973437799e-06,
+      "loss": 0.7509,
+      "step": 24931
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6225874259506976e-06,
+      "loss": 0.5693,
+      "step": 24932
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6224918784635967e-06,
+      "loss": 0.7465,
+      "step": 24933
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6223963309764957e-06,
+      "loss": 0.7517,
+      "step": 24934
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6223007834893943e-06,
+      "loss": 0.6995,
+      "step": 24935
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6222052360022933e-06,
+      "loss": 0.9964,
+      "step": 24936
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6221096885151924e-06,
+      "loss": 0.7879,
+      "step": 24937
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6220141410280914e-06,
+      "loss": 0.6601,
+      "step": 24938
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.62191859354099e-06,
+      "loss": 0.7078,
+      "step": 24939
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.621823046053889e-06,
+      "loss": 0.5717,
+      "step": 24940
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6217274985667877e-06,
+      "loss": 0.8296,
+      "step": 24941
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6216319510796867e-06,
+      "loss": 0.9285,
+      "step": 24942
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6215364035925853e-06,
+      "loss": 1.0156,
+      "step": 24943
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6214408561054844e-06,
+      "loss": 0.5437,
+      "step": 24944
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.621345308618384e-06,
+      "loss": 0.8031,
+      "step": 24945
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6212497611312825e-06,
+      "loss": 0.8022,
+      "step": 24946
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6211542136441815e-06,
+      "loss": 0.7748,
+      "step": 24947
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.62105866615708e-06,
+      "loss": 1.0337,
+      "step": 24948
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.620963118669979e-06,
+      "loss": 0.8645,
+      "step": 24949
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6208675711828778e-06,
+      "loss": 0.458,
+      "step": 24950
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.620772023695777e-06,
+      "loss": 0.8075,
+      "step": 24951
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6206764762086763e-06,
+      "loss": 0.9262,
+      "step": 24952
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.620580928721575e-06,
+      "loss": 0.6512,
+      "step": 24953
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.620485381234474e-06,
+      "loss": 0.9596,
+      "step": 24954
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6203898337473725e-06,
+      "loss": 0.7938,
+      "step": 24955
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6202942862602716e-06,
+      "loss": 0.7666,
+      "step": 24956
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.62019873877317e-06,
+      "loss": 1.2432,
+      "step": 24957
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6201031912860692e-06,
+      "loss": 0.7169,
+      "step": 24958
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6200076437989687e-06,
+      "loss": 0.9957,
+      "step": 24959
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6199120963118673e-06,
+      "loss": 0.7362,
+      "step": 24960
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6198165488247663e-06,
+      "loss": 0.8934,
+      "step": 24961
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.619721001337665e-06,
+      "loss": 0.485,
+      "step": 24962
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.619625453850564e-06,
+      "loss": 1.0863,
+      "step": 24963
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6195299063634626e-06,
+      "loss": 0.9361,
+      "step": 24964
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6194343588763617e-06,
+      "loss": 0.755,
+      "step": 24965
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.619338811389261e-06,
+      "loss": 1.1674,
+      "step": 24966
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6192432639021597e-06,
+      "loss": 1.0821,
+      "step": 24967
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6191477164150588e-06,
+      "loss": 1.1985,
+      "step": 24968
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6190521689279574e-06,
+      "loss": 0.59,
+      "step": 24969
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6189566214408564e-06,
+      "loss": 0.7488,
+      "step": 24970
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618861073953755e-06,
+      "loss": 0.478,
+      "step": 24971
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618765526466654e-06,
+      "loss": 0.4306,
+      "step": 24972
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6186699789795535e-06,
+      "loss": 0.421,
+      "step": 24973
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618574431492452e-06,
+      "loss": 0.5121,
+      "step": 24974
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618478884005351e-06,
+      "loss": 0.4747,
+      "step": 24975
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.61838333651825e-06,
+      "loss": 0.8385,
+      "step": 24976
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618287789031149e-06,
+      "loss": 0.9091,
+      "step": 24977
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6181922415440475e-06,
+      "loss": 0.72,
+      "step": 24978
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6180966940569465e-06,
+      "loss": 1.0128,
+      "step": 24979
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.618001146569845e-06,
+      "loss": 0.6498,
+      "step": 24980
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6179055990827446e-06,
+      "loss": 0.8054,
+      "step": 24981
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.617810051595643e-06,
+      "loss": 1.1574,
+      "step": 24982
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6177145041085422e-06,
+      "loss": 0.7451,
+      "step": 24983
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.617618956621441e-06,
+      "loss": 0.8104,
+      "step": 24984
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.61752340913434e-06,
+      "loss": 0.9639,
+      "step": 24985
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.617427861647239e-06,
+      "loss": 0.7967,
+      "step": 24986
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6173323141601375e-06,
+      "loss": 0.8442,
+      "step": 24987
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.617236766673037e-06,
+      "loss": 0.4592,
+      "step": 24988
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6171412191859356e-06,
+      "loss": 1.1442,
+      "step": 24989
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6170456716988347e-06,
+      "loss": 0.7902,
+      "step": 24990
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6169501242117333e-06,
+      "loss": 0.6267,
+      "step": 24991
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6168545767246323e-06,
+      "loss": 0.8974,
+      "step": 24992
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.616759029237531e-06,
+      "loss": 0.4818,
+      "step": 24993
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.61666348175043e-06,
+      "loss": 0.7265,
+      "step": 24994
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6165679342633294e-06,
+      "loss": 0.8881,
+      "step": 24995
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.616472386776228e-06,
+      "loss": 0.8959,
+      "step": 24996
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.616376839289127e-06,
+      "loss": 0.697,
+      "step": 24997
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6162812918020257e-06,
+      "loss": 0.777,
+      "step": 24998
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6161857443149247e-06,
+      "loss": 0.8715,
+      "step": 24999
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6160901968278233e-06,
+      "loss": 0.8166,
+      "step": 25000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6159946493407224e-06,
+      "loss": 0.3299,
+      "step": 25001
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615899101853622e-06,
+      "loss": 1.007,
+      "step": 25002
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6158035543665205e-06,
+      "loss": 0.5945,
+      "step": 25003
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6157080068794195e-06,
+      "loss": 1.0921,
+      "step": 25004
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615612459392318e-06,
+      "loss": 0.2919,
+      "step": 25005
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615516911905217e-06,
+      "loss": 0.9587,
+      "step": 25006
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6154213644181158e-06,
+      "loss": 1.0521,
+      "step": 25007
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615325816931015e-06,
+      "loss": 0.6559,
+      "step": 25008
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6152302694439143e-06,
+      "loss": 0.9983,
+      "step": 25009
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615134721956813e-06,
+      "loss": 0.5199,
+      "step": 25010
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.615039174469712e-06,
+      "loss": 1.0761,
+      "step": 25011
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6149436269826105e-06,
+      "loss": 1.0046,
+      "step": 25012
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6148480794955096e-06,
+      "loss": 0.97,
+      "step": 25013
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.614752532008408e-06,
+      "loss": 0.5141,
+      "step": 25014
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6146569845213072e-06,
+      "loss": 1.1193,
+      "step": 25015
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.614561437034206e-06,
+      "loss": 0.9564,
+      "step": 25016
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6144658895471053e-06,
+      "loss": 1.1828,
+      "step": 25017
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6143703420600043e-06,
+      "loss": 0.8139,
+      "step": 25018
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.614274794572903e-06,
+      "loss": 0.6758,
+      "step": 25019
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.614179247085802e-06,
+      "loss": 0.7908,
+      "step": 25020
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6140836995987006e-06,
+      "loss": 0.5597,
+      "step": 25021
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6139881521115997e-06,
+      "loss": 0.7701,
+      "step": 25022
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6138926046244983e-06,
+      "loss": 0.7613,
+      "step": 25023
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6137970571373977e-06,
+      "loss": 0.6196,
+      "step": 25024
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6137015096502968e-06,
+      "loss": 0.632,
+      "step": 25025
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6136059621631954e-06,
+      "loss": 0.6124,
+      "step": 25026
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6135104146760944e-06,
+      "loss": 0.9052,
+      "step": 25027
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.613414867188993e-06,
+      "loss": 0.9387,
+      "step": 25028
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.613319319701892e-06,
+      "loss": 0.4769,
+      "step": 25029
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6132237722147907e-06,
+      "loss": 0.9764,
+      "step": 25030
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.61312822472769e-06,
+      "loss": 0.8528,
+      "step": 25031
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6130326772405888e-06,
+      "loss": 0.7109,
+      "step": 25032
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612937129753488e-06,
+      "loss": 0.865,
+      "step": 25033
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6128415822663864e-06,
+      "loss": 0.6718,
+      "step": 25034
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6127460347792855e-06,
+      "loss": 0.992,
+      "step": 25035
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6126504872921845e-06,
+      "loss": 0.7605,
+      "step": 25036
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612554939805083e-06,
+      "loss": 0.9122,
+      "step": 25037
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6124593923179826e-06,
+      "loss": 0.8882,
+      "step": 25038
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612363844830881e-06,
+      "loss": 0.6026,
+      "step": 25039
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6122682973437802e-06,
+      "loss": 0.9261,
+      "step": 25040
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612172749856679e-06,
+      "loss": 0.6217,
+      "step": 25041
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612077202369578e-06,
+      "loss": 1.6025,
+      "step": 25042
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6119816548824765e-06,
+      "loss": 0.6362,
+      "step": 25043
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6118861073953755e-06,
+      "loss": 0.732,
+      "step": 25044
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.611790559908275e-06,
+      "loss": 0.9244,
+      "step": 25045
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6116950124211736e-06,
+      "loss": 0.7511,
+      "step": 25046
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6115994649340727e-06,
+      "loss": 0.7545,
+      "step": 25047
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6115039174469713e-06,
+      "loss": 0.5037,
+      "step": 25048
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6114083699598703e-06,
+      "loss": 0.7709,
+      "step": 25049
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.611312822472769e-06,
+      "loss": 1.122,
+      "step": 25050
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.611217274985668e-06,
+      "loss": 0.7965,
+      "step": 25051
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6111217274985666e-06,
+      "loss": 0.959,
+      "step": 25052
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.611026180011466e-06,
+      "loss": 0.5659,
+      "step": 25053
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.610930632524365e-06,
+      "loss": 0.4572,
+      "step": 25054
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6108350850372637e-06,
+      "loss": 1.1698,
+      "step": 25055
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6107395375501627e-06,
+      "loss": 0.9049,
+      "step": 25056
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6106439900630613e-06,
+      "loss": 0.5947,
+      "step": 25057
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6105484425759604e-06,
+      "loss": 1.1129,
+      "step": 25058
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.610452895088859e-06,
+      "loss": 1.0561,
+      "step": 25059
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6103573476017585e-06,
+      "loss": 0.7433,
+      "step": 25060
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6102618001146575e-06,
+      "loss": 0.8312,
+      "step": 25061
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.610166252627556e-06,
+      "loss": 0.6773,
+      "step": 25062
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.610070705140455e-06,
+      "loss": 1.1461,
+      "step": 25063
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6099751576533538e-06,
+      "loss": 0.6806,
+      "step": 25064
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.609879610166253e-06,
+      "loss": 0.8635,
+      "step": 25065
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6097840626791514e-06,
+      "loss": 0.769,
+      "step": 25066
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.609688515192051e-06,
+      "loss": 1.1556,
+      "step": 25067
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.60959296770495e-06,
+      "loss": 1.0866,
+      "step": 25068
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6094974202178485e-06,
+      "loss": 0.565,
+      "step": 25069
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6094018727307476e-06,
+      "loss": 0.8793,
+      "step": 25070
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.609306325243646e-06,
+      "loss": 1.0126,
+      "step": 25071
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6092107777565452e-06,
+      "loss": 0.9261,
+      "step": 25072
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.609115230269444e-06,
+      "loss": 1.1309,
+      "step": 25073
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6090196827823433e-06,
+      "loss": 0.7183,
+      "step": 25074
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6089241352952423e-06,
+      "loss": 0.5648,
+      "step": 25075
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.608828587808141e-06,
+      "loss": 1.1957,
+      "step": 25076
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.60873304032104e-06,
+      "loss": 0.9884,
+      "step": 25077
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6086374928339386e-06,
+      "loss": 0.7151,
+      "step": 25078
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6085419453468376e-06,
+      "loss": 0.5764,
+      "step": 25079
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6084463978597363e-06,
+      "loss": 0.9656,
+      "step": 25080
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6083508503726357e-06,
+      "loss": 0.6627,
+      "step": 25081
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6082553028855343e-06,
+      "loss": 0.6186,
+      "step": 25082
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6081597553984334e-06,
+      "loss": 0.8192,
+      "step": 25083
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.608064207911332e-06,
+      "loss": 0.9549,
+      "step": 25084
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.607968660424231e-06,
+      "loss": 0.8148,
+      "step": 25085
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6078731129371296e-06,
+      "loss": 1.0499,
+      "step": 25086
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6077775654500287e-06,
+      "loss": 0.7851,
+      "step": 25087
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6076820179629277e-06,
+      "loss": 1.0228,
+      "step": 25088
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6075864704758268e-06,
