diff --git "a/trainer_state.json" "b/trainer_state.json"
deleted file mode 100644--- "a/trainer_state.json"
+++ /dev/null
@@ -1,7498 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 0.9994982438534872,
-  "eval_steps": 500,
-  "global_step": 1245,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.6315789473684213e-07,
-      "loss": 3.4856,
-      "step": 1
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5.263157894736843e-07,
-      "loss": 3.5104,
-      "step": 2
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 7.894736842105263e-07,
-      "loss": 3.5029,
-      "step": 3
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.0526315789473685e-06,
-      "loss": 3.5424,
-      "step": 4
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.3157894736842106e-06,
-      "loss": 3.6725,
-      "step": 5
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.5789473684210526e-06,
-      "loss": 3.4155,
-      "step": 6
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.8421052631578948e-06,
-      "loss": 3.3714,
-      "step": 7
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.105263157894737e-06,
-      "loss": 3.313,
-      "step": 8
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.368421052631579e-06,
-      "loss": 3.1553,
-      "step": 9
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.631578947368421e-06,
-      "loss": 2.7603,
-      "step": 10
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 2.8947368421052634e-06,
-      "loss": 2.6668,
-      "step": 11
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 3.157894736842105e-06,
-      "loss": 2.5377,
-      "step": 12
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 3.421052631578948e-06,
-      "loss": 2.0743,
-      "step": 13
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 3.6842105263157896e-06,
-      "loss": 1.938,
-      "step": 14
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 3.947368421052632e-06,
-      "loss": 1.8478,
-      "step": 15
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 4.210526315789474e-06,
-      "loss": 1.728,
-      "step": 16
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 4.473684210526316e-06,
-      "loss": 1.572,
-      "step": 17
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 4.736842105263158e-06,
-      "loss": 1.5123,
-      "step": 18
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5e-06,
-      "loss": 1.3983,
-      "step": 19
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.263157894736842e-06,
-      "loss": 1.3753,
-      "step": 20
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.526315789473685e-06,
-      "loss": 1.3107,
-      "step": 21
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 5.789473684210527e-06,
-      "loss": 1.3095,
-      "step": 22
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 6.0526315789473685e-06,
-      "loss": 1.3143,
-      "step": 23
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 6.31578947368421e-06,
-      "loss": 1.2648,
-      "step": 24
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 6.578947368421054e-06,
-      "loss": 1.2468,
-      "step": 25
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 6.842105263157896e-06,
-      "loss": 1.2034,
-      "step": 26
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 7.1052631578947375e-06,
-      "loss": 1.2304,
-      "step": 27
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 7.368421052631579e-06,
-      "loss": 1.1427,
-      "step": 28
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 7.631578947368423e-06,
-      "loss": 1.2089,
-      "step": 29
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 7.894736842105265e-06,
-      "loss": 1.1489,
-      "step": 30
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 8.157894736842106e-06,
-      "loss": 1.2023,
-      "step": 31
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 8.421052631578948e-06,
-      "loss": 1.153,
-      "step": 32
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 8.68421052631579e-06,
-      "loss": 1.1394,
-      "step": 33
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 8.947368421052632e-06,
-      "loss": 1.1188,
-      "step": 34
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.210526315789474e-06,
-      "loss": 1.1399,
-      "step": 35
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.473684210526315e-06,
-      "loss": 1.1522,
-      "step": 36
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.736842105263159e-06,
-      "loss": 1.1174,
-      "step": 37
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 1e-05,
-      "loss": 1.0725,
-      "step": 38
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.999983063449216e-06,
-      "loss": 1.0799,
-      "step": 39
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.999932253911604e-06,
-      "loss": 1.0771,
-      "step": 40
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.999847571731377e-06,
-      "loss": 1.0669,
-      "step": 41
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.999729017482228e-06,
-      "loss": 1.1253,
-      "step": 42
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 9.999576591967311e-06,
-      "loss": 1.0429,
-      "step": 43
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.999390296219259e-06,
-      "loss": 1.0779,
-      "step": 44
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.999390296219259e-06,
-      "loss": 1.0533,
-      "step": 45
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.999170131500146e-06,
-      "loss": 1.1058,
-      "step": 46
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.998916099301512e-06,
-      "loss": 1.0537,
-      "step": 47
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.998628201344327e-06,
-      "loss": 1.0612,
-      "step": 48
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.998306439578988e-06,
-      "loss": 1.0547,
-      "step": 49
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.99795081618531e-06,
-      "loss": 1.0395,
-      "step": 50
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.997561333572507e-06,
-      "loss": 1.0177,
-      "step": 51
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.997137994379174e-06,
-      "loss": 1.0468,
-      "step": 52
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.996680801473275e-06,
-      "loss": 1.0289,
-      "step": 53
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.99618975795212e-06,
-      "loss": 1.0715,
-      "step": 54
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.99566486714234e-06,
-      "loss": 1.0198,
-      "step": 55
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 9.995106132599869e-06,
-      "loss": 1.024,
-      "step": 56
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.994513558109925e-06,
-      "loss": 0.9955,
-      "step": 57
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.993887147686974e-06,
-      "loss": 1.0324,
-      "step": 58
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.993226905574708e-06,
-      "loss": 1.0029,
-      "step": 59
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.992532836246019e-06,
-      "loss": 1.0413,
-      "step": 60
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.99180494440296e-06,
-      "loss": 0.9969,
-      "step": 61
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.991043234976723e-06,
-      "loss": 1.0329,
-      "step": 62
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.9902477131276e-06,
-      "loss": 1.0264,
-      "step": 63
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.989418384244952e-06,
-      "loss": 1.0259,
-      "step": 64
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.988555253947166e-06,
-      "loss": 1.014,
-      "step": 65
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.987658328081618e-06,
-      "loss": 1.052,
-      "step": 66
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.986727612724645e-06,
-      "loss": 1.0204,
-      "step": 67
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 9.985763114181489e-06,
-      "loss": 1.0299,
-      "step": 68
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.98476483898626e-06,
-      "loss": 1.0163,
-      "step": 69
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.983732793901897e-06,
-      "loss": 1.0167,
-      "step": 70
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.982666985920108e-06,
-      "loss": 1.0076,
-      "step": 71
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.982666985920108e-06,
-      "loss": 1.0161,
-      "step": 72
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.981567422261343e-06,
-      "loss": 0.9813,
-      "step": 73
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.980434110374725e-06,
-      "loss": 1.0051,
-      "step": 74
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.979267057938012e-06,
-      "loss": 1.0075,
-      "step": 75
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.978066272857543e-06,
-      "loss": 1.0083,
-      "step": 76
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.976831763268177e-06,
-      "loss": 1.0487,
-      "step": 77
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.975563537533253e-06,
-      "loss": 1.0091,
-      "step": 78
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.974261604244514e-06,
-      "loss": 0.9705,
-      "step": 79
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 9.972925972222068e-06,
-      "loss": 0.9536,
-      "step": 80
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.971556650514313e-06,
-      "loss": 0.9602,
-      "step": 81
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.970153648397883e-06,
-      "loss": 1.0246,
-      "step": 82
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.968716975377584e-06,
-      "loss": 0.9897,
-      "step": 83
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.968716975377584e-06,
-      "loss": 0.983,
-      "step": 84
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.96724664118633e-06,
-      "loss": 0.9279,
-      "step": 85
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.965742655785081e-06,
-      "loss": 1.0062,
-      "step": 86
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.964205029362764e-06,
-      "loss": 0.98,
-      "step": 87
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.962633772336212e-06,
-      "loss": 0.9849,
-      "step": 88
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.9610288953501e-06,
-      "loss": 0.9891,
-      "step": 89
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.959390409276856e-06,
-      "loss": 1.0236,
-      "step": 90
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.957718325216601e-06,
-      "loss": 0.9913,
-      "step": 91
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.956012654497073e-06,
-      "loss": 0.9626,
-      "step": 92
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.954273408673543e-06,
-      "loss": 0.9829,
-      "step": 93
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.952500599528738e-06,
-      "loss": 0.9724,
-      "step": 94
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.950694239072767e-06,
-      "loss": 0.9571,
-      "step": 95
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.94885433954304e-06,
-      "loss": 0.9815,
-      "step": 96
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.946980913404174e-06,
-      "loss": 1.0121,
-      "step": 97
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.945073973347923e-06,
-      "loss": 0.997,
-      "step": 98
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.943133532293078e-06,
-      "loss": 1.0058,
-      "step": 99
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.941159603385396e-06,
-      "loss": 0.962,
-      "step": 100
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.939152199997488e-06,
-      "loss": 0.9614,
-      "step": 101
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.937111335728758e-06,
-      "loss": 1.0239,
-      "step": 102
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.93503702440528e-06,
-      "loss": 0.9753,
-      "step": 103
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.932929280079728e-06,
-      "loss": 0.9693,
-      "step": 104
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.930788117031271e-06,
-      "loss": 0.9922,
-      "step": 105
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.928613549765476e-06,
-      "loss": 1.0256,
-      "step": 106
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.926405593014208e-06,
-      "loss": 0.9717,
-      "step": 107
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.924164261735536e-06,
-      "loss": 0.9934,
-      "step": 108
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.921889571113629e-06,
-      "loss": 0.997,
-      "step": 109
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.919581536558654e-06,
-      "loss": 1.0072,
-      "step": 110
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.917240173706666e-06,
-      "loss": 1.0104,
-      "step": 111
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.91486549841951e-06,
-      "loss": 0.9454,
-      "step": 112
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.91245752678471e-06,
-      "loss": 0.9665,
-      "step": 113
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.91001627511536e-06,
-      "loss": 1.0083,
-      "step": 114
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.907541759950015e-06,
-      "loss": 0.9561,
-      "step": 115
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.90503399805257e-06,
-      "loss": 0.9705,
-      "step": 116
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.902493006412164e-06,
-      "loss": 0.9597,
-      "step": 117
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.899918802243052e-06,
-      "loss": 0.9804,
-      "step": 118
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.897311402984484e-06,
-      "loss": 0.9177,
-      "step": 119
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.894670826300605e-06,
-      "loss": 0.9876,
-      "step": 120
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.891997090080319e-06,
-      "loss": 0.9496,
-      "step": 121
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.889290212437172e-06,
-      "loss": 0.9683,
-      "step": 122
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.886550211709234e-06,
-      "loss": 0.9657,
-      "step": 123
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.883777106458967e-06,
-      "loss": 0.9615,
-      "step": 124
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.883777106458967e-06,
-      "loss": 0.9544,
-      "step": 125
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.880970915473106e-06,
-      "loss": 0.9544,
-      "step": 126
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.880970915473106e-06,
-      "loss": 0.9497,
-      "step": 127
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.878131657762535e-06,
-      "loss": 0.9445,
-      "step": 128
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.875259352562141e-06,
-      "loss": 0.9823,
-      "step": 129
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.872354019330703e-06,
-      "loss": 0.9633,
-      "step": 130
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.869415677750751e-06,
-      "loss": 0.9746,
-      "step": 131
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.866444347728434e-06,
-      "loss": 0.9951,
-      "step": 132
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.863440049393385e-06,
-      "loss": 0.9424,
-      "step": 133
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.860402803098583e-06,
-      "loss": 0.9431,
-      "step": 134
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.857332629420217e-06,
-      "loss": 0.9979,
-      "step": 135
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.854229549157552e-06,
-      "loss": 0.9388,
-      "step": 136
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.854229549157552e-06,
-      "loss": 0.9487,
-      "step": 137
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.851093583332777e-06,
-      "loss": 0.9354,
-      "step": 138
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.847924753190868e-06,
-      "loss": 0.9264,
-      "step": 139
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.844723080199448e-06,
-      "loss": 0.893,
-      "step": 140
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.841488586048635e-06,
-      "loss": 0.9406,
-      "step": 141
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.841488586048635e-06,
-      "loss": 0.9536,
-      "step": 142
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.8382212926509e-06,
-      "loss": 0.9039,
-      "step": 143
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.834921222140915e-06,
-      "loss": 0.9188,
-      "step": 144
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.831588396875402e-06,
-      "loss": 0.9635,
-      "step": 145
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.828222839432989e-06,
-      "loss": 0.9677,
-      "step": 146
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.82482457261405e-06,
-      "loss": 0.9613,
-      "step": 147
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.821393619440554e-06,
-      "loss": 0.9384,
-      "step": 148
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.821393619440554e-06,
-      "loss": 0.9359,
-      "step": 149
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.8179300031559e-06,
-      "loss": 0.9263,
-      "step": 150
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.81443374722478e-06,
-      "loss": 0.9631,
-      "step": 151
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.810904875332998e-06,
-      "loss": 0.9404,
-      "step": 152
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.80734341138732e-06,
-      "loss": 0.9919,
-      "step": 153
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.80734341138732e-06,
-      "loss": 0.9231,
-      "step": 154
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.803749379515314e-06,
-      "loss": 0.9996,
-      "step": 155
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.80012280406518e-06,