+      "loss": 0.7579,
+      "step": 25089
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.607490922988726e-06,
+      "loss": 1.0045,
+      "step": 25090
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6073953755016244e-06,
+      "loss": 1.1342,
+      "step": 25091
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6072998280145235e-06,
+      "loss": 1.0423,
+      "step": 25092
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.607204280527422e-06,
+      "loss": 0.6271,
+      "step": 25093
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.607108733040321e-06,
+      "loss": 0.6689,
+      "step": 25094
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6070131855532197e-06,
+      "loss": 0.7916,
+      "step": 25095
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.606917638066119e-06,
+      "loss": 0.6021,
+      "step": 25096
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6068220905790182e-06,
+      "loss": 1.0072,
+      "step": 25097
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.606726543091917e-06,
+      "loss": 0.9068,
+      "step": 25098
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.606630995604816e-06,
+      "loss": 1.0325,
+      "step": 25099
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6065354481177145e-06,
+      "loss": 0.6622,
+      "step": 25100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6064399006306135e-06,
+      "loss": 0.8265,
+      "step": 25101
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.606344353143512e-06,
+      "loss": 0.9464,
+      "step": 25102
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6062488056564116e-06,
+      "loss": 0.5588,
+      "step": 25103
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6061532581693106e-06,
+      "loss": 1.2395,
+      "step": 25104
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6060577106822093e-06,
+      "loss": 0.9361,
+      "step": 25105
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6059621631951083e-06,
+      "loss": 0.6571,
+      "step": 25106
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.605866615708007e-06,
+      "loss": 0.8841,
+      "step": 25107
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.605771068220906e-06,
+      "loss": 0.8169,
+      "step": 25108
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6056755207338046e-06,
+      "loss": 0.3857,
+      "step": 25109
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.605579973246704e-06,
+      "loss": 1.0279,
+      "step": 25110
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.605484425759603e-06,
+      "loss": 1.1167,
+      "step": 25111
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6053888782725017e-06,
+      "loss": 0.6018,
+      "step": 25112
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6052933307854007e-06,
+      "loss": 1.3748,
+      "step": 25113
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6051977832982993e-06,
+      "loss": 0.8358,
+      "step": 25114
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6051022358111984e-06,
+      "loss": 0.8999,
+      "step": 25115
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.605006688324097e-06,
+      "loss": 0.8313,
+      "step": 25116
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6049111408369965e-06,
+      "loss": 0.8792,
+      "step": 25117
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6048155933498955e-06,
+      "loss": 0.8611,
+      "step": 25118
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.604720045862794e-06,
+      "loss": 0.5963,
+      "step": 25119
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.604624498375693e-06,
+      "loss": 1.6215,
+      "step": 25120
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6045289508885918e-06,
+      "loss": 0.9817,
+      "step": 25121
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.604433403401491e-06,
+      "loss": 0.7555,
+      "step": 25122
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6043378559143894e-06,
+      "loss": 0.8988,
+      "step": 25123
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6042423084272885e-06,
+      "loss": 0.605,
+      "step": 25124
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6041467609401875e-06,
+      "loss": 0.4724,
+      "step": 25125
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6040512134530865e-06,
+      "loss": 0.9404,
+      "step": 25126
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6039556659659856e-06,
+      "loss": 1.0042,
+      "step": 25127
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.603860118478884e-06,
+      "loss": 1.0108,
+      "step": 25128
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6037645709917832e-06,
+      "loss": 0.6017,
+      "step": 25129
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.603669023504682e-06,
+      "loss": 0.9511,
+      "step": 25130
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.603573476017581e-06,
+      "loss": 1.0246,
+      "step": 25131
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.60347792853048e-06,
+      "loss": 1.0567,
+      "step": 25132
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.603382381043379e-06,
+      "loss": 0.5852,
+      "step": 25133
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6032868335562776e-06,
+      "loss": 0.9411,
+      "step": 25134
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6031912860691766e-06,
+      "loss": 0.5006,
+      "step": 25135
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6030957385820752e-06,
+      "loss": 0.7483,
+      "step": 25136
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6030001910949743e-06,
+      "loss": 0.5799,
+      "step": 25137
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6029046436078733e-06,
+      "loss": 0.5358,
+      "step": 25138
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6028090961207723e-06,
+      "loss": 1.0669,
+      "step": 25139
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6027135486336714e-06,
+      "loss": 0.5316,
+      "step": 25140
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.60261800114657e-06,
+      "loss": 0.9816,
+      "step": 25141
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.602522453659469e-06,
+      "loss": 1.1501,
+      "step": 25142
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6024269061723676e-06,
+      "loss": 0.7951,
+      "step": 25143
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6023313586852667e-06,
+      "loss": 0.8398,
+      "step": 25144
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6022358111981653e-06,
+      "loss": 0.9162,
+      "step": 25145
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6021402637110648e-06,
+      "loss": 1.0338,
+      "step": 25146
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.602044716223964e-06,
+      "loss": 0.6465,
+      "step": 25147
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6019491687368624e-06,
+      "loss": 1.0172,
+      "step": 25148
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6018536212497615e-06,
+      "loss": 1.0851,
+      "step": 25149
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.60175807376266e-06,
+      "loss": 0.8905,
+      "step": 25150
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.601662526275559e-06,
+      "loss": 0.9231,
+      "step": 25151
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6015669787884577e-06,
+      "loss": 0.6825,
+      "step": 25152
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.601471431301357e-06,
+      "loss": 0.7328,
+      "step": 25153
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6013758838142562e-06,
+      "loss": 1.1076,
+      "step": 25154
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.601280336327155e-06,
+      "loss": 0.6245,
+      "step": 25155
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.601184788840054e-06,
+      "loss": 1.1121,
+      "step": 25156
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6010892413529525e-06,
+      "loss": 0.9024,
+      "step": 25157
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6009936938658515e-06,
+      "loss": 0.8591,
+      "step": 25158
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.60089814637875e-06,
+      "loss": 1.1233,
+      "step": 25159
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600802598891649e-06,
+      "loss": 0.9205,
+      "step": 25160
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6007070514045486e-06,
+      "loss": 0.6368,
+      "step": 25161
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6006115039174473e-06,
+      "loss": 0.7632,
+      "step": 25162
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6005159564303463e-06,
+      "loss": 1.2736,
+      "step": 25163
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600420408943245e-06,
+      "loss": 0.6079,
+      "step": 25164
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600324861456144e-06,
+      "loss": 0.7764,
+      "step": 25165
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6002293139690426e-06,
+      "loss": 0.4934,
+      "step": 25166
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6001337664819416e-06,
+      "loss": 0.8144,
+      "step": 25167
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600038218994841e-06,
+      "loss": 1.1734,
+      "step": 25168
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5999426715077397e-06,
+      "loss": 0.648,
+      "step": 25169
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5998471240206387e-06,
+      "loss": 0.8759,
+      "step": 25170
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5997515765335373e-06,
+      "loss": 0.9532,
+      "step": 25171
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5996560290464364e-06,
+      "loss": 1.2515,
+      "step": 25172
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.599560481559335e-06,
+      "loss": 1.1331,
+      "step": 25173
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.599464934072234e-06,
+      "loss": 0.7144,
+      "step": 25174
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.599369386585133e-06,
+      "loss": 1.07,
+      "step": 25175
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.599273839098032e-06,
+      "loss": 0.5584,
+      "step": 25176
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.599178291610931e-06,
+      "loss": 0.7078,
+      "step": 25177
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5990827441238298e-06,
+      "loss": 0.897,
+      "step": 25178
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.598987196636729e-06,
+      "loss": 1.042,
+      "step": 25179
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5988916491496274e-06,
+      "loss": 1.0664,
+      "step": 25180
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5987961016625265e-06,
+      "loss": 0.6495,
+      "step": 25181
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5987005541754255e-06,
+      "loss": 0.6831,
+      "step": 25182
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5986050066883245e-06,
+      "loss": 1.0398,
+      "step": 25183
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.598509459201223e-06,
+      "loss": 0.6862,
+      "step": 25184
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.598413911714122e-06,
+      "loss": 0.5709,
+      "step": 25185
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.598318364227021e-06,
+      "loss": 0.8115,
+      "step": 25186
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.59822281673992e-06,
+      "loss": 0.8036,
+      "step": 25187
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5981272692528185e-06,
+      "loss": 0.6891,
+      "step": 25188
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.598031721765718e-06,
+      "loss": 0.7386,
+      "step": 25189
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.597936174278617e-06,
+      "loss": 0.9736,
+      "step": 25190
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5978406267915156e-06,
+      "loss": 1.1237,
+      "step": 25191
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5977450793044146e-06,
+      "loss": 0.7962,
+      "step": 25192
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5976495318173132e-06,
+      "loss": 0.6236,
+      "step": 25193
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5975539843302123e-06,
+      "loss": 0.6751,
+      "step": 25194
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.597458436843111e-06,
+      "loss": 0.6194,
+      "step": 25195
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.59736288935601e-06,
+      "loss": 0.5266,
+      "step": 25196
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5972673418689094e-06,
+      "loss": 1.0324,
+      "step": 25197
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.597171794381808e-06,
+      "loss": 0.618,
+      "step": 25198
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.597076246894707e-06,
+      "loss": 0.5748,
+      "step": 25199
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5969806994076056e-06,
+      "loss": 0.9318,
+      "step": 25200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5968851519205047e-06,
+      "loss": 1.4747,
+      "step": 25201
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5967896044334033e-06,
+      "loss": 0.5089,
+      "step": 25202
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5966940569463023e-06,
+      "loss": 0.5225,
+      "step": 25203
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.596598509459202e-06,
+      "loss": 0.7564,
+      "step": 25204
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5965029619721004e-06,
+      "loss": 1.1189,
+      "step": 25205
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5964074144849995e-06,
+      "loss": 0.6247,
+      "step": 25206
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.596311866997898e-06,
+      "loss": 0.8061,
+      "step": 25207
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.596216319510797e-06,
+      "loss": 1.1905,
+      "step": 25208
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5961207720236957e-06,
+      "loss": 0.7654,
+      "step": 25209
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5960252245365948e-06,
+      "loss": 0.8279,
+      "step": 25210
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5959296770494942e-06,
+      "loss": 1.0803,
+      "step": 25211
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.595834129562393e-06,
+      "loss": 0.4243,
+      "step": 25212
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.595738582075292e-06,
+      "loss": 0.7104,
+      "step": 25213
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5956430345881905e-06,
+      "loss": 0.962,
+      "step": 25214
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5955474871010895e-06,
+      "loss": 0.77,
+      "step": 25215
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.595451939613988e-06,
+      "loss": 0.7387,
+      "step": 25216
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.595356392126887e-06,
+      "loss": 0.5925,
+      "step": 25217
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5952608446397866e-06,
+      "loss": 0.8622,
+      "step": 25218
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5951652971526853e-06,
+      "loss": 0.6661,
+      "step": 25219
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5950697496655843e-06,
+      "loss": 0.5066,
+      "step": 25220
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.594974202178483e-06,
+      "loss": 0.678,
+      "step": 25221
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.594878654691382e-06,
+      "loss": 0.5948,
+      "step": 25222
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5947831072042806e-06,
+      "loss": 0.9222,
+      "step": 25223
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5946875597171796e-06,
+      "loss": 0.8513,
+      "step": 25224
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5945920122300786e-06,
+      "loss": 1.3054,
+      "step": 25225
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5944964647429777e-06,
+      "loss": 0.696,
+      "step": 25226
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5944009172558763e-06,
+      "loss": 0.6734,
+      "step": 25227
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5943053697687753e-06,
+      "loss": 0.9103,
+      "step": 25228
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5942098222816744e-06,
+      "loss": 1.1364,
+      "step": 25229
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.594114274794573e-06,
+      "loss": 0.69,
+      "step": 25230
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.594018727307472e-06,
+      "loss": 0.8038,
+      "step": 25231
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5939231798203706e-06,
+      "loss": 0.6888,
+      "step": 25232
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.59382763233327e-06,
+      "loss": 0.515,
+      "step": 25233
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5937320848461687e-06,
+      "loss": 0.8832,
+      "step": 25234
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5936365373590678e-06,
+      "loss": 0.6563,
+      "step": 25235
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5935409898719664e-06,
+      "loss": 0.557,
+      "step": 25236
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5934454423848654e-06,
+      "loss": 0.5771,
+      "step": 25237
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.593349894897764e-06,
+      "loss": 0.9324,
+      "step": 25238
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.593254347410663e-06,
+      "loss": 1.1627,
+      "step": 25239
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5931587999235625e-06,
+      "loss": 0.7036,
+      "step": 25240
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.593063252436461e-06,
+      "loss": 0.9324,
+      "step": 25241
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.59296770494936e-06,
+      "loss": 0.8501,
+      "step": 25242
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.592872157462259e-06,
+      "loss": 0.9122,
+      "step": 25243
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.592776609975158e-06,
+      "loss": 0.6086,
+      "step": 25244
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5926810624880565e-06,
+      "loss": 0.8038,
+      "step": 25245
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5925855150009555e-06,
+      "loss": 1.082,
+      "step": 25246
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.592489967513855e-06,
+      "loss": 0.8296,
+      "step": 25247
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5923944200267536e-06,
+      "loss": 0.7259,
+      "step": 25248
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5922988725396526e-06,
+      "loss": 0.6986,
+      "step": 25249
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5922033250525512e-06,
+      "loss": 1.0748,
+      "step": 25250
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5921077775654503e-06,
+      "loss": 0.6922,
+      "step": 25251
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.592012230078349e-06,
+      "loss": 0.7509,
+      "step": 25252
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.591916682591248e-06,
+      "loss": 0.8691,
+      "step": 25253
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5918211351041474e-06,
+      "loss": 0.6772,
+      "step": 25254
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.591725587617046e-06,
+      "loss": 0.949,
+      "step": 25255
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.591630040129945e-06,
+      "loss": 0.6101,
+      "step": 25256
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5915344926428436e-06,
+      "loss": 0.7245,
+      "step": 25257
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5914389451557427e-06,
+      "loss": 0.8134,
+      "step": 25258
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5913433976686413e-06,
+      "loss": 0.7838,
+      "step": 25259
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5912478501815403e-06,
+      "loss": 0.9473,
+      "step": 25260
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.59115230269444e-06,
+      "loss": 0.8765,
+      "step": 25261
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5910567552073384e-06,
+      "loss": 0.9822,
+      "step": 25262
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5909612077202375e-06,
+      "loss": 0.8337,
+      "step": 25263
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.590865660233136e-06,
+      "loss": 0.4505,
+      "step": 25264
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.590770112746035e-06,
+      "loss": 0.4963,
+      "step": 25265
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5906745652589337e-06,
+      "loss": 0.5468,
+      "step": 25266
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5905790177718328e-06,
+      "loss": 0.8252,
+      "step": 25267
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5904834702847314e-06,
+      "loss": 0.9205,
+      "step": 25268
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.590387922797631e-06,
+      "loss": 0.7827,
+      "step": 25269
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.59029237531053e-06,
+      "loss": 1.1343,
+      "step": 25270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5901968278234285e-06,
+      "loss": 0.7628,
+      "step": 25271
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5901012803363275e-06,
+      "loss": 0.7732,
+      "step": 25272
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.590005732849226e-06,
+      "loss": 0.7761,
+      "step": 25273
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.589910185362125e-06,
+      "loss": 1.0614,
+      "step": 25274
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.589814637875024e-06,
+      "loss": 0.6489,
+      "step": 25275
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5897190903879233e-06,
+      "loss": 1.175,
+      "step": 25276
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.589623542900822e-06,
+      "loss": 0.9097,
+      "step": 25277
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.589527995413721e-06,
+      "loss": 0.5677,
+      "step": 25278
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5894324479266195e-06,
+      "loss": 0.663,
+      "step": 25279
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5893369004395186e-06,
+      "loss": 0.9228,
+      "step": 25280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5892413529524176e-06,
+      "loss": 0.7627,
+      "step": 25281
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5891458054653162e-06,
+      "loss": 0.7614,
+      "step": 25282
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5890502579782157e-06,
+      "loss": 0.5524,
+      "step": 25283
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5889547104911143e-06,
+      "loss": 0.5099,
+      "step": 25284
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5888591630040133e-06,
+      "loss": 0.6918,
+      "step": 25285
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.588763615516912e-06,
+      "loss": 0.9234,
+      "step": 25286
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.588668068029811e-06,
+      "loss": 0.6102,
+      "step": 25287
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5885725205427096e-06,
+      "loss": 0.4904,
+      "step": 25288
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5884769730556086e-06,
+      "loss": 0.6213,
+      "step": 25289
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.588381425568508e-06,
+      "loss": 0.98,
+      "step": 25290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5882858780814067e-06,
+      "loss": 1.2653,
+      "step": 25291
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5881903305943058e-06,
+      "loss": 0.9692,
+      "step": 25292
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5880947831072044e-06,
+      "loss": 0.708,
+      "step": 25293
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5879992356201034e-06,
+      "loss": 0.898,
+      "step": 25294
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587903688133002e-06,
+      "loss": 0.5712,
+      "step": 25295
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587808140645901e-06,
+      "loss": 1.1795,
+      "step": 25296
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5877125931588005e-06,
+      "loss": 0.788,
+      "step": 25297
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587617045671699e-06,
+      "loss": 0.6193,
+      "step": 25298
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587521498184598e-06,
+      "loss": 0.581,
+      "step": 25299
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587425950697497e-06,
+      "loss": 0.7673,
+      "step": 25300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587330403210396e-06,
+      "loss": 0.7222,
+      "step": 25301
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5872348557232945e-06,
+      "loss": 1.022,
+      "step": 25302
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5871393082361935e-06,
+      "loss": 0.8817,
+      "step": 25303
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587043760749092e-06,
+      "loss": 1.1637,
+      "step": 25304
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5869482132619916e-06,
+      "loss": 0.8677,
+      "step": 25305
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5868526657748906e-06,
+      "loss": 0.8092,
+      "step": 25306
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5867571182877892e-06,
+      "loss": 1.22,
+      "step": 25307
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5866615708006883e-06,
+      "loss": 0.7851,
+      "step": 25308
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.586566023313587e-06,
+      "loss": 0.6697,
+      "step": 25309
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.586470475826486e-06,
+      "loss": 1.351,
+      "step": 25310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5863749283393845e-06,
+      "loss": 0.9168,
+      "step": 25311
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.586279380852284e-06,
+      "loss": 0.8018,
+      "step": 25312
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.586183833365183e-06,
+      "loss": 0.5601,
+      "step": 25313
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5860882858780816e-06,
+      "loss": 0.6341,
+      "step": 25314
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5859927383909807e-06,
+      "loss": 1.2352,
+      "step": 25315
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5858971909038793e-06,
+      "loss": 0.6308,
+      "step": 25316
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5858016434167783e-06,
+      "loss": 0.7232,
+      "step": 25317
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585706095929677e-06,
+      "loss": 0.959,
+      "step": 25318
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5856105484425764e-06,
+      "loss": 0.7852,
+      "step": 25319
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5855150009554755e-06,
+      "loss": 0.6217,
+      "step": 25320
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585419453468374e-06,
+      "loss": 0.734,
+      "step": 25321
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585323905981273e-06,
+      "loss": 1.7025,
+      "step": 25322
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5852283584941717e-06,
+      "loss": 0.7404,
+      "step": 25323
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5851328110070708e-06,
+      "loss": 0.6614,
+      "step": 25324
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5850372635199694e-06,
+      "loss": 1.0041,
+      "step": 25325
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584941716032869e-06,
+      "loss": 0.5512,
+      "step": 25326
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5848461685457675e-06,
+      "loss": 0.8276,
+      "step": 25327
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5847506210586665e-06,
+      "loss": 0.7589,
+      "step": 25328
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584655073571565e-06,
+      "loss": 1.234,
+      "step": 25329
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584559526084464e-06,
+      "loss": 0.8377,
+      "step": 25330
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584463978597363e-06,
+      "loss": 0.984,
+      "step": 25331
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584368431110262e-06,
+      "loss": 0.7242,
+      "step": 25332
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5842728836231613e-06,
+      "loss": 0.9032,
+      "step": 25333
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58417733613606e-06,
+      "loss": 0.76,
+      "step": 25334
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.584081788648959e-06,
+      "loss": 0.7747,
+      "step": 25335
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5839862411618575e-06,
+      "loss": 0.7785,
+      "step": 25336
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5838906936747566e-06,
+      "loss": 0.7691,
+      "step": 25337
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.583795146187655e-06,
+      "loss": 0.4872,
+      "step": 25338
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5836995987005542e-06,
+      "loss": 0.6393,
+      "step": 25339
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.583604051213453e-06,
+      "loss": 1.1364,
+      "step": 25340
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5835085037263523e-06,
+      "loss": 0.8659,
+      "step": 25341
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5834129562392513e-06,
+      "loss": 0.8176,
+      "step": 25342
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58331740875215e-06,
+      "loss": 1.0137,
+      "step": 25343
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.583221861265049e-06,
+      "loss": 1.2287,
+      "step": 25344
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5831263137779476e-06,
+      "loss": 0.855,
+      "step": 25345
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5830307662908466e-06,
+      "loss": 0.8238,
+      "step": 25346
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5829352188037453e-06,
+      "loss": 0.7254,
+      "step": 25347
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5828396713166447e-06,
+      "loss": 1.0415,
+      "step": 25348
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5827441238295438e-06,
+      "loss": 0.7958,
+      "step": 25349
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5826485763424424e-06,
+      "loss": 0.8296,
+      "step": 25350
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5825530288553414e-06,
+      "loss": 0.7617,
+      "step": 25351
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58245748136824e-06,
+      "loss": 0.9329,
+      "step": 25352
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.582361933881139e-06,
+      "loss": 0.661,
+      "step": 25353
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5822663863940377e-06,
+      "loss": 1.0468,
+      "step": 25354
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.582170838906937e-06,
+      "loss": 1.2325,
+      "step": 25355
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.582075291419836e-06,
+      "loss": 0.6227,
+      "step": 25356
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.581979743932735e-06,
+      "loss": 0.697,
+      "step": 25357
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.581884196445634e-06,
+      "loss": 0.5806,
+      "step": 25358
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5817886489585325e-06,
+      "loss": 0.9261,
+      "step": 25359
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5816931014714315e-06,
+      "loss": 1.155,
+      "step": 25360
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58159755398433e-06,
+      "loss": 0.9719,
+      "step": 25361
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5815020064972296e-06,
+      "loss": 0.5123,
+      "step": 25362
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5814064590101286e-06,
+      "loss": 1.1761,
+      "step": 25363
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5813109115230272e-06,
+      "loss": 0.8,
+      "step": 25364
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5812153640359263e-06,
+      "loss": 0.6876,
+      "step": 25365
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.581119816548825e-06,
+      "loss": 0.4534,
+      "step": 25366
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.581024269061724e-06,
+      "loss": 1.0098,
+      "step": 25367
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5809287215746225e-06,
+      "loss": 1.0592,
+      "step": 25368
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580833174087522e-06,
+      "loss": 0.7751,
+      "step": 25369
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580737626600421e-06,
+      "loss": 0.5913,
+      "step": 25370
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5806420791133196e-06,
+      "loss": 0.9848,
+      "step": 25371
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5805465316262187e-06,
+      "loss": 0.5968,
+      "step": 25372
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5804509841391173e-06,
+      "loss": 0.5137,
+      "step": 25373
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5803554366520163e-06,
+      "loss": 0.4456,
+      "step": 25374
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580259889164915e-06,
+      "loss": 0.5844,
+      "step": 25375
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580164341677814e-06,
+      "loss": 1.099,
+      "step": 25376
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580068794190713e-06,
+      "loss": 0.7766,
+      "step": 25377
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.579973246703612e-06,
+      "loss": 0.775,
+      "step": 25378
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5798776992165107e-06,
+      "loss": 1.1044,
+      "step": 25379
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5797821517294097e-06,
+      "loss": 0.6361,
+      "step": 25380
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5796866042423083e-06,
+      "loss": 0.6981,
+      "step": 25381
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5795910567552074e-06,
+      "loss": 0.5962,
+      "step": 25382
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5794955092681064e-06,
+      "loss": 0.5167,
+      "step": 25383
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5793999617810055e-06,
+      "loss": 1.2797,
+      "step": 25384
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5793044142939045e-06,
+      "loss": 0.9122,
+      "step": 25385
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.579208866806803e-06,
+      "loss": 0.4291,
+      "step": 25386
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.579113319319702e-06,
+      "loss": 0.5192,
+      "step": 25387
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5790177718326008e-06,
+      "loss": 0.9339,
+      "step": 25388
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5789222243455e-06,
+      "loss": 0.6704,
+      "step": 25389
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5788266768583984e-06,
+      "loss": 1.2763,
+      "step": 25390
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.578731129371298e-06,
+      "loss": 0.9734,
+      "step": 25391
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.578635581884197e-06,
+      "loss": 0.601,
+      "step": 25392
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5785400343970955e-06,
+      "loss": 0.852,
+      "step": 25393
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5784444869099946e-06,
+      "loss": 0.8297,
+      "step": 25394
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.578348939422893e-06,
+      "loss": 1.1422,
+      "step": 25395
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5782533919357922e-06,
+      "loss": 0.9254,
+      "step": 25396
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.578157844448691e-06,
+      "loss": 0.7968,
+      "step": 25397
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5780622969615903e-06,
+      "loss": 0.4139,
+      "step": 25398
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5779667494744893e-06,
+      "loss": 0.9352,
+      "step": 25399
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.577871201987388e-06,
+      "loss": 0.9374,
+      "step": 25400
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.577775654500287e-06,
+      "loss": 0.4405,
+      "step": 25401
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5776801070131856e-06,
+      "loss": 0.8189,
+      "step": 25402
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5775845595260846e-06,
+      "loss": 0.9997,
+      "step": 25403
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5774890120389833e-06,
+      "loss": 0.7761,
+      "step": 25404