-      "loss": 0.9179,
-      "step": 156
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.796463709605588e-06,
-      "loss": 0.9576,
-      "step": 157
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.792772120925517e-06,
-      "loss": 0.9167,
-      "step": 158
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.78904806303408e-06,
-      "loss": 0.9663,
-      "step": 159
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.785291561160349e-06,
-      "loss": 0.9776,
-      "step": 160
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.781502640753203e-06,
-      "loss": 0.9025,
-      "step": 161
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.777681327481136e-06,
-      "loss": 0.9522,
-      "step": 162
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.773827647232098e-06,
-      "loss": 0.9131,
-      "step": 163
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.769941626113307e-06,
-      "loss": 0.9618,
-      "step": 164
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.766023290451081e-06,
-      "loss": 0.9541,
-      "step": 165
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.762072666790658e-06,
-      "loss": 0.8985,
-      "step": 166
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.758089781896009e-06,
-      "loss": 0.9911,
-      "step": 167
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.754074662749673e-06,
-      "loss": 0.8872,
-      "step": 168
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.750027336552553e-06,
-      "loss": 0.9406,
-      "step": 169
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.74594783072375e-06,
-      "loss": 0.9582,
-      "step": 170
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.741836172900366e-06,
-      "loss": 0.9059,
-      "step": 171
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.73769239093732e-06,
-      "loss": 0.9674,
-      "step": 172
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.733516512907162e-06,
-      "loss": 0.9528,
-      "step": 173
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.729308567099883e-06,
-      "loss": 0.9688,
-      "step": 174
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.729308567099883e-06,
-      "loss": 0.9596,
-      "step": 175
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.725068582022717e-06,
-      "loss": 0.9123,
-      "step": 176
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.72079658639995e-06,
-      "loss": 0.9552,
-      "step": 177
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.716492609172734e-06,
-      "loss": 0.9516,
-      "step": 178
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.716492609172734e-06,
-      "loss": 0.9495,
-      "step": 179
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.712156679498878e-06,
-      "loss": 0.9549,
-      "step": 180
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.707788826752662e-06,
-      "loss": 0.9495,
-      "step": 181
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.703389080524628e-06,
-      "loss": 0.9064,
-      "step": 182
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.698957470621386e-06,
-      "loss": 0.9569,
-      "step": 183
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.69449402706541e-06,
-      "loss": 0.9372,
-      "step": 184
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.689998780094839e-06,
-      "loss": 0.9253,
-      "step": 185
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.685471760163259e-06,
-      "loss": 0.9245,
-      "step": 186
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.680912997939516e-06,
-      "loss": 0.9747,
-      "step": 187
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.676322524307492e-06,
-      "loss": 0.9452,
-      "step": 188
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.671700370365899e-06,
-      "loss": 0.9171,
-      "step": 189
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.66704656742808e-06,
-      "loss": 0.9069,
-      "step": 190
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.66236114702178e-06,
-      "loss": 0.9575,
-      "step": 191
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.657644140888944e-06,
-      "loss": 0.9508,
-      "step": 192
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.652895580985499e-06,
-      "loss": 0.9437,
-      "step": 193
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.648115499481134e-06,
-      "loss": 0.9637,
-      "step": 194
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.643303928759087e-06,
-      "loss": 0.973,
-      "step": 195
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.643303928759087e-06,
-      "loss": 0.974,
-      "step": 196
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.63846090141592e-06,
-      "loss": 0.9537,
-      "step": 197
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.633586450261309e-06,
-      "loss": 0.9502,
-      "step": 198
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.633586450261309e-06,
-      "loss": 0.8986,
-      "step": 199
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.628680608317806e-06,
-      "loss": 0.9402,
-      "step": 200
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.62374340882063e-06,
-      "loss": 0.9366,
-      "step": 201
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.61877488521743e-06,
-      "loss": 0.9389,
-      "step": 202
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.61877488521743e-06,
-      "loss": 0.9117,
-      "step": 203
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.613775071168072e-06,
-      "loss": 0.9182,
-      "step": 204
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.608744000544392e-06,
-      "loss": 0.9349,
-      "step": 205
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.603681707429989e-06,
-      "loss": 0.9702,
-      "step": 206
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.59858822611997e-06,
-      "loss": 0.9456,
-      "step": 207
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.593463591120742e-06,
-      "loss": 0.9527,
-      "step": 208
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.58830783714976e-06,
-      "loss": 0.9164,
-      "step": 209
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.5831209991353e-06,
-      "loss": 0.9167,
-      "step": 210
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.577903112216221e-06,
-      "loss": 0.9011,
-      "step": 211
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.572654211741721e-06,
-      "loss": 0.9337,
-      "step": 212
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.567374333271115e-06,
-      "loss": 0.932,
-      "step": 213
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.56206351257357e-06,
-      "loss": 0.938,
-      "step": 214
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.55672178562788e-06,
-      "loss": 0.9379,
-      "step": 215
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.551349188622218e-06,
-      "loss": 0.9153,
-      "step": 216
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.54594575795389e-06,
-      "loss": 0.9341,
-      "step": 217
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.540511530229084e-06,
-      "loss": 0.8977,
-      "step": 218
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.535046542262631e-06,
-      "loss": 0.8865,
-      "step": 219
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.529550831077751e-06,
-      "loss": 0.9205,
-      "step": 220
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.524024433905802e-06,
-      "loss": 0.9278,
-      "step": 221
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.51846738818602e-06,
-      "loss": 0.9383,
-      "step": 222
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.512879731565285e-06,
-      "loss": 0.9324,
-      "step": 223
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.50726150189785e-06,
-      "loss": 0.9579,
-      "step": 224
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.501612737245084e-06,
-      "loss": 0.9274,
-      "step": 225
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.495933475875223e-06,
-      "loss": 0.9295,
-      "step": 226
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.49022375626311e-06,
-      "loss": 0.9386,
-      "step": 227
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.484483617089925e-06,
-      "loss": 0.9112,
-      "step": 228
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.478713097242932e-06,
-      "loss": 0.9068,
-      "step": 229
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.47291223581521e-06,
-      "loss": 0.9354,
-      "step": 230
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.467081072105397e-06,
-      "loss": 0.9672,
-      "step": 231
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.46121964561741e-06,
-      "loss": 0.9492,
-      "step": 232
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.46121964561741e-06,
-      "loss": 0.8949,
-      "step": 233
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.455327996060188e-06,
-      "loss": 0.9799,
-      "step": 234
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.449406163347422e-06,
-      "loss": 0.9369,
-      "step": 235
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.449406163347422e-06,
-      "loss": 0.9374,
-      "step": 236
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.443454187597277e-06,
-      "loss": 0.9049,
-      "step": 237
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.437472109132129e-06,
-      "loss": 0.9136,
-      "step": 238
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.431459968478292e-06,
-      "loss": 0.9058,
-      "step": 239
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.425417806365733e-06,
-      "loss": 0.9308,
-      "step": 240
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.419345663727805e-06,
-      "loss": 0.8802,
-      "step": 241
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.413243581700975e-06,
-      "loss": 0.9487,
-      "step": 242
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.407111601624524e-06,
-      "loss": 0.9151,
-      "step": 243
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.400949765040294e-06,
-      "loss": 0.9268,
-      "step": 244
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.394758113692388e-06,
-      "loss": 0.9218,
-      "step": 245
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.38853668952689e-06,
-      "loss": 0.9259,
-      "step": 246
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.382285534691588e-06,
-      "loss": 0.9308,
-      "step": 247
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.376004691535685e-06,
-      "loss": 0.8823,
-      "step": 248
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.369694202609505e-06,
-      "loss": 0.9138,
-      "step": 249
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.363354110664216e-06,
-      "loss": 0.9141,
-      "step": 250
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.356984458651534e-06,
-      "loss": 0.9331,
-      "step": 251
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.350585289723432e-06,
-      "loss": 0.93,
-      "step": 252
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.344156647231852e-06,
-      "loss": 0.8762,
-      "step": 253
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.337698574728404e-06,
-      "loss": 0.9187,
-      "step": 254
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.331211115964075e-06,
-      "loss": 0.9642,
-      "step": 255
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.331211115964075e-06,
-      "loss": 0.9045,
-      "step": 256
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.324694314888941e-06,
-      "loss": 0.9729,
-      "step": 257
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.31814821565185e-06,
-      "loss": 0.9496,
-      "step": 258
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.31157286260014e-06,
-      "loss": 0.9157,
-      "step": 259
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.304968300279332e-06,
-      "loss": 0.8935,
-      "step": 260
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.298334573432827e-06,
-      "loss": 0.918,
-      "step": 261
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.298334573432827e-06,
-      "loss": 0.919,
-      "step": 262
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.291671727001606e-06,
-      "loss": 0.9442,
-      "step": 263
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.284979806123927e-06,
-      "loss": 0.9249,
-      "step": 264
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.278258856135008e-06,
-      "loss": 0.8902,
-      "step": 265
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.271508922566734e-06,
-      "loss": 0.9128,
-      "step": 266
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.264730051147346e-06,
-      "loss": 0.9059,
-      "step": 267
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.257922287801121e-06,
-      "loss": 0.8984,
-      "step": 268
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.251085678648072e-06,
-      "loss": 0.9272,
-      "step": 269
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.244220270003629e-06,
-      "loss": 0.9082,
-      "step": 270
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.23732610837833e-06,
-      "loss": 0.9497,
-      "step": 271
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.230403240477501e-06,
-      "loss": 0.9461,
-      "step": 272
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.223451713200945e-06,
-      "loss": 0.9716,
-      "step": 273
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.216471573642621e-06,
-      "loss": 0.9398,
-      "step": 274
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.209462869090322e-06,
-      "loss": 0.9025,
-      "step": 275
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.20242564702536e-06,
-      "loss": 0.9578,
-      "step": 276
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.195359955122244e-06,
-      "loss": 0.9576,
-      "step": 277
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.188265841248353e-06,
-      "loss": 0.9184,
-      "step": 278
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.181143353463615e-06,
-      "loss": 0.8961,
-      "step": 279
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.173992540020182e-06,
-      "loss": 0.9139,
-      "step": 280
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.166813449362097e-06,
-      "loss": 0.9496,
-      "step": 281
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.159606130124977e-06,
-      "loss": 0.8677,
-      "step": 282
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.15237063113567e-06,
-      "loss": 0.8723,
-      "step": 283
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.145107001411937e-06,
-      "loss": 0.9285,
-      "step": 284
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.13781529016211e-06,
-      "loss": 0.9224,
-      "step": 285
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.130495546784765e-06,
-      "loss": 0.9173,
-      "step": 286
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.123147820868383e-06,
-      "loss": 0.9425,
-      "step": 287
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.115772162191017e-06,
-      "loss": 0.9023,
-      "step": 288
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.108368620719957e-06,
-      "loss": 0.9022,
-      "step": 289
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.100937246611383e-06,
-      "loss": 0.9396,
-      "step": 290
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.093478090210032e-06,
-      "loss": 0.9255,
-      "step": 291
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.08599120204886e-06,
-      "loss": 0.8779,
-      "step": 292
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.078476632848688e-06,
-      "loss": 0.8873,
-      "step": 293
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.070934433517872e-06,
-      "loss": 0.9292,
-      "step": 294
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.063364655151946e-06,
-      "loss": 0.9375,
-      "step": 295
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.055767349033287e-06,
-      "loss": 0.9096,
-      "step": 296
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.048142566630758e-06,
-      "loss": 0.8942,
-      "step": 297
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.040490359599366e-06,
-      "loss": 0.8816,
-      "step": 298
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.040490359599366e-06,
-      "loss": 0.9072,
-      "step": 299
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.032810779779905e-06,
-      "loss": 0.9223,
-      "step": 300
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.025103879198616e-06,
-      "loss": 0.9044,
-      "step": 301
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.017369710066821e-06,
-      "loss": 0.935,
-      "step": 302
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.009608324780583e-06,
-      "loss": 0.9067,
-      "step": 303
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.001819775920338e-06,
-      "loss": 0.9407,
-      "step": 304
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 8.994004116250547e-06,
-      "loss": 0.9137,
-      "step": 305
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.986161398719339e-06,
-      "loss": 0.9136,
-      "step": 306
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.978291676458146e-06,
-      "loss": 0.9492,
-      "step": 307
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.970395002781347e-06,
-      "loss": 0.9625,
-      "step": 308
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.962471431185911e-06,
-      "loss": 0.9094,
-      "step": 309
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.954521015351025e-06,
-      "loss": 0.9275,
-      "step": 310
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.946543809137738e-06,
-      "loss": 0.8835,
-      "step": 311
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.938539866588593e-06,
-      "loss": 0.9105,
-      "step": 312
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.930509241927262e-06,