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5773934645518827e-06,
+      "loss": 0.4712,
+      "step": 25405
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5772979170647818e-06,
+      "loss": 0.982,
+      "step": 25406
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5772023695776804e-06,
+      "loss": 0.7803,
+      "step": 25407
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5771068220905794e-06,
+      "loss": 0.807,
+      "step": 25408
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.577011274603478e-06,
+      "loss": 1.0767,
+      "step": 25409
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576915727116377e-06,
+      "loss": 0.9551,
+      "step": 25410
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5768201796292757e-06,
+      "loss": 0.8596,
+      "step": 25411
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5767246321421747e-06,
+      "loss": 0.6139,
+      "step": 25412
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576629084655074e-06,
+      "loss": 0.5905,
+      "step": 25413
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576533537167973e-06,
+      "loss": 0.8653,
+      "step": 25414
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576437989680872e-06,
+      "loss": 1.0075,
+      "step": 25415
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5763424421937705e-06,
+      "loss": 0.8027,
+      "step": 25416
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5762468947066695e-06,
+      "loss": 0.6678,
+      "step": 25417
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576151347219568e-06,
+      "loss": 0.6359,
+      "step": 25418
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576055799732467e-06,
+      "loss": 1.0074,
+      "step": 25419
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.575960252245366e-06,
+      "loss": 0.9947,
+      "step": 25420
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5758647047582652e-06,
+      "loss": 0.7924,
+      "step": 25421
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5757691572711643e-06,
+      "loss": 0.6095,
+      "step": 25422
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.575673609784063e-06,
+      "loss": 0.7623,
+      "step": 25423
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.575578062296962e-06,
+      "loss": 0.7997,
+      "step": 25424
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5754825148098605e-06,
+      "loss": 0.9335,
+      "step": 25425
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5753869673227596e-06,
+      "loss": 0.8901,
+      "step": 25426
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5752914198356586e-06,
+      "loss": 1.1716,
+      "step": 25427
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5751958723485576e-06,
+      "loss": 0.9454,
+      "step": 25428
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5751003248614563e-06,
+      "loss": 0.9113,
+      "step": 25429
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5750047773743553e-06,
+      "loss": 0.7015,
+      "step": 25430
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574909229887254e-06,
+      "loss": 0.8841,
+      "step": 25431
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574813682400153e-06,
+      "loss": 0.5816,
+      "step": 25432
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574718134913052e-06,
+      "loss": 0.5208,
+      "step": 25433
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574622587425951e-06,
+      "loss": 0.7891,
+      "step": 25434
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.57452703993885e-06,
+      "loss": 1.2358,
+      "step": 25435
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5744314924517487e-06,
+      "loss": 0.5808,
+      "step": 25436
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5743359449646477e-06,
+      "loss": 0.5309,
+      "step": 25437
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5742403974775463e-06,
+      "loss": 0.5948,
+      "step": 25438
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5741448499904454e-06,
+      "loss": 1.1407,
+      "step": 25439
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.574049302503344e-06,
+      "loss": 1.0152,
+      "step": 25440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5739537550162435e-06,
+      "loss": 0.5652,
+      "step": 25441
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5738582075291425e-06,
+      "loss": 0.9172,
+      "step": 25442
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.573762660042041e-06,
+      "loss": 0.8939,
+      "step": 25443
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.57366711255494e-06,
+      "loss": 0.3476,
+      "step": 25444
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5735715650678388e-06,
+      "loss": 0.6311,
+      "step": 25445
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.573476017580738e-06,
+      "loss": 0.7121,
+      "step": 25446
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5733804700936364e-06,
+      "loss": 0.5686,
+      "step": 25447
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5732849226065355e-06,
+      "loss": 0.476,
+      "step": 25448
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.573189375119435e-06,
+      "loss": 0.4727,
+      "step": 25449
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5730938276323335e-06,
+      "loss": 1.0795,
+      "step": 25450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5729982801452326e-06,
+      "loss": 0.8747,
+      "step": 25451
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.572902732658131e-06,
+      "loss": 0.8387,
+      "step": 25452
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5728071851710302e-06,
+      "loss": 0.7596,
+      "step": 25453
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.572711637683929e-06,
+      "loss": 0.9763,
+      "step": 25454
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.572616090196828e-06,
+      "loss": 0.8894,
+      "step": 25455
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5725205427097273e-06,
+      "loss": 0.693,
+      "step": 25456
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.572424995222626e-06,
+      "loss": 0.8407,
+      "step": 25457
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.572329447735525e-06,
+      "loss": 0.2335,
+      "step": 25458
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5722339002484236e-06,
+      "loss": 0.926,
+      "step": 25459
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5721383527613226e-06,
+      "loss": 0.7869,
+      "step": 25460
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5720428052742213e-06,
+      "loss": 0.9879,
+      "step": 25461
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5719472577871203e-06,
+      "loss": 0.5611,
+      "step": 25462
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5718517103000198e-06,
+      "loss": 1.1301,
+      "step": 25463
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5717561628129184e-06,
+      "loss": 0.6042,
+      "step": 25464
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5716606153258174e-06,
+      "loss": 0.8533,
+      "step": 25465
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.571565067838716e-06,
+      "loss": 0.5744,
+      "step": 25466
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.571469520351615e-06,
+      "loss": 0.9987,
+      "step": 25467
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5713739728645137e-06,
+      "loss": 1.1596,
+      "step": 25468
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5712784253774127e-06,
+      "loss": 0.7247,
+      "step": 25469
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5711828778903118e-06,
+      "loss": 0.6159,
+      "step": 25470
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.571087330403211e-06,
+      "loss": 0.509,
+      "step": 25471
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.57099178291611e-06,
+      "loss": 0.6043,
+      "step": 25472
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5708962354290084e-06,
+      "loss": 0.8108,
+      "step": 25473
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5708006879419075e-06,
+      "loss": 1.0339,
+      "step": 25474
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.570705140454806e-06,
+      "loss": 0.4769,
+      "step": 25475
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.570609592967705e-06,
+      "loss": 0.8217,
+      "step": 25476
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.570514045480604e-06,
+      "loss": 0.8257,
+      "step": 25477
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5704184979935032e-06,
+      "loss": 0.7521,
+      "step": 25478
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.570322950506402e-06,
+      "loss": 0.9712,
+      "step": 25479
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.570227403019301e-06,
+      "loss": 0.7426,
+      "step": 25480
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5701318555321995e-06,
+      "loss": 0.5238,
+      "step": 25481
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5700363080450985e-06,
+      "loss": 0.6351,
+      "step": 25482
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569940760557997e-06,
+      "loss": 0.6066,
+      "step": 25483
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569845213070896e-06,
+      "loss": 0.862,
+      "step": 25484
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5697496655837956e-06,
+      "loss": 1.0135,
+      "step": 25485
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5696541180966943e-06,
+      "loss": 0.5104,
+      "step": 25486
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5695585706095933e-06,
+      "loss": 0.4017,
+      "step": 25487
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569463023122492e-06,
+      "loss": 0.8034,
+      "step": 25488
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569367475635391e-06,
+      "loss": 0.7498,
+      "step": 25489
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5692719281482896e-06,
+      "loss": 0.7762,
+      "step": 25490
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5691763806611886e-06,
+      "loss": 0.53,
+      "step": 25491
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.569080833174088e-06,
+      "loss": 0.7256,
+      "step": 25492
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5689852856869867e-06,
+      "loss": 0.6854,
+      "step": 25493
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5688897381998857e-06,
+      "loss": 0.9537,
+      "step": 25494
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5687941907127843e-06,
+      "loss": 0.8729,
+      "step": 25495
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5686986432256834e-06,
+      "loss": 0.9271,
+      "step": 25496
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568603095738582e-06,
+      "loss": 0.7862,
+      "step": 25497
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568507548251481e-06,
+      "loss": 0.6433,
+      "step": 25498
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5684120007643805e-06,
+      "loss": 0.4263,
+      "step": 25499
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568316453277279e-06,
+      "loss": 0.8338,
+      "step": 25500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568220905790178e-06,
+      "loss": 0.7165,
+      "step": 25501
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5681253583030768e-06,
+      "loss": 0.7891,
+      "step": 25502
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.568029810815976e-06,
+      "loss": 1.1816,
+      "step": 25503
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5679342633288744e-06,
+      "loss": 0.811,
+      "step": 25504
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5678387158417734e-06,
+      "loss": 0.4461,
+      "step": 25505
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.567743168354673e-06,
+      "loss": 0.8127,
+      "step": 25506
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5676476208675715e-06,
+      "loss": 1.234,
+      "step": 25507
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5675520733804706e-06,
+      "loss": 0.9421,
+      "step": 25508
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.567456525893369e-06,
+      "loss": 0.5835,
+      "step": 25509
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5673609784062682e-06,
+      "loss": 0.734,
+      "step": 25510
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.567265430919167e-06,
+      "loss": 0.5542,
+      "step": 25511
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.567169883432066e-06,
+      "loss": 0.544,
+      "step": 25512
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5670743359449653e-06,
+      "loss": 0.5034,
+      "step": 25513
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566978788457864e-06,
+      "loss": 1.0265,
+      "step": 25514
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566883240970763e-06,
+      "loss": 0.6188,
+      "step": 25515
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5667876934836616e-06,
+      "loss": 0.7308,
+      "step": 25516
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5666921459965606e-06,
+      "loss": 0.837,
+      "step": 25517
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5665965985094593e-06,
+      "loss": 0.4447,
+      "step": 25518
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5665010510223583e-06,
+      "loss": 1.2065,
+      "step": 25519
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566405503535257e-06,
+      "loss": 1.0722,
+      "step": 25520
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5663099560481564e-06,
+      "loss": 1.1929,
+      "step": 25521
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566214408561055e-06,
+      "loss": 1.0634,
+      "step": 25522
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566118861073954e-06,
+      "loss": 1.1557,
+      "step": 25523
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.566023313586853e-06,
+      "loss": 0.5228,
+      "step": 25524
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5659277660997517e-06,
+      "loss": 0.5867,
+      "step": 25525
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5658322186126507e-06,
+      "loss": 0.8295,
+      "step": 25526
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5657366711255493e-06,
+      "loss": 0.7544,
+      "step": 25527
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.565641123638449e-06,
+      "loss": 0.5756,
+      "step": 25528
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5655455761513474e-06,
+      "loss": 0.5663,
+      "step": 25529
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5654500286642464e-06,
+      "loss": 1.12,
+      "step": 25530
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.565354481177145e-06,
+      "loss": 0.528,
+      "step": 25531
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.565258933690044e-06,
+      "loss": 1.0652,
+      "step": 25532
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5651633862029427e-06,
+      "loss": 1.1724,
+      "step": 25533
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5650678387158418e-06,
+      "loss": 1.1475,
+      "step": 25534
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5649722912287412e-06,
+      "loss": 0.7156,
+      "step": 25535
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.56487674374164e-06,
+      "loss": 0.9205,
+      "step": 25536
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.564781196254539e-06,
+      "loss": 1.2353,
+      "step": 25537
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5646856487674375e-06,
+      "loss": 0.8357,
+      "step": 25538
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5645901012803365e-06,
+      "loss": 1.266,
+      "step": 25539
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.564494553793235e-06,
+      "loss": 1.0344,
+      "step": 25540
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.564399006306134e-06,
+      "loss": 0.9667,
+      "step": 25541
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5643034588190336e-06,
+      "loss": 0.9322,
+      "step": 25542
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5642079113319323e-06,
+      "loss": 0.6322,
+      "step": 25543
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5641123638448313e-06,
+      "loss": 0.7005,
+      "step": 25544
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.56401681635773e-06,
+      "loss": 1.0404,
+      "step": 25545
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.563921268870629e-06,
+      "loss": 0.8845,
+      "step": 25546
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5638257213835276e-06,
+      "loss": 0.8591,
+      "step": 25547
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5637301738964266e-06,
+      "loss": 0.3755,
+      "step": 25548
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.563634626409326e-06,
+      "loss": 1.5382,
+      "step": 25549
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5635390789222247e-06,
+      "loss": 1.0566,
+      "step": 25550
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5634435314351237e-06,
+      "loss": 0.9867,
+      "step": 25551
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5633479839480223e-06,
+      "loss": 0.5063,
+      "step": 25552
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5632524364609214e-06,
+      "loss": 0.6414,
+      "step": 25553
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.56315688897382e-06,
+      "loss": 0.7697,
+      "step": 25554
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.563061341486719e-06,
+      "loss": 1.1645,
+      "step": 25555
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5629657939996176e-06,
+      "loss": 0.5447,
+      "step": 25556
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.562870246512517e-06,
+      "loss": 0.892,
+      "step": 25557
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.562774699025416e-06,
+      "loss": 1.2672,
+      "step": 25558
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5626791515383148e-06,
+      "loss": 0.7308,
+      "step": 25559
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.562583604051214e-06,
+      "loss": 1.0119,
+      "step": 25560
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5624880565641124e-06,
+      "loss": 1.3562,
+      "step": 25561
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5623925090770114e-06,
+      "loss": 0.7036,
+      "step": 25562
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.56229696158991e-06,
+      "loss": 0.9905,
+      "step": 25563
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5622014141028095e-06,
+      "loss": 1.4029,
+      "step": 25564
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5621058666157086e-06,
+      "loss": 0.8542,
+      "step": 25565
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.562010319128607e-06,
+      "loss": 0.5203,
+      "step": 25566
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5619147716415062e-06,
+      "loss": 1.3306,
+      "step": 25567
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.561819224154405e-06,
+      "loss": 0.909,
+      "step": 25568
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.561723676667304e-06,
+      "loss": 0.913,
+      "step": 25569
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5616281291802025e-06,
+      "loss": 0.5448,
+      "step": 25570
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.561532581693102e-06,
+      "loss": 0.5568,
+      "step": 25571
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5614370342060006e-06,
+      "loss": 1.0327,
+      "step": 25572
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5613414867188996e-06,
+      "loss": 0.6464,
+      "step": 25573
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5612459392317982e-06,
+      "loss": 0.4894,
+      "step": 25574
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5611503917446973e-06,
+      "loss": 0.8273,
+      "step": 25575
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5610548442575963e-06,
+      "loss": 0.5079,
+      "step": 25576
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560959296770495e-06,
+      "loss": 0.6988,
+      "step": 25577
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5608637492833944e-06,
+      "loss": 0.9366,
+      "step": 25578
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560768201796293e-06,
+      "loss": 1.3493,
+      "step": 25579
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560672654309192e-06,
+      "loss": 0.7519,
+      "step": 25580
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5605771068220906e-06,
+      "loss": 1.1333,
+      "step": 25581
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5604815593349897e-06,
+      "loss": 0.6226,
+      "step": 25582
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5603860118478883e-06,
+      "loss": 0.6389,
+      "step": 25583
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5602904643607873e-06,
+      "loss": 0.9408,
+      "step": 25584
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560194916873687e-06,
+      "loss": 0.7554,
+      "step": 25585
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5600993693865854e-06,
+      "loss": 0.8462,
+      "step": 25586
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5600038218994844e-06,
+      "loss": 0.6471,
+      "step": 25587
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.559908274412383e-06,
+      "loss": 0.7569,
+      "step": 25588
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.559812726925282e-06,
+      "loss": 0.9534,
+      "step": 25589
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5597171794381807e-06,
+      "loss": 0.7942,
+      "step": 25590
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5596216319510798e-06,
+      "loss": 0.8745,
+      "step": 25591
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5595260844639784e-06,
+      "loss": 0.9847,
+      "step": 25592
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.559430536976878e-06,
+      "loss": 0.6135,
+      "step": 25593
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.559334989489777e-06,
+      "loss": 1.1609,
+      "step": 25594
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5592394420026755e-06,
+      "loss": 0.7608,
+      "step": 25595
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5591438945155745e-06,
+      "loss": 0.889,
+      "step": 25596
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.559048347028473e-06,
+      "loss": 0.7274,
+      "step": 25597
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.558952799541372e-06,
+      "loss": 0.96,
+      "step": 25598
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.558857252054271e-06,
+      "loss": 0.9782,
+      "step": 25599
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5587617045671703e-06,
+      "loss": 0.4617,
+      "step": 25600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5586661570800693e-06,
+      "loss": 0.6552,
+      "step": 25601
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.558570609592968e-06,
+      "loss": 0.6253,
+      "step": 25602
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.558475062105867e-06,
+      "loss": 0.749,
+      "step": 25603
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5583795146187656e-06,
+      "loss": 1.1979,
+      "step": 25604
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5582839671316646e-06,
+      "loss": 1.0133,
+      "step": 25605
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5581884196445632e-06,
+      "loss": 0.7271,
+      "step": 25606
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5580928721574627e-06,
+      "loss": 0.7918,
+      "step": 25607
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5579973246703617e-06,
+      "loss": 0.5992,
+      "step": 25608
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5579017771832603e-06,
+      "loss": 0.9613,
+      "step": 25609
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5578062296961594e-06,
+      "loss": 1.0229,
+      "step": 25610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.557710682209058e-06,
+      "loss": 0.452,
+      "step": 25611
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.557615134721957e-06,
+      "loss": 0.7587,
+      "step": 25612
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5575195872348556e-06,
+      "loss": 0.9905,
+      "step": 25613
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.557424039747755e-06,
+      "loss": 1.0284,
+      "step": 25614
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.557328492260654e-06,
+      "loss": 0.8688,
+      "step": 25615
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5572329447735528e-06,
+      "loss": 0.801,
+      "step": 25616
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.557137397286452e-06,
+      "loss": 0.8789,
+      "step": 25617
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5570418497993504e-06,
+      "loss": 0.6664,
+      "step": 25618
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5569463023122494e-06,
+      "loss": 0.6604,
+      "step": 25619
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556850754825148e-06,
+      "loss": 1.3393,
+      "step": 25620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5567552073380475e-06,
+      "loss": 0.9144,
+      "step": 25621
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556659659850946e-06,
+      "loss": 0.7269,
+      "step": 25622
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556564112363845e-06,
+      "loss": 0.7037,
+      "step": 25623
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556468564876744e-06,
+      "loss": 0.8889,
+      "step": 25624
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556373017389643e-06,
+      "loss": 0.6674,
+      "step": 25625
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.556277469902542e-06,
+      "loss": 1.1951,
+      "step": 25626
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5561819224154405e-06,
+      "loss": 0.9016,
+      "step": 25627
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5560863749283395e-06,
+      "loss": 0.9862,
+      "step": 25628
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5559908274412386e-06,
+      "loss": 0.7401,
+      "step": 25629
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5558952799541376e-06,
+      "loss": 1.3842,
+      "step": 25630
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5557997324670362e-06,
+      "loss": 1.1717,
+      "step": 25631
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5557041849799353e-06,
+      "loss": 0.6975,
+      "step": 25632
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.555608637492834e-06,
+      "loss": 0.9744,
+      "step": 25633
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.555513090005733e-06,
+      "loss": 0.9265,
+      "step": 25634
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5554175425186315e-06,
+      "loss": 0.8169,
+      "step": 25635
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.555321995031531e-06,
+      "loss": 1.2013,
+      "step": 25636
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.55522644754443e-06,
+      "loss": 1.1486,
+      "step": 25637
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5551309000573286e-06,
+      "loss": 0.6284,
+      "step": 25638
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5550353525702277e-06,
+      "loss": 0.4825,
+      "step": 25639
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5549398050831263e-06,
+      "loss": 0.9254,
+      "step": 25640
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5548442575960253e-06,
+      "loss": 1.1529,
+      "step": 25641
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.554748710108924e-06,
+      "loss": 0.8896,
+      "step": 25642
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5546531626218234e-06,
+      "loss": 0.605,
+      "step": 25643
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5545576151347224e-06,
+      "loss": 1.1083,
+      "step": 25644
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.554462067647621e-06,
+      "loss": 0.8546,
+      "step": 25645
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.55436652016052e-06,
+      "loss": 0.7393,
+      "step": 25646
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5542709726734187e-06,
+      "loss": 0.8085,
+      "step": 25647
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5541754251863178e-06,
+      "loss": 0.7996,
+      "step": 25648
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5540798776992164e-06,
+      "loss": 0.9682,
+      "step": 25649
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553984330212116e-06,
+      "loss": 0.5762,
+      "step": 25650
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553888782725015e-06,
+      "loss": 0.9623,
+      "step": 25651
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5537932352379135e-06,
+      "loss": 0.6757,
+      "step": 25652
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5536976877508125e-06,
+      "loss": 0.8104,
+      "step": 25653
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553602140263711e-06,
+      "loss": 0.925,
+      "step": 25654
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.55350659277661e-06,
+      "loss": 0.732,
+      "step": 25655
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553411045289509e-06,
+      "loss": 0.7624,
+      "step": 25656
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5533154978024083e-06,
+      "loss": 0.6659,
+      "step": 25657
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5532199503153073e-06,
+      "loss": 0.8867,
+      "step": 25658
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553124402828206e-06,
+      "loss": 0.966,
+      "step": 25659
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.553028855341105e-06,
+      "loss": 1.2683,
+      "step": 25660
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5529333078540036e-06,
+      "loss": 0.6007,
+      "step": 25661
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5528377603669026e-06,
+      "loss": 0.5899,
+      "step": 25662
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5527422128798012e-06,
+      "loss": 1.1847,
+      "step": 25663
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5526466653927003e-06,
+      "loss": 1.0155,
+      "step": 25664
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5525511179055997e-06,
+      "loss": 1.0966,
+      "step": 25665
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5524555704184983e-06,
+      "loss": 0.8771,
+      "step": 25666
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5523600229313974e-06,
+      "loss": 0.476,
+      "step": 25667
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.552264475444296e-06,
+      "loss": 0.8218,
+      "step": 25668
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.552168927957195e-06,
+      "loss": 0.7488,
+      "step": 25669
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5520733804700936e-06,
+      "loss": 0.6863,
+      "step": 25670
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5519778329829927e-06,
+      "loss": 0.8755,
+      "step": 25671
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5518822854958917e-06,
+      "loss": 1.2333,
+      "step": 25672
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5517867380087908e-06,
+      "loss": 0.3972,
+      "step": 25673
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5516911905216894e-06,
+      "loss": 0.6449,
+      "step": 25674
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5515956430345884e-06,
+      "loss": 0.7776,
+      "step": 25675
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551500095547487e-06,
+      "loss": 0.544,
+      "step": 25676
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551404548060386e-06,
+      "loss": 1.1738,
+      "step": 25677
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551309000573285e-06,
+      "loss": 1.2818,
+      "step": 25678
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551213453086184e-06,
+      "loss": 0.6939,
+      "step": 25679
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551117905599083e-06,
+      "loss": 0.8697,
+      "step": 25680
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551022358111982e-06,
+      "loss": 0.6671,
+      "step": 25681
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550926810624881e-06,
+      "loss": 0.594,
+      "step": 25682
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5508312631377794e-06,
+      "loss": 0.9198,
+      "step": 25683
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5507357156506785e-06,
+      "loss": 0.9309,
+      "step": 25684
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550640168163577e-06,
+      "loss": 1.1169,
+      "step": 25685
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5505446206764766e-06,
+      "loss": 0.8023,
+      "step": 25686
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5504490731893756e-06,
+      "loss": 1.1595,
+      "step": 25687
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5503535257022742e-06,
+      "loss": 1.1614,
+      "step": 25688
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5502579782151733e-06,
+      "loss": 1.0254,
+      "step": 25689
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550162430728072e-06,
+      "loss": 0.581,
+      "step": 25690
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550066883240971e-06,
+      "loss": 0.6822,
+      "step": 25691
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5499713357538695e-06,
+      "loss": 0.9109,
+      "step": 25692
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549875788266769e-06,
+      "loss": 0.7566,
+      "step": 25693
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549780240779668e-06,
+      "loss": 1.0083,
+      "step": 25694
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5496846932925666e-06,
+      "loss": 0.7392,
+      "step": 25695
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5495891458054657e-06,
+      "loss": 0.5249,
+      "step": 25696
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5494935983183643e-06,
+      "loss": 0.667,
+      "step": 25697
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5493980508312633e-06,
+      "loss": 0.7087,
+      "step": 25698
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549302503344162e-06,
+      "loss": 0.8887,
+      "step": 25699
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549206955857061e-06,
+      "loss": 1.0005,
+      "step": 25700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5491114083699604e-06,
+      "loss": 0.8984,
+      "step": 25701
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.549015860882859e-06,
+      "loss": 0.7938,
+      "step": 25702
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548920313395758e-06,
+      "loss": 0.8139,
+      "step": 25703
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5488247659086567e-06,
+      "loss": 0.8674,
+      "step": 25704
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5487292184215558e-06,
+      "loss": 0.7476,
+      "step": 25705
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5486336709344544e-06,
+      "loss": 0.6752,
+      "step": 25706
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5485381234473534e-06,
+      "loss": 0.8984,
+      "step": 25707
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548442575960253e-06,
+      "loss": 0.8762,
+      "step": 25708
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5483470284731515e-06,
+      "loss": 0.6712,
+      "step": 25709
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5482514809860505e-06,
+      "loss": 1.0911,
+      "step": 25710
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548155933498949e-06,
+      "loss": 0.3387,
+      "step": 25711
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548060386011848e-06,
+      "loss": 0.8456,
+      "step": 25712
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547964838524747e-06,
+      "loss": 0.8072,
+      "step": 25713
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547869291037646e-06,
+      "loss": 1.0456,
+      "step": 25714
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547773743550545e-06,
+      "loss": 0.7672,
+      "step": 25715
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547678196063444e-06,
+      "loss": 0.7964,
+      "step": 25716
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547582648576343e-06,
+      "loss": 0.6448,
+      "step": 25717
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5474871010892416e-06,
+      "loss": 1.5601,
+      "step": 25718
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5473915536021406e-06,
+      "loss": 0.5608,
+      "step": 25719
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5472960061150392e-06,
+      "loss": 0.6846,
+      "step": 25720