-      "loss": 0.8901,
-      "step": 313
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.930509241927262e-06,
-      "loss": 0.9291,
-      "step": 314
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.930509241927262e-06,
-      "loss": 0.8956,
-      "step": 315
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.922451989558178e-06,
-      "loss": 0.8785,
-      "step": 316
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 8.914368164066167e-06,
-      "loss": 0.906,
-      "step": 317
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.906257820216076e-06,
-      "loss": 0.8928,
-      "step": 318
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.906257820216076e-06,
-      "loss": 0.9062,
-      "step": 319
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.898121012952406e-06,
-      "loss": 0.9245,
-      "step": 320
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.889957797398938e-06,
-      "loss": 0.9114,
-      "step": 321
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.881768228858356e-06,
-      "loss": 0.9269,
-      "step": 322
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.873552362811878e-06,
-      "loss": 0.9447,
-      "step": 323
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.865310254918879e-06,
-      "loss": 0.9136,
-      "step": 324
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.857041961016508e-06,
-      "loss": 0.9212,
-      "step": 325
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.848747537119317e-06,
-      "loss": 0.867,
-      "step": 326
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.84042703941888e-06,
-      "loss": 0.9411,
-      "step": 327
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.83208052428341e-06,
-      "loss": 0.914,
-      "step": 328
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.823708048257376e-06,
-      "loss": 0.8958,
-      "step": 329
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 8.815309668061124e-06,
-      "loss": 0.9098,
-      "step": 330
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.806885440590494e-06,
-      "loss": 0.924,
-      "step": 331
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.798435422916425e-06,
-      "loss": 0.9023,
-      "step": 332
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.789959672284582e-06,
-      "loss": 0.9374,
-      "step": 333
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.781458246114954e-06,
-      "loss": 0.9165,
-      "step": 334
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.772931202001478e-06,
-      "loss": 0.8565,
-      "step": 335
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.76437859771164e-06,
-      "loss": 0.8931,
-      "step": 336
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.755800491186083e-06,
-      "loss": 0.9164,
-      "step": 337
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.747196940538228e-06,
-      "loss": 0.9083,
-      "step": 338
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.73856800405386e-06,
-      "loss": 0.9133,
-      "step": 339
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.729913740190747e-06,
-      "loss": 0.8953,
-      "step": 340
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.721234207578245e-06,
-      "loss": 0.8489,
-      "step": 341
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.712529465016884e-06,
-      "loss": 0.9452,
-      "step": 342
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.712529465016884e-06,
-      "loss": 0.8986,
-      "step": 343
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.703799571477998e-06,
-      "loss": 0.8933,
-      "step": 344
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.695044586103297e-06,
-      "loss": 0.9365,
-      "step": 345
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.686264568204483e-06,
-      "loss": 0.9007,
-      "step": 346
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.677459577262844e-06,
-      "loss": 0.9133,
-      "step": 347
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.668629672928852e-06,
-      "loss": 0.9267,
-      "step": 348
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.659774915021753e-06,
-      "loss": 0.8966,
-      "step": 349
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.650895363529172e-06,
-      "loss": 0.9303,
-      "step": 350
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.641991078606699e-06,
-      "loss": 0.9103,
-      "step": 351
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.633062120577483e-06,
-      "loss": 0.9269,
-      "step": 352
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.624108549931824e-06,
-      "loss": 0.8924,
-      "step": 353
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.615130427326765e-06,
-      "loss": 0.9283,
-      "step": 354
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.606127813585675e-06,
-      "loss": 0.9241,
-      "step": 355
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.59710076969785e-06,
-      "loss": 0.9014,
-      "step": 356
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.588049356818078e-06,
-      "loss": 0.899,
-      "step": 357
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.578973636266245e-06,
-      "loss": 0.906,
-      "step": 358
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.569873669526919e-06,
-      "loss": 0.8873,
-      "step": 359
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.560749518248911e-06,
-      "loss": 0.9347,
-      "step": 360
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.551601244244885e-06,
-      "loss": 0.9208,
-      "step": 361
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.542428909490927e-06,
-      "loss": 0.8866,
-      "step": 362
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.533232576126116e-06,
-      "loss": 0.949,
-      "step": 363
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.524012306452124e-06,
-      "loss": 0.876,
-      "step": 364
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.514768162932775e-06,
-      "loss": 0.9485,
-      "step": 365
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.505500208193633e-06,
-      "loss": 0.9019,
-      "step": 366
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.496208505021572e-06,
-      "loss": 0.9223,
-      "step": 367
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.48689311636435e-06,
-      "loss": 0.9289,
-      "step": 368
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.477554105330193e-06,
-      "loss": 0.9076,
-      "step": 369
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.468191535187351e-06,
-      "loss": 0.8675,
-      "step": 370
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.458805469363683e-06,
-      "loss": 0.8676,
-      "step": 371
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.449395971446223e-06,
-      "loss": 0.9265,
-      "step": 372
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.439963105180744e-06,
-      "loss": 0.8737,
-      "step": 373
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.430506934471337e-06,
-      "loss": 0.9099,
-      "step": 374
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.421027523379962e-06,
-      "loss": 0.9092,
-      "step": 375
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.411524936126039e-06,
-      "loss": 0.8854,
-      "step": 376
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.40199923708598e-06,
-      "loss": 0.8825,
-      "step": 377
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.392450490792783e-06,
-      "loss": 0.9178,
-      "step": 378
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.38287876193558e-06,
-      "loss": 0.9385,
-      "step": 379
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.373284115359195e-06,
-      "loss": 0.8687,
-      "step": 380
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.363666616063722e-06,
-      "loss": 0.9073,
-      "step": 381
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.35402632920406e-06,
-      "loss": 0.912,
-      "step": 382
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.35402632920406e-06,
-      "loss": 0.9293,
-      "step": 383
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.3443633200895e-06,
-      "loss": 0.8936,
-      "step": 384
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.334677654183254e-06,
-      "loss": 0.895,
-      "step": 385
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.324969397102033e-06,
-      "loss": 0.9302,
-      "step": 386
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.315238614615592e-06,
-      "loss": 0.9211,
-      "step": 387
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.30548537264629e-06,
-      "loss": 0.8805,
-      "step": 388
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.295709737268635e-06,
-      "loss": 0.915,
-      "step": 389
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.295709737268635e-06,
-      "loss": 0.8803,
-      "step": 390
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.285911774708846e-06,
-      "loss": 0.8924,
-      "step": 391
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.2760915513444e-06,
-      "loss": 0.8835,
-      "step": 392
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.266249133703581e-06,
-      "loss": 0.8825,
-      "step": 393
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.256384588465031e-06,
-      "loss": 0.8741,
-      "step": 394
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.2464979824573e-06,
-      "loss": 0.9065,
-      "step": 395
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.23658938265839e-06,
-      "loss": 0.8987,
-      "step": 396
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.23658938265839e-06,
-      "loss": 0.9026,
-      "step": 397
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.2266588561953e-06,
-      "loss": 0.9068,
-      "step": 398
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.216706470343579e-06,
-      "loss": 0.92,
-      "step": 399
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.20673229252686e-06,
-      "loss": 0.921,
-      "step": 400
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.196736390316412e-06,
-      "loss": 0.9338,
-      "step": 401
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.186718831430678e-06,
-      "loss": 0.9043,
-      "step": 402
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.176679683734814e-06,
-      "loss": 0.9327,
-      "step": 403
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.166619015240236e-06,
-      "loss": 0.9047,
-      "step": 404
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.156536894104151e-06,
-      "loss": 0.9013,
-      "step": 405
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.146433388629103e-06,
-      "loss": 0.9095,
-      "step": 406
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.136308567262507e-06,
-      "loss": 0.9034,
-      "step": 407
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.12616249859618e-06,
-      "loss": 0.8971,
-      "step": 408
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.115995251365887e-06,
-      "loss": 0.8751,
-      "step": 409
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.105806894450866e-06,
-      "loss": 0.8906,
-      "step": 410
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.095597496873368e-06,
-      "loss": 0.9146,
-      "step": 411
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.085367127798187e-06,
-      "loss": 0.9239,
-      "step": 412
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.075115856532186e-06,
-      "loss": 0.8862,
-      "step": 413
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.064843752523835e-06,
-      "loss": 0.8998,
-      "step": 414
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.054550885362742e-06,
-      "loss": 0.8887,
-      "step": 415
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.054550885362742e-06,
-      "loss": 0.9023,
-      "step": 416
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.044237324779174e-06,
-      "loss": 0.8777,
-      "step": 417
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.033903140643583e-06,
-      "loss": 0.8971,
-      "step": 418
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.023548402966148e-06,
-      "loss": 0.9298,
-      "step": 419
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.013173181896283e-06,
-      "loss": 0.8968,
-      "step": 420
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.00277754772217e-06,
-      "loss": 0.8565,
-      "step": 421
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.992361570870289e-06,
-      "loss": 0.892,
-      "step": 422
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.981925321904921e-06,
-      "loss": 0.9263,
-      "step": 423
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.971468871527694e-06,
-      "loss": 0.9182,
-      "step": 424
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.96099229057709e-06,
-      "loss": 0.8845,
-      "step": 425
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.950495650027965e-06,
-      "loss": 0.8989,
-      "step": 426
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.939979020991075e-06,
-      "loss": 0.9355,
-      "step": 427
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.929442474712587e-06,
-      "loss": 0.9147,
-      "step": 428
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 7.918886082573605e-06,
-      "loss": 0.9054,
-      "step": 429
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.908309916089673e-06,
-      "loss": 0.8821,
-      "step": 430
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.897714046910305e-06,
-      "loss": 0.8957,
-      "step": 431
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.897714046910305e-06,
-      "loss": 0.9261,
-      "step": 432
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.887098546818493e-06,
-      "loss": 0.9289,
-      "step": 433
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.876463487730218e-06,
-      "loss": 0.9206,
-      "step": 434
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.865808941693967e-06,
-      "loss": 0.9142,
-      "step": 435
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.855134980890247e-06,
-      "loss": 0.9447,
-      "step": 436
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.855134980890247e-06,
-      "loss": 0.8982,
-      "step": 437
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.844441677631085e-06,
-      "loss": 0.8983,
-      "step": 438
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.833729104359555e-06,
-      "loss": 0.9212,
-      "step": 439
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.822997333649271e-06,
-      "loss": 0.8663,
-      "step": 440
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.812246438203905e-06,
-      "loss": 0.8726,
-      "step": 441
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 7.80147649085669e-06,
-      "loss": 0.8989,
-      "step": 442
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.790687564569934e-06,
-      "loss": 0.9078,
-      "step": 443
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.790687564569934e-06,
-      "loss": 0.8898,
-      "step": 444
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.779879732434515e-06,
-      "loss": 0.878,
-      "step": 445
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.76905306766939e-06,
-      "loss": 0.9122,
-      "step": 446
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.758207643621106e-06,
-      "loss": 0.892,
-      "step": 447
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.747343533763287e-06,
-      "loss": 0.9085,
-      "step": 448
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.736460811696158e-06,
-      "loss": 0.8891,
-      "step": 449
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.725559551146026e-06,
-      "loss": 0.8974,
-      "step": 450
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.714639825964794e-06,
-      "loss": 0.9051,
-      "step": 451
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.703701710129451e-06,
-      "loss": 0.8989,
-      "step": 452
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.692745277741583e-06,
-      "loss": 0.8965,
-      "step": 453
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 7.681770603026856e-06,
-      "loss": 0.8801,
-      "step": 454
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.670777760334527e-06,
-      "loss": 0.9018,
-      "step": 455
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.659766824136928e-06,
-      "loss": 0.8823,
-      "step": 456
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.648737869028975e-06,
-      "loss": 0.8475,
-      "step": 457
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.637690969727647e-06,
-      "loss": 0.8958,
-      "step": 458
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.626626201071494e-06,
-      "loss": 0.8961,
-      "step": 459
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.615543638020125e-06,
-      "loss": 0.9136,
-      "step": 460
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.6044433556536945e-06,
-      "loss": 0.9029,
-      "step": 461
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.593325429172401e-06,
-      "loss": 0.9118,
-      "step": 462
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.582189933895976e-06,
-      "loss": 0.9015,
-      "step": 463
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.57103694526317e-06,
-      "loss": 0.8695,
-      "step": 464
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.559866538831248e-06,
-      "loss": 0.9029,
-      "step": 465
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.5486787902754735e-06,
-      "loss": 0.8878,
-      "step": 466
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 7.537473775388593e-06,
-      "loss": 0.9173,
-      "step": 467
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.526251570080327e-06,
-      "loss": 0.8604,
-      "step": 468
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.515012250376859e-06,
-      "loss": 0.9063,
-      "step": 469
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.515012250376859e-06,