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5472004586279383e-06,
+      "loss": 0.7203,
+      "step": 25721
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5471049111408373e-06,
+      "loss": 0.7412,
+      "step": 25722
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5470093636537363e-06,
+      "loss": 0.8985,
+      "step": 25723
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.546913816166635e-06,
+      "loss": 0.6517,
+      "step": 25724
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.546818268679534e-06,
+      "loss": 1.0556,
+      "step": 25725
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5467227211924326e-06,
+      "loss": 1.0497,
+      "step": 25726
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5466271737053316e-06,
+      "loss": 1.0227,
+      "step": 25727
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5465316262182307e-06,
+      "loss": 0.762,
+      "step": 25728
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5464360787311297e-06,
+      "loss": 0.6149,
+      "step": 25729
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5463405312440288e-06,
+      "loss": 0.8287,
+      "step": 25730
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5462449837569274e-06,
+      "loss": 0.8133,
+      "step": 25731
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5461494362698264e-06,
+      "loss": 0.7348,
+      "step": 25732
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.546053888782725e-06,
+      "loss": 0.9994,
+      "step": 25733
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545958341295624e-06,
+      "loss": 0.9191,
+      "step": 25734
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5458627938085227e-06,
+      "loss": 0.5091,
+      "step": 25735
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5457672463214217e-06,
+      "loss": 0.6079,
+      "step": 25736
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545671698834321e-06,
+      "loss": 1.0559,
+      "step": 25737
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.54557615134722e-06,
+      "loss": 0.6563,
+      "step": 25738
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545480603860119e-06,
+      "loss": 1.0704,
+      "step": 25739
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5453850563730174e-06,
+      "loss": 0.835,
+      "step": 25740
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5452895088859165e-06,
+      "loss": 0.9922,
+      "step": 25741
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545193961398815e-06,
+      "loss": 1.3164,
+      "step": 25742
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545098413911714e-06,
+      "loss": 0.8952,
+      "step": 25743
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5450028664246136e-06,
+      "loss": 0.7682,
+      "step": 25744
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5449073189375122e-06,
+      "loss": 1.5149,
+      "step": 25745
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5448117714504113e-06,
+      "loss": 0.8818,
+      "step": 25746
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.54471622396331e-06,
+      "loss": 1.1228,
+      "step": 25747
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.544620676476209e-06,
+      "loss": 0.6518,
+      "step": 25748
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5445251289891075e-06,
+      "loss": 0.6295,
+      "step": 25749
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5444295815020066e-06,
+      "loss": 0.8646,
+      "step": 25750
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.544334034014906e-06,
+      "loss": 0.5062,
+      "step": 25751
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5442384865278046e-06,
+      "loss": 0.7283,
+      "step": 25752
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5441429390407037e-06,
+      "loss": 0.9485,
+      "step": 25753
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5440473915536023e-06,
+      "loss": 0.834,
+      "step": 25754
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5439518440665013e-06,
+      "loss": 1.2261,
+      "step": 25755
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5438562965794e-06,
+      "loss": 0.7798,
+      "step": 25756
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.543760749092299e-06,
+      "loss": 0.6676,
+      "step": 25757
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5436652016051984e-06,
+      "loss": 0.8191,
+      "step": 25758
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.543569654118097e-06,
+      "loss": 0.5829,
+      "step": 25759
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.543474106630996e-06,
+      "loss": 0.9465,
+      "step": 25760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5433785591438947e-06,
+      "loss": 0.4008,
+      "step": 25761
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5432830116567938e-06,
+      "loss": 0.929,
+      "step": 25762
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5431874641696924e-06,
+      "loss": 0.9095,
+      "step": 25763
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5430919166825914e-06,
+      "loss": 0.5083,
+      "step": 25764
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5429963691954904e-06,
+      "loss": 0.9644,
+      "step": 25765
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5429008217083895e-06,
+      "loss": 0.4712,
+      "step": 25766
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5428052742212885e-06,
+      "loss": 0.8046,
+      "step": 25767
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.542709726734187e-06,
+      "loss": 0.9897,
+      "step": 25768
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.542614179247086e-06,
+      "loss": 0.7332,
+      "step": 25769
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.542518631759985e-06,
+      "loss": 0.9866,
+      "step": 25770
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.542423084272884e-06,
+      "loss": 0.8413,
+      "step": 25771
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5423275367857824e-06,
+      "loss": 0.8141,
+      "step": 25772
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.542231989298682e-06,
+      "loss": 0.9871,
+      "step": 25773
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5421364418115805e-06,
+      "loss": 0.93,
+      "step": 25774
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5420408943244796e-06,
+      "loss": 0.9413,
+      "step": 25775
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541945346837378e-06,
+      "loss": 1.1088,
+      "step": 25776
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5418497993502772e-06,
+      "loss": 0.7499,
+      "step": 25777
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541754251863176e-06,
+      "loss": 0.6729,
+      "step": 25778
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541658704376075e-06,
+      "loss": 0.6281,
+      "step": 25779
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5415631568889743e-06,
+      "loss": 0.5863,
+      "step": 25780
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541467609401873e-06,
+      "loss": 0.8868,
+      "step": 25781
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541372061914772e-06,
+      "loss": 0.7561,
+      "step": 25782
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5412765144276706e-06,
+      "loss": 0.6401,
+      "step": 25783
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5411809669405696e-06,
+      "loss": 0.9499,
+      "step": 25784
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5410854194534683e-06,
+      "loss": 0.5634,
+      "step": 25785
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5409898719663673e-06,
+      "loss": 0.4619,
+      "step": 25786
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5408943244792668e-06,
+      "loss": 0.369,
+      "step": 25787
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5407987769921654e-06,
+      "loss": 0.7145,
+      "step": 25788
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5407032295050644e-06,
+      "loss": 0.7769,
+      "step": 25789
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.540607682017963e-06,
+      "loss": 0.6709,
+      "step": 25790
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.540512134530862e-06,
+      "loss": 0.999,
+      "step": 25791
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5404165870437607e-06,
+      "loss": 1.2173,
+      "step": 25792
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5403210395566597e-06,
+      "loss": 0.8339,
+      "step": 25793
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.540225492069559e-06,
+      "loss": 0.5501,
+      "step": 25794
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.540129944582458e-06,
+      "loss": 0.6269,
+      "step": 25795
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.540034397095357e-06,
+      "loss": 1.2865,
+      "step": 25796
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5399388496082554e-06,
+      "loss": 0.7739,
+      "step": 25797
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5398433021211545e-06,
+      "loss": 0.6193,
+      "step": 25798
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.539747754634053e-06,
+      "loss": 0.6739,
+      "step": 25799
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.539652207146952e-06,
+      "loss": 0.8641,
+      "step": 25800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5395566596598516e-06,
+      "loss": 0.955,
+      "step": 25801
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5394611121727502e-06,
+      "loss": 0.7124,
+      "step": 25802
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5393655646856493e-06,
+      "loss": 1.6065,
+      "step": 25803
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.539270017198548e-06,
+      "loss": 0.5886,
+      "step": 25804
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.539174469711447e-06,
+      "loss": 0.5972,
+      "step": 25805
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5390789222243455e-06,
+      "loss": 0.9166,
+      "step": 25806
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5389833747372446e-06,
+      "loss": 0.2794,
+      "step": 25807
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.538887827250143e-06,
+      "loss": 0.8174,
+      "step": 25808
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5387922797630426e-06,
+      "loss": 0.6793,
+      "step": 25809
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5386967322759417e-06,
+      "loss": 0.8345,
+      "step": 25810
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5386011847888403e-06,
+      "loss": 0.4126,
+      "step": 25811
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5385056373017393e-06,
+      "loss": 1.0122,
+      "step": 25812
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.538410089814638e-06,
+      "loss": 0.5981,
+      "step": 25813
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.538314542327537e-06,
+      "loss": 0.5336,
+      "step": 25814
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5382189948404356e-06,
+      "loss": 0.8223,
+      "step": 25815
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.538123447353335e-06,
+      "loss": 0.4858,
+      "step": 25816
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5380278998662337e-06,
+      "loss": 0.7178,
+      "step": 25817
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5379323523791327e-06,
+      "loss": 1.3089,
+      "step": 25818
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5378368048920318e-06,
+      "loss": 0.4977,
+      "step": 25819
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5377412574049304e-06,
+      "loss": 0.7795,
+      "step": 25820
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5376457099178294e-06,
+      "loss": 0.7976,
+      "step": 25821
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.537550162430728e-06,
+      "loss": 0.5465,
+      "step": 25822
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5374546149436275e-06,
+      "loss": 0.3569,
+      "step": 25823
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.537359067456526e-06,
+      "loss": 0.8265,
+      "step": 25824
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.537263519969425e-06,
+      "loss": 0.8143,
+      "step": 25825
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5371679724823238e-06,
+      "loss": 1.352,
+      "step": 25826
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.537072424995223e-06,
+      "loss": 0.7255,
+      "step": 25827
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5369768775081214e-06,
+      "loss": 1.0668,
+      "step": 25828
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5368813300210204e-06,
+      "loss": 0.638,
+      "step": 25829
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.53678578253392e-06,
+      "loss": 1.0258,
+      "step": 25830
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5366902350468185e-06,
+      "loss": 0.8917,
+      "step": 25831
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5365946875597176e-06,
+      "loss": 0.9521,
+      "step": 25832
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.536499140072616e-06,
+      "loss": 0.4376,
+      "step": 25833
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5364035925855152e-06,
+      "loss": 0.6224,
+      "step": 25834
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.536308045098414e-06,
+      "loss": 0.6984,
+      "step": 25835
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.536212497611313e-06,
+      "loss": 0.5789,
+      "step": 25836
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5361169501242123e-06,
+      "loss": 0.5632,
+      "step": 25837
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.536021402637111e-06,
+      "loss": 0.724,
+      "step": 25838
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.53592585515001e-06,
+      "loss": 0.5626,
+      "step": 25839
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5358303076629086e-06,
+      "loss": 0.7312,
+      "step": 25840
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5357347601758076e-06,
+      "loss": 0.8982,
+      "step": 25841
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5356392126887063e-06,
+      "loss": 1.2853,
+      "step": 25842
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5355436652016053e-06,
+      "loss": 0.7347,
+      "step": 25843
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.535448117714504e-06,
+      "loss": 0.8202,
+      "step": 25844
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5353525702274034e-06,
+      "loss": 0.9233,
+      "step": 25845
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5352570227403024e-06,
+      "loss": 0.5968,
+      "step": 25846
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.535161475253201e-06,
+      "loss": 0.8456,
+      "step": 25847
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5350659277661e-06,
+      "loss": 0.8266,
+      "step": 25848
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5349703802789987e-06,
+      "loss": 0.7701,
+      "step": 25849
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5348748327918977e-06,
+      "loss": 0.8308,
+      "step": 25850
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5347792853047963e-06,
+      "loss": 0.8908,
+      "step": 25851
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.534683737817696e-06,
+      "loss": 0.982,
+      "step": 25852
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.534588190330595e-06,
+      "loss": 0.7128,
+      "step": 25853
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5344926428434934e-06,
+      "loss": 0.5475,
+      "step": 25854
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5343970953563925e-06,
+      "loss": 0.5161,
+      "step": 25855
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.534301547869291e-06,
+      "loss": 1.1482,
+      "step": 25856
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.53420600038219e-06,
+      "loss": 0.5539,
+      "step": 25857
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5341104528950887e-06,
+      "loss": 0.6252,
+      "step": 25858
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.534014905407988e-06,
+      "loss": 0.8983,
+      "step": 25859
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5339193579208873e-06,
+      "loss": 0.9864,
+      "step": 25860
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.533823810433786e-06,
+      "loss": 0.8757,
+      "step": 25861
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.533728262946685e-06,
+      "loss": 1.3167,
+      "step": 25862
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5336327154595835e-06,
+      "loss": 0.7772,
+      "step": 25863
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5335371679724826e-06,
+      "loss": 0.6894,
+      "step": 25864
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.533441620485381e-06,
+      "loss": 0.6599,
+      "step": 25865
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5333460729982806e-06,
+      "loss": 1.0913,
+      "step": 25866
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5332505255111792e-06,
+      "loss": 0.6045,
+      "step": 25867
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5331549780240783e-06,
+      "loss": 1.2315,
+      "step": 25868
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5330594305369773e-06,
+      "loss": 0.9482,
+      "step": 25869
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.532963883049876e-06,
+      "loss": 0.6912,
+      "step": 25870
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.532868335562775e-06,
+      "loss": 1.1058,
+      "step": 25871
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5327727880756736e-06,
+      "loss": 0.6718,
+      "step": 25872
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.532677240588573e-06,
+      "loss": 0.8517,
+      "step": 25873
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5325816931014717e-06,
+      "loss": 0.5566,
+      "step": 25874
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5324861456143707e-06,
+      "loss": 1.2105,
+      "step": 25875
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5323905981272693e-06,
+      "loss": 0.9231,
+      "step": 25876
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5322950506401684e-06,
+      "loss": 0.9435,
+      "step": 25877
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.532199503153067e-06,
+      "loss": 0.6087,
+      "step": 25878