-      "loss": 0.9009,
-      "step": 470
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.50375589242031e-06,
-      "loss": 0.9157,
-      "step": 471
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.492482572468233e-06,
-      "loss": 0.8932,
-      "step": 472
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.481192366893087e-06,
-      "loss": 0.9319,
-      "step": 473
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.469885352181733e-06,
-      "loss": 0.9015,
-      "step": 474
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.458561604934898e-06,
-      "loss": 0.9175,
-      "step": 475
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.447221201866674e-06,
-      "loss": 0.889,
-      "step": 476
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.4358642198039835e-06,
-      "loss": 0.9339,
-      "step": 477
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.4244907356860685e-06,
-      "loss": 0.9018,
-      "step": 478
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 7.413100826563965e-06,
-      "loss": 0.887,
-      "step": 479
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.401694569599982e-06,
-      "loss": 0.8921,
-      "step": 480
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.390272042067183e-06,
-      "loss": 0.8663,
-      "step": 481
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.378833321348852e-06,
-      "loss": 0.8599,
-      "step": 482
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.367378484937979e-06,
-      "loss": 0.8837,
-      "step": 483
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.367378484937979e-06,
-      "loss": 0.8997,
-      "step": 484
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.355907610436732e-06,
-      "loss": 0.907,
-      "step": 485
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.3444207755559315e-06,
-      "loss": 0.8898,
-      "step": 486
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.332918058114522e-06,
-      "loss": 0.8923,
-      "step": 487
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.321399536039046e-06,
-      "loss": 0.8711,
-      "step": 488
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.309865287363115e-06,
-      "loss": 0.9064,
-      "step": 489
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.298315390226889e-06,
-      "loss": 0.9404,
-      "step": 490
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.286749922876531e-06,
-      "loss": 0.9144,
-      "step": 491
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 7.275168963663696e-06,
-      "loss": 0.8771,
-      "step": 492
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.275168963663696e-06,
-      "loss": 0.878,
-      "step": 493
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.263572591044982e-06,
-      "loss": 0.8796,
-      "step": 494
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.25196088358141e-06,
-      "loss": 0.8957,
-      "step": 495
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.240333919937893e-06,
-      "loss": 0.8774,
-      "step": 496
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.2286917788826926e-06,
-      "loss": 0.882,
-      "step": 497
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.217034539286893e-06,
-      "loss": 0.8695,
-      "step": 498
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.2053622801238685e-06,
-      "loss": 0.8791,
-      "step": 499
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.193675080468743e-06,
-      "loss": 0.8882,
-      "step": 500
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.181973019497854e-06,
-      "loss": 0.9328,
-      "step": 501
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.1702561764882264e-06,
-      "loss": 0.8739,
-      "step": 502
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.158524630817017e-06,
-      "loss": 0.9061,
-      "step": 503
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 7.146778461960997e-06,
-      "loss": 0.9252,
-      "step": 504
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.135017749496001e-06,
-      "loss": 0.8864,
-      "step": 505
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.123242573096388e-06,
-      "loss": 0.912,
-      "step": 506
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.1114530125345115e-06,
-      "loss": 0.8918,
-      "step": 507
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.1114530125345115e-06,
-      "loss": 0.8691,
-      "step": 508
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.1114530125345115e-06,
-      "loss": 0.8933,
-      "step": 509
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.0996491476801635e-06,
-      "loss": 0.9224,
-      "step": 510
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.0878310585000475e-06,
-      "loss": 0.9019,
-      "step": 511
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.075998825057232e-06,
-      "loss": 0.8982,
-      "step": 512
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.0641525275106045e-06,
-      "loss": 0.9031,
-      "step": 513
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.052292246114335e-06,
-      "loss": 0.9089,
-      "step": 514
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.040418061217325e-06,
-      "loss": 0.8716,
-      "step": 515
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 7.02853005326267e-06,
-      "loss": 0.8693,
-      "step": 516
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 7.016628302787108e-06,
-      "loss": 0.8521,
-      "step": 517
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 7.004712890420483e-06,
-      "loss": 0.8608,
-      "step": 518
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.992783896885187e-06,
-      "loss": 0.8764,
-      "step": 519
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.980841402995623e-06,
-      "loss": 0.8827,
-      "step": 520
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.9688854896576504e-06,
-      "loss": 0.8895,
-      "step": 521
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.956916237868047e-06,
-      "loss": 0.9096,
-      "step": 522
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.944933728713945e-06,
-      "loss": 0.8856,
-      "step": 523
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.932938043372296e-06,
-      "loss": 0.8758,
-      "step": 524
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.920929263109314e-06,
-      "loss": 0.9149,
-      "step": 525
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.908907469279925e-06,
-      "loss": 0.9201,
-      "step": 526
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.896872743327219e-06,
-      "loss": 0.9316,
-      "step": 527
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.8848251667818925e-06,
-      "loss": 0.8971,
-      "step": 528
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 6.872764821261703e-06,
-      "loss": 0.909,
-      "step": 529
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.860691788470914e-06,
-      "loss": 0.9222,
-      "step": 530
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.8486061501997384e-06,
-      "loss": 0.9312,
-      "step": 531
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.836507988323785e-06,
-      "loss": 0.8967,
-      "step": 532
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.824397384803507e-06,
-      "loss": 0.8954,
-      "step": 533
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.812274421683648e-06,
-      "loss": 0.8779,
-      "step": 534
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.800139181092676e-06,
-      "loss": 0.8855,
-      "step": 535
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.7879917452422405e-06,
-      "loss": 0.9422,
-      "step": 536
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.775832196426608e-06,
-      "loss": 0.8835,
-      "step": 537
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.763660617022105e-06,
-      "loss": 0.8964,
-      "step": 538
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.75147708948656e-06,
-      "loss": 0.9095,
-      "step": 539
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.75147708948656e-06,
-      "loss": 0.9165,
-      "step": 540
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 6.739281696358745e-06,
-      "loss": 0.9082,
-      "step": 541
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.7270745202578184e-06,
-      "loss": 0.8863,
-      "step": 542
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.7148556438827625e-06,
-      "loss": 0.8764,
-      "step": 543
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.702625150011829e-06,
-      "loss": 0.8882,
-      "step": 544
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.6903831215019666e-06,
-      "loss": 0.8665,
-      "step": 545
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.678129641288272e-06,
-      "loss": 0.8874,
-      "step": 546
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.665864792383421e-06,
-      "loss": 0.9007,
-      "step": 547
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.653588657877108e-06,
-      "loss": 0.9311,
-      "step": 548
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.6413013209354835e-06,
-      "loss": 0.9113,
-      "step": 549
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.629002864800589e-06,
-      "loss": 0.9044,
-      "step": 550
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.616693372789795e-06,
-      "loss": 0.9073,
-      "step": 551
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.6043729282952374e-06,
-      "loss": 0.8814,
-      "step": 552
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.592041614783249e-06,
-      "loss": 0.9034,
-      "step": 553
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 6.5796995157937985e-06,
-      "loss": 0.8653,
-      "step": 554
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.567346714939917e-06,
-      "loss": 0.8652,
-      "step": 555
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.5549832959071435e-06,
-      "loss": 0.8564,
-      "step": 556
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.542609342452946e-06,
-      "loss": 0.8575,
-      "step": 557
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.53022493840616e-06,
-      "loss": 0.8635,
-      "step": 558
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.517830167666424e-06,
-      "loss": 0.8978,
-      "step": 559
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.5054251142036e-06,
-      "loss": 0.8986,
-      "step": 560
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.493009862057218e-06,
-      "loss": 0.8711,
-      "step": 561
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.480584495335896e-06,
-      "loss": 0.8756,
-      "step": 562
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.468149098216775e-06,
-      "loss": 0.8983,
-      "step": 563
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.45570375494495e-06,
-      "loss": 0.8629,
-      "step": 564
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.443248549832895e-06,
-      "loss": 0.8925,
-      "step": 565
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 6.4307835672599e-06,
-      "loss": 0.8753,
-      "step": 566
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.4307835672599e-06,
-      "loss": 0.9098,
-      "step": 567
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.418308891671484e-06,
-      "loss": 0.9146,
-      "step": 568
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.405824607578841e-06,
-      "loss": 0.855,
-      "step": 569
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.405824607578841e-06,
-      "loss": 0.9191,
-      "step": 570
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.393330799558253e-06,
-      "loss": 0.9064,
-      "step": 571
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.380827552250527e-06,
-      "loss": 0.8877,
-      "step": 572
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.380827552250527e-06,
-      "loss": 0.9,
-      "step": 573
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.368314950360416e-06,
-      "loss": 0.8792,
-      "step": 574
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.355793078656046e-06,
-      "loss": 0.8573,
-      "step": 575
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.343262021968345e-06,
-      "loss": 0.8839,
-      "step": 576
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.330721865190463e-06,
-      "loss": 0.9042,
-      "step": 577
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.318172693277202e-06,
-      "loss": 0.9061,
-      "step": 578
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 6.305614591244438e-06,
-      "loss": 0.8919,
-      "step": 579
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.29304764416854e-06,
-      "loss": 0.859,
-      "step": 580
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.280471937185808e-06,
-      "loss": 0.9029,
-      "step": 581
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.2678875554918776e-06,
-      "loss": 0.9273,
-      "step": 582
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.255294584341157e-06,
-      "loss": 0.8754,
-      "step": 583
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.242693109046249e-06,
-      "loss": 0.9063,
-      "step": 584
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.2300832149773585e-06,
-      "loss": 0.8989,
-      "step": 585
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.2300832149773585e-06,
-      "loss": 0.8815,
-      "step": 586
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.217464987561733e-06,
-      "loss": 0.8709,
-      "step": 587
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.204838512283073e-06,
-      "loss": 0.873,
-      "step": 588
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.192203874680952e-06,
-      "loss": 0.8927,
-      "step": 589
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.179561160350244e-06,
-      "loss": 0.9038,
-      "step": 590
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 6.166910454940538e-06,
-      "loss": 0.8999,
-      "step": 591
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.15425184415556e-06,
-      "loss": 0.898,
-      "step": 592
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.15425184415556e-06,
-      "loss": 0.877,
-      "step": 593
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.141585413752591e-06,
-      "loss": 0.878,
-      "step": 594
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.128911249541889e-06,
-      "loss": 0.8816,
-      "step": 595
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.116229437386103e-06,
-      "loss": 0.887,
-      "step": 596
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.103540063199697e-06,
-      "loss": 0.9211,
-      "step": 597
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.090843212948361e-06,
-      "loss": 0.8553,
-      "step": 598
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.078138972648435e-06,
-      "loss": 0.903,
-      "step": 599
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.065427428366325e-06,
-      "loss": 0.8657,
-      "step": 600
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.0527086662179166e-06,
-      "loss": 0.8913,
-      "step": 601
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.039982772367994e-06,
-      "loss": 0.899,
-      "step": 602
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.027249833029655e-06,
-      "loss": 0.8362,
-      "step": 603
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 6.027249833029655e-06,
-      "loss": 0.8839,
-      "step": 604
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 6.014509934463731e-06,
-      "loss": 0.8869,
-      "step": 605
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 6.001763162978196e-06,
-      "loss": 0.9152,
-      "step": 606
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.989009604927587e-06,
-      "loss": 0.9079,
-      "step": 607
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.976249346712418e-06,
-      "loss": 0.9086,
-      "step": 608
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.963482474778595e-06,
-      "loss": 0.8914,
-      "step": 609
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.950709075616826e-06,
-      "loss": 0.877,
-      "step": 610
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.9379292357620414e-06,
-      "loss": 0.9008,
-      "step": 611
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.925143041792805e-06,
-      "loss": 0.8818,
-      "step": 612
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.9123505803307225e-06,
-      "loss": 0.8598,
-      "step": 613
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.899551938039866e-06,
-      "loss": 0.8964,
-      "step": 614
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.886747201626176e-06,
-      "loss": 0.9058,
-      "step": 615
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 5.8739364578368815e-06,
-      "loss": 0.8745,
-      "step": 616
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.861119793459908e-06,
-      "loss": 0.8539,
-      "step": 617
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.848297295323288e-06,
-      "loss": 0.8922,
-      "step": 618
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.83546905029458e-06,
-      "loss": 0.9016,
-      "step": 619
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.822635145280273e-06,
-      "loss": 0.8839,
-      "step": 620
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.822635145280273e-06,
-      "loss": 0.8855,
-      "step": 621
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.8097956672252e-06,
-      "loss": 0.9035,
-      "step": 622
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.796950703111949e-06,
-      "loss": 0.8817,
-      "step": 623
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.784100339960276e-06,
-      "loss": 0.8706,
-      "step": 624
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.771244664826512e-06,
-      "loss": 0.8964,
-      "step": 625
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.758383764802975e-06,
-      "loss": 0.8886,
-      "step": 626
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.74551772701738e-06,
-      "loss": 0.8416,
-      "step": 627
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.732646638632247e-06,
-      "loss": 0.8846,
-      "step": 628
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 5.732646638632247e-06,
-      "loss": 0.8833,
-      "step": 629
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.7197705868443145e-06,
-      "loss": 0.8806,
-      "step": 630
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.706889658883941e-06,
-      "loss": 0.8787,
-      "step": 631
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.694003942014525e-06,
-      "loss": 0.9382,
-      "step": 632
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.681113523531909e-06,
-      "loss": 0.8592,
-      "step": 633
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.668218490763778e-06,
-      "loss": 0.8877,
-      "step": 634
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.6553189310690874e-06,
-      "loss": 0.9039,
-      "step": 635
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.642414931837452e-06,
-      "loss": 0.878,
-      "step": 636
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.629506580488571e-06,
-      "loss": 0.9106,
-      "step": 637
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.616593964471623e-06,
-      "loss": 0.8546,
-      "step": 638
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.6036771712646755e-06,
-      "loss": 0.8897,
-      "step": 639
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.590756288374103e-06,
-      "loss": 0.8776,
-      "step": 640
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 5.577831403333977e-06,
-      "loss": 0.8884,
-      "step": 641
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.564902603705491e-06,
-      "loss": 0.8471,
-      "step": 642
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.55196997707635e-06,
-      "loss": 0.8257,
-      "step": 643
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.539033611060189e-06,
-      "loss": 0.8889,
-      "step": 644
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.526093593295977e-06,
-      "loss": 0.8903,
-      "step": 645
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.51315001144742e-06,
-      "loss": 0.8429,
-      "step": 646
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.500202953202373e-06,
-      "loss": 0.8917,
-      "step": 647
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.487252506272238e-06,
-      "loss": 0.8639,
-      "step": 648
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.474298758391375e-06,
-      "loss": 0.8941,
-      "step": 649
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.46134179731651e-06,
-      "loss": 0.8127,
-      "step": 650
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.448381710826134e-06,
-      "loss": 0.8743,
-      "step": 651
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.43541858671991e-06,
-      "loss": 0.8406,
-      "step": 652
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 5.422452512818085e-06,
-      "loss": 0.8862,
-      "step": 653
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.4094835769608845e-06,
-      "loss": 0.8991,
-      "step": 654
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.3965118670079256e-06,
-      "loss": 0.8762,
-      "step": 655
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.383537470837617e-06,
-      "loss": 0.9043,
-      "step": 656
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.370560476346568e-06,
-      "loss": 0.9134,
-      "step": 657
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.357580971448989e-06,
-      "loss": 0.8619,
-      "step": 658
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.3445990440760966e-06,
-      "loss": 0.8905,
-      "step": 659
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.33161478217552e-06,
-      "loss": 0.8673,
-      "step": 660
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.3186282737107044e-06,
-      "loss": 0.8487,
-      "step": 661
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.305639606660313e-06,
-      "loss": 0.8678,
-      "step": 662
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.292648869017634e-06,
-      "loss": 0.8384,
-      "step": 663
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.279656148789982e-06,
-      "loss": 0.8881,
-      "step": 664
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.266661533998103e-06,
-      "loss": 0.8935,
-      "step": 665
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 5.25366511267558e-06,
-      "loss": 0.8329,
-      "step": 666
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.240666972868231e-06,
-      "loss": 0.8946,
-      "step": 667
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.227667202633518e-06,
-      "loss": 0.918,
-      "step": 668
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.21466589003995e-06,
-      "loss": 0.8719,
-      "step": 669
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.201663123166484e-06,
-      "loss": 0.8943,
-      "step": 670
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.188658990101926e-06,
-      "loss": 0.8395,
-      "step": 671
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.175653578944343e-06,
-      "loss": 0.8476,
-      "step": 672
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.162646977800454e-06,
-      "loss": 0.8474,
-      "step": 673
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.149639274785047e-06,
-      "loss": 0.8625,
-      "step": 674
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.136630558020367e-06,
-      "loss": 0.9026,
-      "step": 675
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.123620915635533e-06,
-      "loss": 0.8499,
-      "step": 676
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.110610435765935e-06,
-      "loss": 0.8834,
-      "step": 677
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 5.110610435765935e-06,
-      "loss": 0.9014,
-      "step": 678
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.097599206552631e-06,
-      "loss": 0.8807,
-      "step": 679
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.08458731614176e-06,
-      "loss": 0.8418,
-      "step": 680
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.071574852683938e-06,
-      "loss": 0.8619,
-      "step": 681
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.058561904333667e-06,
-      "loss": 0.9048,
-      "step": 682
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.045548559248728e-06,
-      "loss": 0.8762,
-      "step": 683
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.032534905589593e-06,
-      "loss": 0.8862,
-      "step": 684
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.019521031518827e-06,
-      "loss": 0.9432,
-      "step": 685
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 5.006507025200484e-06,
-      "loss": 0.8724,
-      "step": 686
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.993492974799516e-06,
-      "loss": 0.8876,
-      "step": 687
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.980478968481173e-06,
-      "loss": 0.8384,
-      "step": 688
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.967465094410407e-06,
-      "loss": 0.903,
-      "step": 689
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.954451440751273e-06,
-      "loss": 0.8932,
-      "step": 690
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 4.941438095666336e-06,
-      "loss": 0.8772,
-      "step": 691
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.941438095666336e-06,
-      "loss": 0.8829,
-      "step": 692
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.928425147316063e-06,
-      "loss": 0.8989,
-      "step": 693
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.915412683858242e-06,
-      "loss": 0.8504,
-      "step": 694
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.90240079344737e-06,
-      "loss": 0.88,
-      "step": 695
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.8893895642340665e-06,
-      "loss": 0.9037,
-      "step": 696
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.876379084364468e-06,
-      "loss": 0.8526,
-      "step": 697
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.863369441979635e-06,
-      "loss": 0.9073,
-      "step": 698
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.8503607252149565e-06,
-      "loss": 0.9173,
-      "step": 699
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.837353022199548e-06,
-      "loss": 0.8623,
-      "step": 700
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.8243464210556605e-06,
-      "loss": 0.8496,
-      "step": 701
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.811341009898076e-06,
-      "loss": 0.8762,
-      "step": 702
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 4.798336876833517e-06,
-      "loss": 0.9179,
-      "step": 703
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.78533410996005e-06,
-      "loss": 0.8681,
-      "step": 704
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.772332797366483e-06,
-      "loss": 0.8438,
-      "step": 705
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.75933302713177e-06,
-      "loss": 0.8779,
-      "step": 706
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.746334887324421e-06,
-      "loss": 0.8968,
-      "step": 707
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.733338466001898e-06,
-      "loss": 0.8947,
-      "step": 708
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.7203438512100196e-06,
-      "loss": 0.8646,
-      "step": 709
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.707351130982367e-06,
-      "loss": 0.8338,
-      "step": 710
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.694360393339688e-06,
-      "loss": 0.8586,
-      "step": 711
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.681371726289296e-06,
-      "loss": 0.8641,
-      "step": 712
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.668385217824482e-06,
-      "loss": 0.8905,
-      "step": 713
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.655400955923905e-06,
-      "loss": 0.8516,
-      "step": 714
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.642419028551013e-06,
-      "loss": 0.8927,
-      "step": 715
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 4.629439523653434e-06,
-      "loss": 0.888,
-      "step": 716
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.6164625291623835e-06,
-      "loss": 0.8644,
-      "step": 717
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.603488132992075e-06,
-      "loss": 0.8834,
-      "step": 718
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.590516423039116e-06,
-      "loss": 0.868,
-      "step": 719
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.590516423039116e-06,
-      "loss": 0.8541,
-      "step": 720
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.577547487181916e-06,
-      "loss": 0.8945,
-      "step": 721
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.564581413280091e-06,
-      "loss": 0.8734,
-      "step": 722
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.551618289173867e-06,
-      "loss": 0.873,
-      "step": 723
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.53865820268349e-06,
-      "loss": 0.8534,
-      "step": 724
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.525701241608626e-06,
-      "loss": 0.9209,
-      "step": 725
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.512747493727764e-06,
-      "loss": 0.8965,
-      "step": 726
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.499797046797629e-06,
-      "loss": 0.8805,
-      "step": 727
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 4.486849988552581e-06,
-      "loss": 0.8586,
-      "step": 728
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.473906406704025e-06,
-      "loss": 0.8702,
-      "step": 729
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.460966388939813e-06,
-      "loss": 0.8761,
-      "step": 730
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.4480300229236525e-06,
-      "loss": 0.8468,
-      "step": 731
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.435097396294511e-06,
-      "loss": 0.8565,
-      "step": 732
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.422168596666022e-06,
-      "loss": 0.8463,
-      "step": 733
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.409243711625898e-06,
-      "loss": 0.8992,
-      "step": 734
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.3963228287353245e-06,
-      "loss": 0.8979,
-      "step": 735
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.383406035528379e-06,
-      "loss": 0.8587,
-      "step": 736
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.37049341951143e-06,
-      "loss": 0.8871,
-      "step": 737
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.357585068162549e-06,
-      "loss": 0.9029,
-      "step": 738
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.344681068930915e-06,
-      "loss": 0.8956,
-      "step": 739
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.3317815092362234e-06,
-      "loss": 0.894,
-      "step": 740
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 4.318886476468092e-06,
-      "loss": 0.886,
-      "step": 741
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.305996057985476e-06,
-      "loss": 0.8986,
-      "step": 742
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.293110341116061e-06,
-      "loss": 0.8537,
-      "step": 743
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.280229413155689e-06,
-      "loss": 0.8578,
-      "step": 744
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.267353361367755e-06,
-      "loss": 0.8899,
-      "step": 745
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.254482272982622e-06,
-      "loss": 0.8833,
-      "step": 746
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.241616235197025e-06,
-      "loss": 0.8929,
-      "step": 747
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.228755335173488e-06,
-      "loss": 0.859,
-      "step": 748
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.215899660039725e-06,
-      "loss": 0.8778,
-      "step": 749
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.2030492968880526e-06,
-      "loss": 0.8937,
-      "step": 750
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.190204332774802e-06,
-      "loss": 0.8846,
-      "step": 751
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.177364854719729e-06,
-      "loss": 0.8968,
-      "step": 752
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 4.164530949705421e-06,
-      "loss": 0.8607,
-      "step": 753
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.1517027046767125e-06,
-      "loss": 0.8776,
-      "step": 754
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.1388802065400935e-06,
-      "loss": 0.855,
-      "step": 755
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.126063542163119e-06,
-      "loss": 0.8785,
-      "step": 756
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.1132527983738254e-06,
-      "loss": 0.8236,
-      "step": 757
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.100448061960136e-06,
-      "loss": 0.8213,
-      "step": 758
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.087649419669279e-06,
-      "loss": 0.8633,
-      "step": 759
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.0748569582071975e-06,
-      "loss": 0.9179,
-      "step": 760
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.062070764237959e-06,
-      "loss": 0.9195,
-      "step": 761
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.049290924383174e-06,
-      "loss": 0.8432,
-      "step": 762
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.036517525221405e-06,
-      "loss": 0.849,
-      "step": 763
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.023750653287582e-06,
-      "loss": 0.8606,
-      "step": 764
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 4.010990395072414e-06,
-      "loss": 0.857,
-      "step": 765
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.998236837021806e-06,
-      "loss": 0.8662,
-      "step": 766
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.98549006553627e-06,
-      "loss": 0.8808,
-      "step": 767
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.972750166970346e-06,
-      "loss": 0.875,
-      "step": 768
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.960017227632008e-06,
-      "loss": 0.8878,
-      "step": 769
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.947291333782084e-06,
-      "loss": 0.8853,
-      "step": 770
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.9345725716336765e-06,
-      "loss": 0.8817,
-      "step": 771
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.9345725716336765e-06,
-      "loss": 0.8977,
-      "step": 772
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.921861027351567e-06,
-      "loss": 0.8481,
-      "step": 773
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.921861027351567e-06,
-      "loss": 0.8818,
-      "step": 774
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.909156787051642e-06,
-      "loss": 0.8674,
-      "step": 775
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.896459936800306e-06,
-      "loss": 0.8597,
-      "step": 776
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.8837705626138995e-06,
-      "loss": 0.8395,
-      "step": 777
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 3.871088750458112e-06,
-      "loss": 0.8483,
-      "step": 778
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.858414586247409e-06,
-      "loss": 0.8354,
-      "step": 779
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.845748155844441e-06,
-      "loss": 0.8513,
-      "step": 780
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.845748155844441e-06,
-      "loss": 0.869,
-      "step": 781
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.833089545059464e-06,
-      "loss": 0.8607,
-      "step": 782
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.820438839649758e-06,