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.532103955665966e-06,
+      "loss": 0.4976,
+      "step": 25879
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5320084081788646e-06,
+      "loss": 0.8003,
+      "step": 25880
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.531912860691764e-06,
+      "loss": 0.684,
+      "step": 25881
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.531817313204663e-06,
+      "loss": 0.9479,
+      "step": 25882
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5317217657175617e-06,
+      "loss": 1.2735,
+      "step": 25883
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5316262182304608e-06,
+      "loss": 0.7816,
+      "step": 25884
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5315306707433594e-06,
+      "loss": 1.2329,
+      "step": 25885
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5314351232562584e-06,
+      "loss": 0.8121,
+      "step": 25886
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.531339575769157e-06,
+      "loss": 0.8555,
+      "step": 25887
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5312440282820565e-06,
+      "loss": 0.7875,
+      "step": 25888
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5311484807949556e-06,
+      "loss": 0.5003,
+      "step": 25889
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.531052933307854e-06,
+      "loss": 0.9317,
+      "step": 25890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530957385820753e-06,
+      "loss": 0.4805,
+      "step": 25891
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530861838333652e-06,
+      "loss": 0.7764,
+      "step": 25892
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530766290846551e-06,
+      "loss": 0.721,
+      "step": 25893
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5306707433594495e-06,
+      "loss": 0.4837,
+      "step": 25894
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530575195872349e-06,
+      "loss": 1.0837,
+      "step": 25895
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530479648385248e-06,
+      "loss": 0.5222,
+      "step": 25896
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5303841008981466e-06,
+      "loss": 0.4333,
+      "step": 25897
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5302885534110456e-06,
+      "loss": 0.774,
+      "step": 25898
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5301930059239442e-06,
+      "loss": 0.5686,
+      "step": 25899
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5300974584368433e-06,
+      "loss": 1.1564,
+      "step": 25900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.530001910949742e-06,
+      "loss": 0.5031,
+      "step": 25901
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5299063634626414e-06,
+      "loss": 0.7859,
+      "step": 25902
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5298108159755404e-06,
+      "loss": 0.7934,
+      "step": 25903
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.529715268488439e-06,
+      "loss": 0.4862,
+      "step": 25904
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.529619721001338e-06,
+      "loss": 0.8181,
+      "step": 25905
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5295241735142367e-06,
+      "loss": 0.6537,
+      "step": 25906
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5294286260271357e-06,
+      "loss": 1.0559,
+      "step": 25907
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5293330785400343e-06,
+      "loss": 1.1346,
+      "step": 25908
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5292375310529338e-06,
+      "loss": 0.6187,
+      "step": 25909
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.529141983565833e-06,
+      "loss": 0.7464,
+      "step": 25910
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5290464360787314e-06,
+      "loss": 0.6034,
+      "step": 25911
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5289508885916305e-06,
+      "loss": 0.8297,
+      "step": 25912
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528855341104529e-06,
+      "loss": 0.5558,
+      "step": 25913
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528759793617428e-06,
+      "loss": 0.9299,
+      "step": 25914
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5286642461303267e-06,
+      "loss": 0.8185,
+      "step": 25915
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5285686986432258e-06,
+      "loss": 1.2511,
+      "step": 25916
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528473151156125e-06,
+      "loss": 0.785,
+      "step": 25917
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528377603669024e-06,
+      "loss": 0.6582,
+      "step": 25918
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5282820561819225e-06,
+      "loss": 0.56,
+      "step": 25919
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5281865086948215e-06,
+      "loss": 0.6169,
+      "step": 25920
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5280909612077206e-06,
+      "loss": 1.0274,
+      "step": 25921
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.527995413720619e-06,
+      "loss": 0.8504,
+      "step": 25922
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.527899866233518e-06,
+      "loss": 0.9836,
+      "step": 25923
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5278043187464172e-06,
+      "loss": 1.0527,
+      "step": 25924
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5277087712593163e-06,
+      "loss": 0.4256,
+      "step": 25925
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.527613223772215e-06,
+      "loss": 1.014,
+      "step": 25926
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.527517676285114e-06,
+      "loss": 0.6472,
+      "step": 25927
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5274221287980126e-06,
+      "loss": 0.8522,
+      "step": 25928
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5273265813109116e-06,
+      "loss": 0.6676,
+      "step": 25929
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.52723103382381e-06,
+      "loss": 0.816,
+      "step": 25930
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5271354863367097e-06,
+      "loss": 1.0927,
+      "step": 25931
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5270399388496087e-06,
+      "loss": 1.0165,
+      "step": 25932
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5269443913625073e-06,
+      "loss": 1.2274,
+      "step": 25933
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5268488438754064e-06,
+      "loss": 0.7508,
+      "step": 25934
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.526753296388305e-06,
+      "loss": 0.9423,
+      "step": 25935
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.526657748901204e-06,
+      "loss": 0.8429,
+      "step": 25936
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5265622014141026e-06,
+      "loss": 0.9481,
+      "step": 25937
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.526466653927002e-06,
+      "loss": 0.5445,
+      "step": 25938
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.526371106439901e-06,
+      "loss": 0.9438,
+      "step": 25939
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5262755589527997e-06,
+      "loss": 0.9289,
+      "step": 25940
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5261800114656988e-06,
+      "loss": 0.6389,
+      "step": 25941
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5260844639785974e-06,
+      "loss": 1.1736,
+      "step": 25942
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5259889164914964e-06,
+      "loss": 1.0073,
+      "step": 25943
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525893369004395e-06,
+      "loss": 0.8479,
+      "step": 25944
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5257978215172945e-06,
+      "loss": 0.6898,
+      "step": 25945
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5257022740301936e-06,
+      "loss": 0.7555,
+      "step": 25946
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525606726543092e-06,
+      "loss": 0.6094,
+      "step": 25947
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525511179055991e-06,
+      "loss": 0.7668,
+      "step": 25948
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.52541563156889e-06,
+      "loss": 0.4018,
+      "step": 25949
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525320084081789e-06,
+      "loss": 0.5034,
+      "step": 25950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5252245365946875e-06,
+      "loss": 0.6002,
+      "step": 25951
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5251289891075865e-06,
+      "loss": 0.8209,
+      "step": 25952
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525033441620486e-06,
+      "loss": 0.9081,
+      "step": 25953
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5249378941333846e-06,
+      "loss": 1.442,
+      "step": 25954
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5248423466462836e-06,
+      "loss": 0.5961,
+      "step": 25955
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5247467991591822e-06,
+      "loss": 0.7743,
+      "step": 25956
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5246512516720813e-06,
+      "loss": 0.8964,
+      "step": 25957
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.52455570418498e-06,
+      "loss": 0.7292,
+      "step": 25958
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.524460156697879e-06,
+      "loss": 0.894,
+      "step": 25959
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5243646092107784e-06,
+      "loss": 0.5486,
+      "step": 25960
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.524269061723677e-06,
+      "loss": 0.6675,
+      "step": 25961
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.524173514236576e-06,
+      "loss": 1.0898,
+      "step": 25962
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5240779667494747e-06,
+      "loss": 0.9217,
+      "step": 25963
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5239824192623737e-06,
+      "loss": 0.7113,
+      "step": 25964
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5238868717752723e-06,
+      "loss": 1.2357,
+      "step": 25965
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5237913242881714e-06,
+      "loss": 1.1996,
+      "step": 25966
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5236957768010704e-06,
+      "loss": 0.7883,
+      "step": 25967
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5236002293139694e-06,
+      "loss": 1.3861,
+      "step": 25968
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.523504681826868e-06,
+      "loss": 0.5972,
+      "step": 25969
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.523409134339767e-06,
+      "loss": 1.2171,
+      "step": 25970
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5233135868526657e-06,
+      "loss": 0.9667,
+      "step": 25971
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5232180393655647e-06,
+      "loss": 1.1088,
+      "step": 25972
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5231224918784638e-06,
+      "loss": 1.2193,
+      "step": 25973
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.523026944391363e-06,
+      "loss": 0.5573,
+      "step": 25974
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.522931396904262e-06,
+      "loss": 0.8341,
+      "step": 25975
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5228358494171605e-06,
+      "loss": 0.6859,
+      "step": 25976
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5227403019300595e-06,
+      "loss": 0.6394,
+      "step": 25977
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.522644754442958e-06,
+      "loss": 1.6059,
+      "step": 25978
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.522549206955857e-06,
+      "loss": 0.6898,
+      "step": 25979
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5224536594687558e-06,
+      "loss": 0.7299,
+      "step": 25980
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5223581119816552e-06,
+      "loss": 0.8875,
+      "step": 25981
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5222625644945543e-06,
+      "loss": 0.7421,
+      "step": 25982
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.522167017007453e-06,
+      "loss": 0.5787,
+      "step": 25983
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.522071469520352e-06,
+      "loss": 0.6402,
+      "step": 25984
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5219759220332506e-06,
+      "loss": 0.5202,
+      "step": 25985
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5218803745461496e-06,
+      "loss": 0.8031,
+      "step": 25986
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.521784827059048e-06,
+      "loss": 0.7856,
+      "step": 25987
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5216892795719472e-06,
+      "loss": 0.8645,
+      "step": 25988
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5215937320848467e-06,
+      "loss": 1.4735,
+      "step": 25989
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5214981845977453e-06,
+      "loss": 0.9007,
+      "step": 25990
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5214026371106444e-06,
+      "loss": 0.9206,
+      "step": 25991
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.521307089623543e-06,
+      "loss": 0.7077,
+      "step": 25992
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.521211542136442e-06,
+      "loss": 0.7555,
+      "step": 25993
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5211159946493406e-06,
+      "loss": 0.532,
+      "step": 25994
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5210204471622397e-06,
+      "loss": 0.7065,
+      "step": 25995
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.520924899675139e-06,
+      "loss": 0.7928,
+      "step": 25996
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5208293521880377e-06,
+      "loss": 1.1944,
+      "step": 25997
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5207338047009368e-06,
+      "loss": 0.8449,
+      "step": 25998
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5206382572138354e-06,
+      "loss": 0.8757,
+      "step": 25999
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5205427097267344e-06,
+      "loss": 0.6936,
+      "step": 26000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.520447162239633e-06,
+      "loss": 0.5707,
+      "step": 26001
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.520351614752532e-06,
+      "loss": 0.7283,
+      "step": 26002
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5202560672654316e-06,
+      "loss": 0.7975,
+      "step": 26003
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.52016051977833e-06,
+      "loss": 1.2967,
+      "step": 26004
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.520064972291229e-06,
+      "loss": 1.0733,
+      "step": 26005
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.519969424804128e-06,
+      "loss": 1.0712,
+      "step": 26006
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.519873877317027e-06,
+      "loss": 0.9936,
+      "step": 26007
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5197783298299255e-06,
+      "loss": 0.7537,
+      "step": 26008
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5196827823428245e-06,
+      "loss": 0.6324,
+      "step": 26009
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5195872348557236e-06,
+      "loss": 0.6642,
+      "step": 26010
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5194916873686226e-06,
+      "loss": 0.9586,
+      "step": 26011
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5193961398815216e-06,
+      "loss": 0.63,
+      "step": 26012
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5193005923944202e-06,
+      "loss": 0.8588,
+      "step": 26013
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5192050449073193e-06,
+      "loss": 0.7926,
+      "step": 26014
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.519109497420218e-06,
+      "loss": 0.6993,
+      "step": 26015
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.519013949933117e-06,
+      "loss": 0.7887,
+      "step": 26016
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518918402446016e-06,
+      "loss": 0.2719,
+      "step": 26017
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518822854958915e-06,
+      "loss": 0.9449,
+      "step": 26018
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5187273074718136e-06,
+      "loss": 1.2537,
+      "step": 26019
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5186317599847127e-06,
+      "loss": 0.9811,
+      "step": 26020
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5185362124976113e-06,
+      "loss": 0.3717,
+      "step": 26021
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5184406650105103e-06,
+      "loss": 0.7462,
+      "step": 26022
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5183451175234094e-06,
+      "loss": 1.0105,
+      "step": 26023
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518249570036308e-06,
+      "loss": 0.7554,
+      "step": 26024
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5181540225492074e-06,
+      "loss": 0.6864,
+      "step": 26025
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.518058475062106e-06,
+      "loss": 0.5934,
+      "step": 26026
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.517962927575005e-06,
+      "loss": 1.1267,
+      "step": 26027
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5178673800879037e-06,
+      "loss": 0.5713,
+      "step": 26028
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5177718326008027e-06,
+      "loss": 0.6035,
+      "step": 26029
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5176762851137014e-06,
+      "loss": 1.024,
+      "step": 26030
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5175807376266004e-06,
+      "loss": 0.9371,
+      "step": 26031
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5174851901395e-06,
+      "loss": 1.03,
+      "step": 26032
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5173896426523985e-06,
+      "loss": 0.7404,
+      "step": 26033
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5172940951652975e-06,
+      "loss": 0.351,
+      "step": 26034
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.517198547678196e-06,
+      "loss": 0.994,
+      "step": 26035
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.517103000191095e-06,
+      "loss": 0.8278,