-      "loss": 0.8286,
-      "step": 783
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.80779612531905e-06,
-      "loss": 0.8661,
-      "step": 784
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7951614877169285e-06,
-      "loss": 0.87,
-      "step": 785
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7825350124382677e-06,
-      "loss": 0.8462,
-      "step": 786
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7699167850226428e-06,
-      "loss": 0.8454,
-      "step": 787
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7573068909537534e-06,
-      "loss": 0.8695,
-      "step": 788
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7447054156588436e-06,
-      "loss": 0.8528,
-      "step": 789
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.7321124445081245e-06,
-      "loss": 0.8642,
-      "step": 790
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.719528062814195e-06,
-      "loss": 0.8708,
-      "step": 791
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.719528062814195e-06,
-      "loss": 0.885,
-      "step": 792
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.7069523558314614e-06,
-      "loss": 0.8518,
-      "step": 793
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.694385408755564e-06,
-      "loss": 0.8513,
-      "step": 794
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6818273067227976e-06,
-      "loss": 0.8387,
-      "step": 795
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6692781348095368e-06,
-      "loss": 0.8654,
-      "step": 796
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6567379780316557e-06,
-      "loss": 0.9041,
-      "step": 797
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6442069213439545e-06,
-      "loss": 0.8549,
-      "step": 798
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6316850496395863e-06,
-      "loss": 0.8757,
-      "step": 799
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6191724477494748e-06,
-      "loss": 0.834,
-      "step": 800
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.6066692004417486e-06,
-      "loss": 0.8431,
-      "step": 801
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.594175392421161e-06,
-      "loss": 0.8753,
-      "step": 802
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 3.5816911083285165e-06,
-      "loss": 0.8619,
-      "step": 803
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.5692164327401026e-06,
-      "loss": 0.8844,
-      "step": 804
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.556751450167106e-06,
-      "loss": 0.8779,
-      "step": 805
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.5442962450550524e-06,
-      "loss": 0.8721,
-      "step": 806
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.531850901783228e-06,
-      "loss": 0.8672,
-      "step": 807
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.5194155046641078e-06,
-      "loss": 0.8519,
-      "step": 808
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.506990137942783e-06,
-      "loss": 0.8753,
-      "step": 809
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.4945748857964e-06,
-      "loss": 0.8447,
-      "step": 810
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.482169832333578e-06,
-      "loss": 0.9121,
-      "step": 811
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.4697750615938407e-06,
-      "loss": 0.864,
-      "step": 812
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.4573906575470562e-06,
-      "loss": 0.8792,
-      "step": 813
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.445016704092858e-06,
-      "loss": 0.8788,
-      "step": 814
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.4326532850600843e-06,
-      "loss": 0.8903,
-      "step": 815
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.4203004842062036e-06,
-      "loss": 0.8964,
-      "step": 816
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.407958385216752e-06,
-      "loss": 0.862,
-      "step": 817
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.3956270717047642e-06,
-      "loss": 0.8714,
-      "step": 818
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.3833066272102068e-06,
-      "loss": 0.8948,
-      "step": 819
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.370997135199413e-06,
-      "loss": 0.8738,
-      "step": 820
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.3586986790645186e-06,
-      "loss": 0.8842,
-      "step": 821
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.346411342122894e-06,
-      "loss": 0.841,
-      "step": 822
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.33413520761658e-06,
-      "loss": 0.8883,
-      "step": 823
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.3218703587117285e-06,
-      "loss": 0.8624,
-      "step": 824
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.309616878498034e-06,
-      "loss": 0.8977,
-      "step": 825
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.297374849988172e-06,
-      "loss": 0.8683,
-      "step": 826
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.2851443561172375e-06,
-      "loss": 0.8482,
-      "step": 827
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 3.2729254797421832e-06,
-      "loss": 0.9168,
-      "step": 828
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.2607183036412573e-06,
-      "loss": 0.8677,
-      "step": 829
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.248522910513442e-06,
-      "loss": 0.843,
-      "step": 830
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.2363393829778955e-06,
-      "loss": 0.8662,
-      "step": 831
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.2241678035733924e-06,
-      "loss": 0.868,
-      "step": 832
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.2120082547577603e-06,
-      "loss": 0.8724,
-      "step": 833
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.199860818907326e-06,
-      "loss": 0.8518,
-      "step": 834
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.187725578316355e-06,
-      "loss": 0.8677,
-      "step": 835
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.1756026151964947e-06,
-      "loss": 0.8493,
-      "step": 836
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.1634920116762175e-06,
-      "loss": 0.846,
-      "step": 837
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.1513938498002615e-06,
-      "loss": 0.8801,
-      "step": 838
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.139308211529085e-06,
-      "loss": 0.8669,
-      "step": 839
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 3.127235178738297e-06,
-      "loss": 0.8951,
-      "step": 840
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.1151748332181088e-06,
-      "loss": 0.8686,
-      "step": 841
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.1031272566727828e-06,
-      "loss": 0.8726,
-      "step": 842
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.0910925307200756e-06,
-      "loss": 0.9013,
-      "step": 843
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.079070736890687e-06,
-      "loss": 0.8863,
-      "step": 844
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.067061956627705e-06,
-      "loss": 0.912,
-      "step": 845
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.055066271286057e-06,
-      "loss": 0.8724,
-      "step": 846
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.0430837621319553e-06,
-      "loss": 0.8581,
-      "step": 847
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.031114510342351e-06,
-      "loss": 0.8994,
-      "step": 848
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.019158597004379e-06,
-      "loss": 0.8761,
-      "step": 849
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.007216103114815e-06,
-      "loss": 0.841,
-      "step": 850
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.995287109579519e-06,
-      "loss": 0.8615,
-      "step": 851
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.9833716972128922e-06,
-      "loss": 0.8803,
-      "step": 852
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 2.9833716972128922e-06,
-      "loss": 0.8817,
-      "step": 853
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.971469946737331e-06,
-      "loss": 0.8259,
-      "step": 854
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.9595819387826753e-06,
-      "loss": 0.8601,
-      "step": 855
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.947707753885666e-06,
-      "loss": 0.8745,
-      "step": 856
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.935847472489396e-06,
-      "loss": 0.9127,
-      "step": 857
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.924001174942769e-06,
-      "loss": 0.8566,
-      "step": 858
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.9121689414999533e-06,
-      "loss": 0.8908,
-      "step": 859
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.900350852319838e-06,
-      "loss": 0.8992,
-      "step": 860
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.900350852319838e-06,
-      "loss": 0.8929,
-      "step": 861
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.88854698746549e-06,
-      "loss": 0.8425,
-      "step": 862
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.8767574269036114e-06,
-      "loss": 0.8478,
-      "step": 863
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.8649822505039997e-06,
-      "loss": 0.8273,
-      "step": 864
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 2.8532215380390034e-06,
-      "loss": 0.8526,
-      "step": 865
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.841475369182984e-06,
-      "loss": 0.8658,
-      "step": 866
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.8297438235117756e-06,
-      "loss": 0.8635,
-      "step": 867
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.8180269805021465e-06,
-      "loss": 0.8712,
-      "step": 868
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.8063249195312608e-06,
-      "loss": 0.8681,
-      "step": 869
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.794637719876133e-06,
-      "loss": 0.863,
-      "step": 870
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.782965460713108e-06,
-      "loss": 0.8525,
-      "step": 871
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.771308221117309e-06,
-      "loss": 0.8672,
-      "step": 872
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.7596660800621076e-06,
-      "loss": 0.8686,
-      "step": 873
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.7480391164185903e-06,
-      "loss": 0.8384,
-      "step": 874
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.7364274089550196e-06,
-      "loss": 0.8616,
-      "step": 875
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.7248310363363056e-06,
-      "loss": 0.86,
-      "step": 876
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.7132500771234694e-06,
-      "loss": 0.8514,
-      "step": 877
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 2.701684609773113e-06,
-      "loss": 0.8592,
-      "step": 878
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6901347126368858e-06,
-      "loss": 0.863,
-      "step": 879
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6786004639609565e-06,
-      "loss": 0.8383,
-      "step": 880
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.66708194188548e-06,
-      "loss": 0.8345,
-      "step": 881
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6555792244440694e-06,
-      "loss": 0.8863,
-      "step": 882
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6440923895632685e-06,
-      "loss": 0.8786,
-      "step": 883
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6326215150620204e-06,
-      "loss": 0.9029,
-      "step": 884
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.6211666786511503e-06,
-      "loss": 0.8397,
-      "step": 885
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.609727957932819e-06,
-      "loss": 0.8621,
-      "step": 886
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.5983054304000187e-06,
-      "loss": 0.9083,
-      "step": 887
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.5868991734360373e-06,
-      "loss": 0.8816,
-      "step": 888
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.5868991734360373e-06,
-      "loss": 0.8691,
-      "step": 889
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 2.575509264313934e-06,
-      "loss": 0.8791,
-      "step": 890
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.5641357801960186e-06,
-      "loss": 0.9034,
-      "step": 891
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.5527787981333275e-06,
-      "loss": 0.8379,
-      "step": 892
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.541438395065103e-06,
-      "loss": 0.925,
-      "step": 893
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.5301146478182686e-06,
-      "loss": 0.84,
-      "step": 894
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.5188076331069133e-06,
-      "loss": 0.8509,
-      "step": 895
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.5075174275317693e-06,
-      "loss": 0.857,
-      "step": 896
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.496244107579692e-06,
-      "loss": 0.8354,
-      "step": 897
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.496244107579692e-06,
-      "loss": 0.8622,
-      "step": 898
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.484987749623143e-06,
-      "loss": 0.8946,
-      "step": 899
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.4737484299196744e-06,
-      "loss": 0.8619,
-      "step": 900
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.462526224611408e-06,
-      "loss": 0.864,
-      "step": 901
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.451321209724527e-06,
-      "loss": 0.9043,
-      "step": 902
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 2.4401334611687517e-06,
-      "loss": 0.8348,
-      "step": 903
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.4289630547368305e-06,
-      "loss": 0.8678,
-      "step": 904
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.417810066104025e-06,
-      "loss": 0.8596,
-      "step": 905
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.406674570827599e-06,
-      "loss": 0.8989,
-      "step": 906
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.395556644346306e-06,
-      "loss": 0.8651,
-      "step": 907
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.395556644346306e-06,
-      "loss": 0.8733,
-      "step": 908
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.384456361979877e-06,
-      "loss": 0.8672,
-      "step": 909
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.373373798928507e-06,
-      "loss": 0.8583,
-      "step": 910
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.3623090302723556e-06,
-      "loss": 0.8699,
-      "step": 911
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.3512621309710283e-06,
-      "loss": 0.8525,
-      "step": 912
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.3402331758630734e-06,
-      "loss": 0.8725,
-      "step": 913
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.3292222396654756e-06,
-      "loss": 0.8906,
-      "step": 914
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 2.318229396973145e-06,
-      "loss": 0.8509,
-      "step": 915
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.307254722258417e-06,
-      "loss": 0.8781,
-      "step": 916
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.307254722258417e-06,
-      "loss": 0.8555,
-      "step": 917
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.307254722258417e-06,
-      "loss": 0.8478,
-      "step": 918
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.2962982898705483e-06,
-      "loss": 0.849,
-      "step": 919
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.2962982898705483e-06,
-      "loss": 0.8896,
-      "step": 920
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.2962982898705483e-06,
-      "loss": 0.8509,
-      "step": 921
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.2853601740352064e-06,
-      "loss": 0.8955,
-      "step": 922
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.274440448853974e-06,
-      "loss": 0.8307,
-      "step": 923
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.263539188303842e-06,
-      "loss": 0.8773,
-      "step": 924
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.252656466236713e-06,
-      "loss": 0.8499,
-      "step": 925
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.252656466236713e-06,
-      "loss": 0.8314,
-      "step": 926
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 2.2417923563788957e-06,
-      "loss": 0.8724,
-      "step": 927
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.2309469323306098e-06,
-      "loss": 0.8553,
-      "step": 928
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.220120267565485e-06,
-      "loss": 0.8641,
-      "step": 929
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.209312435430066e-06,
-      "loss": 0.895,
-      "step": 930
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1985235091433098e-06,
-      "loss": 0.834,
-      "step": 931
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.187753561796097e-06,
-      "loss": 0.8355,
-      "step": 932
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.177002666350731e-06,
-      "loss": 0.8968,
-      "step": 933
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1662708956404466e-06,
-      "loss": 0.8777,
-      "step": 934
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1555583223689175e-06,
-      "loss": 0.8748,
-      "step": 935
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1448650191097546e-06,
-      "loss": 0.8649,
-      "step": 936
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1341910583060336e-06,
-      "loss": 0.8518,
-      "step": 937
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1235365122697834e-06,
-      "loss": 0.8705,
-      "step": 938
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.1129014531815084e-06,
-      "loss": 0.868,
-      "step": 939
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 2.102285953089696e-06,
-      "loss": 0.8421,
-      "step": 940
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0916900839103283e-06,
-      "loss": 0.9007,
-      "step": 941
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0811139174263967e-06,
-      "loss": 0.8751,
-      "step": 942
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0705575252874128e-06,
-      "loss": 0.8778,
-      "step": 943
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0600209790089258e-06,
-      "loss": 0.8681,
-      "step": 944
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0495043499720356e-06,
-      "loss": 0.8513,
-      "step": 945
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.039007709422911e-06,
-      "loss": 0.8957,
-      "step": 946
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.028531128472307e-06,
-      "loss": 0.8711,
-      "step": 947
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0180746780950803e-06,
-      "loss": 0.8791,
-      "step": 948
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 2.0076384291297134e-06,
-      "loss": 0.8848,
-      "step": 949
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.9972224522778304e-06,
-      "loss": 0.8526,
-      "step": 950
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.9868268181037186e-06,
-      "loss": 0.8488,
-      "step": 951
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 1.976451597033854e-06,
-      "loss": 0.8732,
-      "step": 952
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.966096859356418e-06,
-      "loss": 0.8146,
-      "step": 953
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.9557626752208285e-06,
-      "loss": 0.839,
-      "step": 954
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.945449114637259e-06,
-      "loss": 0.8712,
-      "step": 955
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.935156247476166e-06,
-      "loss": 0.8578,
-      "step": 956
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.924884143467817e-06,
-      "loss": 0.8183,
-      "step": 957
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.914632872201816e-06,
-      "loss": 0.8492,
-      "step": 958
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.9044025031266333e-06,
-      "loss": 0.8693,
-      "step": 959
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.8941931055491364e-06,
-      "loss": 0.8638,
-      "step": 960
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.8840047486341161e-06,
-      "loss": 0.898,
-      "step": 961
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.873837501403823e-06,
-      "loss": 0.9038,
-      "step": 962
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.863691432737496e-06,
-      "loss": 0.8797,
-      "step": 963
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.8535666113708984e-06,
-      "loss": 0.864,
-      "step": 964
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 1.843463105895849e-06,
-      "loss": 0.8521,
-      "step": 965
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.8333809847597644e-06,
-      "loss": 0.8779,
-      "step": 966
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.8233203162651863e-06,
-      "loss": 0.8859,
-      "step": 967
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.813281168569323e-06,
-      "loss": 0.8418,
-      "step": 968
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.8032636096835888e-06,
-      "loss": 0.8636,
-      "step": 969
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7932677074731413e-06,
-      "loss": 0.8902,
-      "step": 970
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7832935296564246e-06,
-      "loss": 0.862,
-      "step": 971
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7733411438047032e-06,
-      "loss": 0.8553,
-      "step": 972
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7733411438047032e-06,
-      "loss": 0.8721,
-      "step": 973
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7634106173416137e-06,
-      "loss": 0.8355,
-      "step": 974
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7535020175427025e-06,
-      "loss": 0.872,
-      "step": 975
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.7436154115349712e-06,
-      "loss": 0.8774,
-      "step": 976
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 1.733750866296422e-06,
-      "loss": 0.8561,
-      "step": 977
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.723908448655603e-06,
-      "loss": 0.8406,
-      "step": 978
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.7140882252911568e-06,
-      "loss": 0.9005,
-      "step": 979
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.7042902627313661e-06,
-      "loss": 0.8729,
-      "step": 980
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.694514627353711e-06,
-      "loss": 0.885,
-      "step": 981
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6847613853844074e-06,
-      "loss": 0.8511,
-      "step": 982
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6750306028979668e-06,
-      "loss": 0.8632,
-      "step": 983
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.665322345816746e-06,
-      "loss": 0.8498,
-      "step": 984
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6556366799105005e-06,
-      "loss": 0.8486,
-      "step": 985
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6459736707959388e-06,
-      "loss": 0.8983,
-      "step": 986
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6363333839362793e-06,
-      "loss": 0.826,
-      "step": 987
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6267158846408048e-06,
-      "loss": 0.8531,
-      "step": 988
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.6171212380644218e-06,
-      "loss": 0.8583,
-      "step": 989
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 1.607549509207218e-06,
-      "loss": 0.8627,
-      "step": 990
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5980007629140216e-06,
-      "loss": 0.8499,
-      "step": 991
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5884750638739633e-06,
-      "loss": 0.8758,
-      "step": 992
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5789724766200393e-06,
-      "loss": 0.9033,
-      "step": 993
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5694930655286666e-06,
-      "loss": 0.8444,
-      "step": 994
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5600368948192567e-06,
-      "loss": 0.8764,
-      "step": 995
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5506040285537783e-06,
-      "loss": 0.8327,
-      "step": 996
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5506040285537783e-06,
-      "loss": 0.8482,
-      "step": 997
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.541194530636318e-06,
-      "loss": 0.8943,
-      "step": 998
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.5318084648126507e-06,
-      "loss": 0.8827,
-      "step": 999
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.522445894669809e-06,
-      "loss": 0.8618,
-      "step": 1000
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.513106883635651e-06,
-      "loss": 0.8544,
-      "step": 1001
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 1.50379149497843e-06,
-      "loss": 0.8608,
-      "step": 1002
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4944997918063676e-06,
-      "loss": 0.8589,
-      "step": 1003
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4944997918063676e-06,
-      "loss": 0.8861,
-      "step": 1004
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4852318370672254e-06,
-      "loss": 0.8573,
-      "step": 1005
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4759876935478767e-06,
-      "loss": 0.8672,
-      "step": 1006
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4667674238738843e-06,
-      "loss": 0.8419,
-      "step": 1007
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.457571090509075e-06,
-      "loss": 0.8905,
-      "step": 1008
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4483987557551155e-06,
-      "loss": 0.8493,
-      "step": 1009
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4392504817510906e-06,
-      "loss": 0.8514,
-      "step": 1010
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4301263304730834e-06,
-      "loss": 0.8825,
-      "step": 1011
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4210263637337546e-06,
-      "loss": 0.8466,
-      "step": 1012
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.411950643181924e-06,
-      "loss": 0.908,
-      "step": 1013
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.4028992303021522e-06,
-      "loss": 0.8549,
-      "step": 1014
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 1.3938721864143245e-06,
-      "loss": 0.8452,
-      "step": 1015
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3848695726732364e-06,
-      "loss": 0.8326,
-      "step": 1016
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.375891450068177e-06,
-      "loss": 0.876,
-      "step": 1017
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3669378794225185e-06,
-      "loss": 0.862,
-      "step": 1018
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3580089213933024e-06,
-      "loss": 0.8774,
-      "step": 1019
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3491046364708294e-06,
-      "loss": 0.8618,
-      "step": 1020
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.340225084978249e-06,
-      "loss": 0.8559,
-      "step": 1021
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3313703270711509e-06,
-      "loss": 0.8364,
-      "step": 1022
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3225404227371575e-06,
-      "loss": 0.8493,
-      "step": 1023
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3137354317955193e-06,
-      "loss": 0.8508,
-      "step": 1024
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.3049554138967052e-06,
-      "loss": 0.8646,
-      "step": 1025
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2962004285220025e-06,
-      "loss": 0.8343,
-      "step": 1026
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 1.2874705349831158e-06,
-      "loss": 0.8482,
-      "step": 1027
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2787657924217573e-06,
-      "loss": 0.8517,
-      "step": 1028
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2700862598092529e-06,
-      "loss": 0.8864,
-      "step": 1029
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2614319959461403e-06,
-      "loss": 0.8474,
-      "step": 1030
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2528030594617734e-06,
-      "loss": 0.8743,
-      "step": 1031
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2441995088139176e-06,
-      "loss": 0.911,
-      "step": 1032
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2356214022883622e-06,
-      "loss": 0.8357,
-      "step": 1033
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2270687979985236e-06,
-      "loss": 0.8643,
-      "step": 1034
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2185417538850474e-06,
-      "loss": 0.8755,
-      "step": 1035
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2100403277154198e-06,
-      "loss": 0.8895,
-      "step": 1036
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.2015645770835765e-06,
-      "loss": 0.8663,
-      "step": 1037
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1931145594095083e-06,
-      "loss": 0.853,
-      "step": 1038
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1846903319388774e-06,
-      "loss": 0.8613,
-      "step": 1039
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 1.1762919517426253e-06,
-      "loss": 0.8473,
-      "step": 1040
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1679194757165912e-06,
-      "loss": 0.936,
-      "step": 1041
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1595729605811201e-06,
-      "loss": 0.8502,
-      "step": 1042
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.151252462880683e-06,
-      "loss": 0.8828,
-      "step": 1043
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1429580389834926e-06,
-      "loss": 0.8524,
-      "step": 1044
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1346897450811213e-06,
-      "loss": 0.8832,
-      "step": 1045
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1264476371881216e-06,
-      "loss": 0.8614,
-      "step": 1046
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1182317711416441e-06,
-      "loss": 0.8438,
-      "step": 1047
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1100422026010626e-06,
-      "loss": 0.8277,
-      "step": 1048
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.1018789870475942e-06,
-      "loss": 0.8418,
-      "step": 1049
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.0937421797839254e-06,
-      "loss": 0.8889,
-      "step": 1050
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.0856318359338348e-06,
-      "loss": 0.8782,
-      "step": 1051
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 1.077548010441825e-06,
-      "loss": 0.8975,
-      "step": 1052
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0694907580727409e-06,
-      "loss": 0.8657,
-      "step": 1053
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0614601334114099e-06,
-      "loss": 0.845,
-      "step": 1054
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0534561908622637e-06,
-      "loss": 0.9132,
-      "step": 1055
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.045478984648976e-06,
-      "loss": 0.8208,
-      "step": 1056
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0375285688140902e-06,
-      "loss": 0.8893,
-      "step": 1057
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0375285688140902e-06,
-      "loss": 0.8591,
-      "step": 1058
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0296049972186534e-06,
-      "loss": 0.8423,
-      "step": 1059
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0217083235418551e-06,
-      "loss": 0.8524,
-      "step": 1060
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0138386012806617e-06,
-      "loss": 0.8506,
-      "step": 1061
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 1.0059958837494532e-06,
-      "loss": 0.9042,
-      "step": 1062
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 9.981802240796629e-07,
-      "loss": 0.8546,
-      "step": 1063
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 9.903916752194182e-07,
-      "loss": 0.891,
-      "step": 1064
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 9.826302899331792e-07,
-      "loss": 0.8472,
-      "step": 1065
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.748961208013858e-07,
-      "loss": 0.8679,
-      "step": 1066
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.671892202200961e-07,
-      "loss": 0.8856,
-      "step": 1067
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.59509640400636e-07,
-      "loss": 0.8549,
-      "step": 1068
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.518574333692432e-07,
-      "loss": 0.8734,
-      "step": 1069
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.442326509667138e-07,
-      "loss": 0.8767,
-      "step": 1070
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.366353448480547e-07,
-      "loss": 0.8649,
-      "step": 1071
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.290655664821296e-07,
-      "loss": 0.8458,
-      "step": 1072
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.21523367151313e-07,
-      "loss": 0.8605,
-      "step": 1073
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.140087979511414e-07,
-      "loss": 0.8891,
-      "step": 1074
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 9.065219097899685e-07,
-      "loss": 0.8641,
-      "step": 1075
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 8.990627533886187e-07,
-      "loss": 0.8951,
-      "step": 1076
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 8.916313792800441e-07,
-      "loss": 0.8558,
-      "step": 1077
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.842278378089836e-07,
-      "loss": 0.876,
-      "step": 1078
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.768521791316181e-07,
-      "loss": 0.8875,
-      "step": 1079
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.695044532152363e-07,
-      "loss": 0.8408,
-      "step": 1080
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.62184709837891e-07,
-      "loss": 0.8627,
-      "step": 1081
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.62184709837891e-07,
-      "loss": 0.8775,
-      "step": 1082
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.548929985880638e-07,
-      "loss": 0.8831,
-      "step": 1083
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.476293688643306e-07,
-      "loss": 0.8462,
-      "step": 1084
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.403938698750241e-07,
-      "loss": 0.8349,
-      "step": 1085
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.331865506379022e-07,
-      "loss": 0.8713,
-      "step": 1086
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.260074599798179e-07,
-      "loss": 0.8599,
-      "step": 1087
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.188566465363845e-07,
-      "loss": 0.8582,
-      "step": 1088
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 8.11734158751647e-07,
-      "loss": 0.8543,
-      "step": 1089
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 8.046400448777575e-07,
-      "loss": 0.8718,
-      "step": 1090
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.975743529746416e-07,
-      "loss": 0.8509,
-      "step": 1091
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.905371309096798e-07,
-      "loss": 0.8862,
-      "step": 1092
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.835284263573805e-07,
-      "loss": 0.8679,
-      "step": 1093
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.765482867990553e-07,
-      "loss": 0.8851,
-      "step": 1094