+      "step": 26036
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5170074527039938e-06,
+      "loss": 0.5688,
+      "step": 26037
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.516911905216893e-06,
+      "loss": 0.7922,
+      "step": 26038
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5168163577297923e-06,
+      "loss": 0.7266,
+      "step": 26039
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.516720810242691e-06,
+      "loss": 1.0674,
+      "step": 26040
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.51662526275559e-06,
+      "loss": 0.5095,
+      "step": 26041
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5165297152684886e-06,
+      "loss": 0.832,
+      "step": 26042
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5164341677813876e-06,
+      "loss": 0.8051,
+      "step": 26043
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.516338620294286e-06,
+      "loss": 0.935,
+      "step": 26044
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5162430728071852e-06,
+      "loss": 0.8514,
+      "step": 26045
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5161475253200847e-06,
+      "loss": 0.7447,
+      "step": 26046
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5160519778329833e-06,
+      "loss": 0.9491,
+      "step": 26047
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5159564303458824e-06,
+      "loss": 0.795,
+      "step": 26048
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.515860882858781e-06,
+      "loss": 0.7985,
+      "step": 26049
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.51576533537168e-06,
+      "loss": 0.5263,
+      "step": 26050
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5156697878845786e-06,
+      "loss": 0.8801,
+      "step": 26051
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5155742403974777e-06,
+      "loss": 0.6046,
+      "step": 26052
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.515478692910377e-06,
+      "loss": 0.7112,
+      "step": 26053
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5153831454232757e-06,
+      "loss": 0.9917,
+      "step": 26054
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5152875979361748e-06,
+      "loss": 0.6827,
+      "step": 26055
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5151920504490734e-06,
+      "loss": 1.1612,
+      "step": 26056
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5150965029619724e-06,
+      "loss": 0.5891,
+      "step": 26057
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.515000955474871e-06,
+      "loss": 0.5166,
+      "step": 26058
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.51490540798777e-06,
+      "loss": 0.6114,
+      "step": 26059
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5148098605006687e-06,
+      "loss": 0.9493,
+      "step": 26060
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514714313013568e-06,
+      "loss": 0.7418,
+      "step": 26061
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514618765526467e-06,
+      "loss": 0.8558,
+      "step": 26062
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514523218039366e-06,
+      "loss": 0.6007,
+      "step": 26063
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514427670552265e-06,
+      "loss": 0.6362,
+      "step": 26064
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5143321230651635e-06,
+      "loss": 0.5943,
+      "step": 26065
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5142365755780625e-06,
+      "loss": 0.9604,
+      "step": 26066
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.514141028090961e-06,
+      "loss": 0.6755,
+      "step": 26067
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5140454806038606e-06,
+      "loss": 0.9581,
+      "step": 26068
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513949933116759e-06,
+      "loss": 0.8151,
+      "step": 26069
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5138543856296582e-06,
+      "loss": 0.6552,
+      "step": 26070
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513758838142557e-06,
+      "loss": 0.4901,
+      "step": 26071
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513663290655456e-06,
+      "loss": 1.0119,
+      "step": 26072
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5135677431683545e-06,
+      "loss": 0.8838,
+      "step": 26073
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5134721956812536e-06,
+      "loss": 0.9766,
+      "step": 26074
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.513376648194153e-06,
+      "loss": 0.895,
+      "step": 26075
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5132811007070516e-06,
+      "loss": 0.7341,
+      "step": 26076
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5131855532199507e-06,
+      "loss": 1.0702,
+      "step": 26077
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5130900057328493e-06,
+      "loss": 0.4489,
+      "step": 26078
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5129944582457483e-06,
+      "loss": 0.9905,
+      "step": 26079
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.512898910758647e-06,
+      "loss": 0.796,
+      "step": 26080
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.512803363271546e-06,
+      "loss": 0.7952,
+      "step": 26081
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5127078157844454e-06,
+      "loss": 0.6526,
+      "step": 26082
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.512612268297344e-06,
+      "loss": 1.0052,
+      "step": 26083
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.512516720810243e-06,
+      "loss": 1.2596,
+      "step": 26084
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5124211733231417e-06,
+      "loss": 0.4827,
+      "step": 26085
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5123256258360407e-06,
+      "loss": 0.8434,
+      "step": 26086
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5122300783489394e-06,
+      "loss": 1.0243,
+      "step": 26087
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5121345308618384e-06,
+      "loss": 0.8978,
+      "step": 26088
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.512038983374738e-06,
+      "loss": 0.7352,
+      "step": 26089
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5119434358876365e-06,
+      "loss": 0.5172,
+      "step": 26090
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5118478884005355e-06,
+      "loss": 0.7471,
+      "step": 26091
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511752340913434e-06,
+      "loss": 1.1795,
+      "step": 26092
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511656793426333e-06,
+      "loss": 0.8113,
+      "step": 26093
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5115612459392318e-06,
+      "loss": 1.0392,
+      "step": 26094
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511465698452131e-06,
+      "loss": 0.8127,
+      "step": 26095
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5113701509650294e-06,
+      "loss": 1.219,
+      "step": 26096
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511274603477929e-06,
+      "loss": 0.9963,
+      "step": 26097
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511179055990828e-06,
+      "loss": 0.5116,
+      "step": 26098
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5110835085037266e-06,
+      "loss": 0.6051,
+      "step": 26099
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5109879610166256e-06,
+      "loss": 0.6929,
+      "step": 26100
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.510892413529524e-06,
+      "loss": 1.0536,
+      "step": 26101
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5107968660424232e-06,
+      "loss": 0.4715,
+      "step": 26102
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.510701318555322e-06,
+      "loss": 0.8289,
+      "step": 26103
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5106057710682213e-06,
+      "loss": 0.9715,
+      "step": 26104
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5105102235811204e-06,
+      "loss": 0.9154,
+      "step": 26105
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.510414676094019e-06,
+      "loss": 0.6925,
+      "step": 26106
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.510319128606918e-06,
+      "loss": 0.8591,
+      "step": 26107
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5102235811198166e-06,
+      "loss": 0.7532,
+      "step": 26108
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5101280336327157e-06,
+      "loss": 0.9747,
+      "step": 26109
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5100324861456143e-06,
+      "loss": 0.4996,
+      "step": 26110
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5099369386585137e-06,
+      "loss": 1.0638,
+      "step": 26111
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5098413911714124e-06,
+      "loss": 0.9146,
+      "step": 26112
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5097458436843114e-06,
+      "loss": 0.5433,
+      "step": 26113
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5096502961972104e-06,
+      "loss": 0.7691,
+      "step": 26114
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509554748710109e-06,
+      "loss": 1.0199,
+      "step": 26115
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509459201223008e-06,
+      "loss": 0.9406,
+      "step": 26116
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5093636537359067e-06,
+      "loss": 1.3091,
+      "step": 26117
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509268106248806e-06,
+      "loss": 0.6724,
+      "step": 26118
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5091725587617048e-06,
+      "loss": 0.3849,
+      "step": 26119
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.509077011274604e-06,
+      "loss": 0.8071,
+      "step": 26120
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5089814637875024e-06,
+      "loss": 0.4421,
+      "step": 26121
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5088859163004015e-06,
+      "loss": 0.6183,
+      "step": 26122
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5087903688133e-06,
+      "loss": 0.8082,
+      "step": 26123
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.508694821326199e-06,
+      "loss": 1.0161,
+      "step": 26124
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5085992738390986e-06,
+      "loss": 0.7651,
+      "step": 26125
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.508503726351997e-06,
+      "loss": 0.8957,
+      "step": 26126
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5084081788648962e-06,
+      "loss": 0.9398,
+      "step": 26127
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.508312631377795e-06,
+      "loss": 0.91,
+      "step": 26128
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.508217083890694e-06,
+      "loss": 0.7644,
+      "step": 26129
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5081215364035925e-06,
+      "loss": 0.8195,
+      "step": 26130
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5080259889164916e-06,
+      "loss": 0.6279,
+      "step": 26131
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.50793044142939e-06,
+      "loss": 0.7127,
+      "step": 26132
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5078348939422896e-06,
+      "loss": 0.8278,
+      "step": 26133
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5077393464551887e-06,
+      "loss": 0.8323,
+      "step": 26134
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5076437989680873e-06,
+      "loss": 0.9389,
+      "step": 26135
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5075482514809863e-06,
+      "loss": 0.9673,
+      "step": 26136
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.507452703993885e-06,
+      "loss": 0.4856,
+      "step": 26137
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.507357156506784e-06,
+      "loss": 1.0783,
+      "step": 26138
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5072616090196826e-06,
+      "loss": 1.153,
+      "step": 26139
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.507166061532582e-06,
+      "loss": 0.6405,
+      "step": 26140
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.507070514045481e-06,
+      "loss": 1.123,
+      "step": 26141
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5069749665583797e-06,
+      "loss": 0.3658,
+      "step": 26142
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5068794190712787e-06,
+      "loss": 0.8755,
+      "step": 26143
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5067838715841774e-06,
+      "loss": 0.8227,
+      "step": 26144
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5066883240970764e-06,
+      "loss": 1.0645,
+      "step": 26145
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.506592776609975e-06,
+      "loss": 0.6635,
+      "step": 26146
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5064972291228745e-06,
+      "loss": 1.2391,
+      "step": 26147
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5064016816357735e-06,
+      "loss": 0.6836,
+      "step": 26148
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.506306134148672e-06,
+      "loss": 0.4142,
+      "step": 26149
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.506210586661571e-06,
+      "loss": 0.8826,
+      "step": 26150
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5061150391744698e-06,
+      "loss": 0.834,
+      "step": 26151
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.506019491687369e-06,
+      "loss": 0.9914,
+      "step": 26152
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5059239442002674e-06,
+      "loss": 0.7525,
+      "step": 26153
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.505828396713167e-06,
+      "loss": 1.114,
+      "step": 26154
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.505732849226066e-06,
+      "loss": 1.0729,
+      "step": 26155
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5056373017389646e-06,
+      "loss": 1.3571,
+      "step": 26156
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5055417542518636e-06,
+      "loss": 1.1197,
+      "step": 26157
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.505446206764762e-06,
+      "loss": 0.3853,
+      "step": 26158
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5053506592776612e-06,
+      "loss": 0.5391,
+      "step": 26159
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.50525511179056e-06,
+      "loss": 0.7968,
+      "step": 26160
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5051595643034593e-06,
+      "loss": 0.607,
+      "step": 26161
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.505064016816358e-06,
+      "loss": 1.4522,
+      "step": 26162
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.504968469329257e-06,
+      "loss": 0.5308,
+      "step": 26163
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.504872921842156e-06,
+      "loss": 0.6107,
+      "step": 26164
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5047773743550546e-06,
+      "loss": 0.7293,
+      "step": 26165
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5046818268679537e-06,
+      "loss": 1.184,
+      "step": 26166
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5045862793808523e-06,
+      "loss": 0.587,
+      "step": 26167
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5044907318937513e-06,
+      "loss": 0.3216,
+      "step": 26168
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5043951844066504e-06,
+      "loss": 0.9248,
+      "step": 26169
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5042996369195494e-06,
+      "loss": 0.7962,
+      "step": 26170
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.504204089432448e-06,
+      "loss": 0.9279,
+      "step": 26171
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.504108541945347e-06,
+      "loss": 0.5615,
+      "step": 26172
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5040129944582457e-06,
+      "loss": 0.977,
+      "step": 26173
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5039174469711447e-06,
+      "loss": 1.0542,
+      "step": 26174
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5038218994840433e-06,
+      "loss": 0.7153,
+      "step": 26175
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5037263519969428e-06,
+      "loss": 0.7253,
+      "step": 26176
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.503630804509842e-06,
+      "loss": 0.743,
+      "step": 26177
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5035352570227404e-06,
+      "loss": 0.5233,
+      "step": 26178
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5034397095356395e-06,
+      "loss": 0.6886,
+      "step": 26179
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.503344162048538e-06,
+      "loss": 1.1918,
+      "step": 26180
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.503248614561437e-06,
+      "loss": 0.7511,
+      "step": 26181
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5031530670743357e-06,
+      "loss": 1.3257,
+      "step": 26182
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.503057519587235e-06,
+      "loss": 0.7284,
+      "step": 26183
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5029619721001342e-06,
+      "loss": 0.391,
+      "step": 26184
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.502866424613033e-06,
+      "loss": 0.7557,
+      "step": 26185
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.502770877125932e-06,
+      "loss": 0.9616,
+      "step": 26186
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5026753296388305e-06,
+      "loss": 0.8279,
+      "step": 26187
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5025797821517296e-06,
+      "loss": 0.8291,
+      "step": 26188
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.502484234664628e-06,
+      "loss": 0.8518,
+      "step": 26189
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5023886871775276e-06,
+      "loss": 1.0157,
+      "step": 26190
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7072524640240555,
+      "eval_loss": 1.0227203369140625,
+      "eval_runtime": 141.144,
+      "eval_samples_per_second": 296.881,
+      "eval_steps_per_second": 2.324,
+      "step": 26190
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 52380,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 3.590426029622454e+16,
+  "trial_name": null,
+  "trial_params": null
+}