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.695967595224996e-07,
-      "loss": 0.8783,
-      "step": 1095
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.62673891621672e-07,
-      "loss": 0.8524,
-      "step": 1096
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.62673891621672e-07,
-      "loss": 0.8831,
-      "step": 1097
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.557797299963726e-07,
-      "loss": 0.8261,
-      "step": 1098
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.489143213519301e-07,
-      "loss": 0.8528,
-      "step": 1099
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.420777121988809e-07,
-      "loss": 0.7968,
-      "step": 1100
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.35269948852656e-07,
-      "loss": 0.8254,
-      "step": 1101
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 7.28491077433266e-07,
-      "loss": 0.8348,
-      "step": 1102
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.21741143864994e-07,
-      "loss": 0.8515,
-      "step": 1103
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.150201938760748e-07,
-      "loss": 0.8273,
-      "step": 1104
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.150201938760748e-07,
-      "loss": 0.8418,
-      "step": 1105
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.083282729983937e-07,
-      "loss": 0.8537,
-      "step": 1106
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.083282729983937e-07,
-      "loss": 0.8216,
-      "step": 1107
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 7.016654265671735e-07,
-      "loss": 0.8638,
-      "step": 1108
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.950316997206691e-07,
-      "loss": 0.8501,
-      "step": 1109
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.884271373998608e-07,
-      "loss": 0.8631,
-      "step": 1110
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.818517843481509e-07,
-      "loss": 0.8558,
-      "step": 1111
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.753056851110601e-07,
-      "loss": 0.8818,
-      "step": 1112
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.687888840359241e-07,
-      "loss": 0.8749,
-      "step": 1113
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 6.623014252715981e-07,
-      "loss": 0.8567,
-      "step": 1114
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.558433527681496e-07,
-      "loss": 0.8551,
-      "step": 1115
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.494147102765685e-07,
-      "loss": 0.8853,
-      "step": 1116
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.430155413484673e-07,
-      "loss": 0.8435,
-      "step": 1117
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.366458893357841e-07,
-      "loss": 0.8415,
-      "step": 1118
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.303057973904952e-07,
-      "loss": 0.8375,
-      "step": 1119
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.239953084643157e-07,
-      "loss": 0.8163,
-      "step": 1120
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.177144653084116e-07,
-      "loss": 0.888,
-      "step": 1121
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.114633104731105e-07,
-      "loss": 0.8443,
-      "step": 1122
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 6.052418863076132e-07,
-      "loss": 0.8416,
-      "step": 1123
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.99050234959706e-07,
-      "loss": 0.8369,
-      "step": 1124
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.928883983754763e-07,
-      "loss": 0.8746,
-      "step": 1125
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.867564182990271e-07,
-      "loss": 0.8512,
-      "step": 1126
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.806543362721945e-07,
-      "loss": 0.8491,
-      "step": 1127
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.745821936342688e-07,
-      "loss": 0.8461,
-      "step": 1128
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.685400315217093e-07,
-      "loss": 0.8829,
-      "step": 1129
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.625278908678716e-07,
-      "loss": 0.8719,
-      "step": 1130
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.625278908678716e-07,
-      "loss": 0.8508,
-      "step": 1131
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.565458124027251e-07,
-      "loss": 0.8884,
-      "step": 1132
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.505938366525804e-07,
-      "loss": 0.8516,
-      "step": 1133
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.446720039398129e-07,
-      "loss": 0.8938,
-      "step": 1134
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.387803543825909e-07,
-      "loss": 0.88,
-      "step": 1135
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.387803543825909e-07,
-      "loss": 0.8951,
-      "step": 1136
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.329189278946034e-07,
-      "loss": 0.8445,
-      "step": 1137
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.2708776418479e-07,
-      "loss": 0.8262,
-      "step": 1138
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.212869027570694e-07,
-      "loss": 0.8604,
-      "step": 1139
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.155163829100762e-07,
-      "loss": 0.8709,
-      "step": 1140
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.097762437368908e-07,
-      "loss": 0.8822,
-      "step": 1141
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.040665241247778e-07,
-      "loss": 0.8564,
-      "step": 1142
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.98387262754918e-07,
-      "loss": 0.8851,
-      "step": 1143
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.927384981021521e-07,
-      "loss": 0.8727,
-      "step": 1144
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.871202684347154e-07,
-      "loss": 0.8862,
-      "step": 1145
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.815326118139813e-07,
-      "loss": 0.8619,
-      "step": 1146
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.759755660942006e-07,
-      "loss": 0.8657,
-      "step": 1147
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.7044916892224956e-07,
-      "loss": 0.8848,
-      "step": 1148
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.649534577373688e-07,
-      "loss": 0.8369,
-      "step": 1149
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.5948846977091677e-07,
-      "loss": 0.8665,
-      "step": 1150
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.5405424204611136e-07,
-      "loss": 0.859,
-      "step": 1151
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.4865081137778264e-07,
-      "loss": 0.8767,
-      "step": 1152
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.43278214372122e-07,
-      "loss": 0.869,
-      "step": 1153
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.3793648742643247e-07,
-      "loss": 0.8268,
-      "step": 1154
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.3262566672888697e-07,
-      "loss": 0.8436,
-      "step": 1155
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.2734578825827934e-07,
-      "loss": 0.8711,
-      "step": 1156
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.2209688778378165e-07,
-      "loss": 0.811,
-      "step": 1157
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.1687900086470114e-07,
-      "loss": 0.8627,
-      "step": 1158
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.116921628502407e-07,
-      "loss": 0.8758,
-      "step": 1159
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.0653640887925893e-07,
-      "loss": 0.8675,
-      "step": 1160
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.0141177388003104e-07,
-      "loss": 0.8699,
-      "step": 1161
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.963182925700138e-07,
-      "loss": 0.8838,
-      "step": 1162
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.9125599945560866e-07,
-      "loss": 0.8263,
-      "step": 1163
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 3.862249288319286e-07,
-      "loss": 0.8649,
-      "step": 1164
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.8122511478256907e-07,
-      "loss": 0.8702,
-      "step": 1165
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.762565911793703e-07,
-      "loss": 0.8739,
-      "step": 1166
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.713193916821939e-07,
-      "loss": 0.8358,
-      "step": 1167
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.664135497386917e-07,
-      "loss": 0.8188,
-      "step": 1168
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.6153909858408e-07,
-      "loss": 0.8816,
-      "step": 1169
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.5669607124091454e-07,
-      "loss": 0.8666,
-      "step": 1170
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.5188450051886646e-07,
-      "loss": 0.894,
-      "step": 1171
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.471044190145012e-07,
-      "loss": 0.8772,
-      "step": 1172
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.4235585911105705e-07,
-      "loss": 0.8281,
-      "step": 1173
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.3763885297822153e-07,
-      "loss": 0.8412,
-      "step": 1174
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.3295343257192245e-07,
-      "loss": 0.8934,
-      "step": 1175
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.2829962963410287e-07,
-      "loss": 0.8728,
-      "step": 1176
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 3.2829962963410287e-07,
-      "loss": 0.8244,
-      "step": 1177
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.2367747569251153e-07,
-      "loss": 0.8582,
-      "step": 1178
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.1908700206048526e-07,
-      "loss": 0.836,
-      "step": 1179
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.145282398367416e-07,
-      "loss": 0.882,
-      "step": 1180
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.100012199051627e-07,
-      "loss": 0.8619,
-      "step": 1181
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.0550597293458994e-07,
-      "loss": 0.8377,
-      "step": 1182
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.010425293786151e-07,
-      "loss": 0.8927,
-      "step": 1183
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.966109194753736e-07,
-      "loss": 0.8737,
-      "step": 1184
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.922111732473393e-07,
-      "loss": 0.8266,
-      "step": 1185
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.878433205011222e-07,
-      "loss": 0.8565,
-      "step": 1186
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.835073908272673e-07,
-      "loss": 0.858,
-      "step": 1187
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.79203413600051e-07,
-      "loss": 0.8644,
-      "step": 1188
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 2.7493141797728495e-07,
-      "loss": 0.8798,
-      "step": 1189
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.7069143290011815e-07,
-      "loss": 0.862,
-      "step": 1190
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.6648348709283845e-07,
-      "loss": 0.8608,
-      "step": 1191
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.6230760906268195e-07,
-      "loss": 0.903,
-      "step": 1192
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.5816382709963674e-07,
-      "loss": 0.869,
-      "step": 1193
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.5405216927625155e-07,
-      "loss": 0.8849,
-      "step": 1194
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.499726634474475e-07,
-      "loss": 0.7991,
-      "step": 1195
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.4592533725032783e-07,
-      "loss": 0.8434,
-      "step": 1196
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.4191021810399063e-07,
-      "loss": 0.8619,
-      "step": 1197
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.3792733320934348e-07,
-      "loss": 0.8582,
-      "step": 1198
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.3397670954891926e-07,
-      "loss": 0.9126,
-      "step": 1199
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.300583738866935e-07,
-      "loss": 0.8429,
-      "step": 1200
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.2617235276790273e-07,
-      "loss": 0.8531,
-      "step": 1201
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.223186725188642e-07,
-      "loss": 0.8404,
-      "step": 1202
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.1849735924679873e-07,
-      "loss": 0.8919,
-      "step": 1203
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.1470843883965197e-07,
-      "loss": 0.8286,
-      "step": 1204
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.1095193696592241e-07,
-      "loss": 0.8302,
-      "step": 1205
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.0722787907448304e-07,
-      "loss": 0.852,
-      "step": 1206
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 2.0353629039441213e-07,
-      "loss": 0.9069,
-      "step": 1207
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.9987719593482114e-07,
-      "loss": 0.8603,
-      "step": 1208
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.9625062048468702e-07,
-      "loss": 0.8265,
-      "step": 1209
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.9265658861268023e-07,
-      "loss": 0.8709,
-      "step": 1210
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.890951246670031e-07,
-      "loss": 0.8495,
-      "step": 1211
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.8556625277522056e-07,
-      "loss": 0.8287,
-      "step": 1212
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.820699968441003e-07,
-      "loss": 0.83,
-      "step": 1213
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.7860638055944846e-07,
-      "loss": 0.8694,
-      "step": 1214
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.7517542738595071e-07,
-      "loss": 0.8369,
-      "step": 1215
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.7177716056701155e-07,
-      "loss": 0.8512,
-      "step": 1216
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.6841160312459913e-07,
-      "loss": 0.8394,
-      "step": 1217
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.6507877785908732e-07,
-      "loss": 0.8656,
-      "step": 1218
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.6177870734910117e-07,
-      "loss": 0.8585,
-      "step": 1219
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.585114139513655e-07,
-      "loss": 0.8474,
-      "step": 1220
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.552769198005527e-07,
-      "loss": 0.8785,
-      "step": 1221
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.52075246809133e-07,
-      "loss": 0.8431,
-      "step": 1222
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.4890641666722384e-07,
-      "loss": 0.837,
-      "step": 1223
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.4577045084244788e-07,
-      "loss": 0.8698,
-      "step": 1224
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.4266737057978263e-07,
-      "loss": 0.8663,
-      "step": 1225
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 1.3959719690141815e-07,
-      "loss": 0.8485,
-      "step": 1226
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.3655995060661565e-07,
-      "loss": 0.8175,
-      "step": 1227
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.3355565227156587e-07,
-      "loss": 0.8548,
-      "step": 1228
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.3058432224924866e-07,
-      "loss": 0.8379,
-      "step": 1229
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.2764598066929702e-07,
-      "loss": 0.8583,
-      "step": 1230
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.247406474378593e-07,
-      "loss": 0.8828,
-      "step": 1231
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.2186834223746612e-07,
-      "loss": 0.8581,
-      "step": 1232
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1902908452689376e-07,
-      "loss": 0.8477,
-      "step": 1233
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1622289354103478e-07,
-      "loss": 0.8449,
-      "step": 1234
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1344978829076813e-07,
-      "loss": 0.8532,
-      "step": 1235
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.1070978756282936e-07,
-      "loss": 0.8202,
-      "step": 1236
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.0800290991968221e-07,
-      "loss": 0.8603,
-      "step": 1237
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.05329173699395e-07,
-      "loss": 0.8479,
-      "step": 1238
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.0268859701551614e-07,
-      "loss": 0.8455,
-      "step": 1239
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 1.0008119775694992e-07,
-      "loss": 0.8531,
-      "step": 1240
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.750699358783589e-08,
-      "loss": 0.8533,
-      "step": 1241
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.496600194743078e-08,
-      "loss": 0.8295,
-      "step": 1242
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 9.245824004998682e-08,
-      "loss": 0.8504,
-      "step": 1243
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 8.998372488464024e-08,
-      "loss": 0.86,
-      "step": 1244
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 8.754247321529019e-08,
-      "loss": 0.8675,
-      "step": 1245
-    },
-    {
-      "epoch": 1.0,
-      "step": 1245,
-      "total_flos": 1.0675083773315383e+20,
-      "train_loss": 0.9266132452880522,
-      "train_runtime": 24349.4137,
-      "train_samples_per_second": 39.288,
-      "train_steps_per_second": 0.051
-    }
-  ],
-  "logging_steps": 1.0,
-  "max_steps": 1245,
-  "num_train_epochs": 1,
-  "save_steps": 250,
-  "total_flos": 1.0675083773315383e+20,
-  "trial_name": null,
-  "trial_params": null
-}