diff --git "a/checkpoint-11196/trainer_state.json" "b/checkpoint-11196/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11196/trainer_state.json"
@@ -0,0 +1,67210 @@
+{
+  "best_metric": 0.15645097196102142,
+  "best_model_checkpoint": "./medium-article-titles-engagement/checkpoint-11196",
+  "epoch": 2.0,
+  "global_step": 11196,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6922,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6961,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.6841,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.663,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.71,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.7028,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6921,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.698,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.67,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7024,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7012,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6888,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.6896,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6943,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6861,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6908,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.699,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6951,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6903,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.6857,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6834,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6985,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.7025,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6788,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6884,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6994,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6891,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.687,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6885,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.6974,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6887,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6911,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 0.6865,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6845,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6826,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 0.682,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6791,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6908,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.6739,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.6821,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.683,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6774,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6731,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6889,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6401,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.6513,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 0.656,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6556,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6523,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.6719,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999821045096636e-05,
+      "loss": 0.7211,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999642090193272e-05,
+      "loss": 0.7396,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994631352899075e-05,
+      "loss": 0.6635,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999284180386542e-05,
+      "loss": 0.6957,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991052254831785e-05,
+      "loss": 0.6602,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998926270579814e-05,
+      "loss": 0.6495,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987473156764496e-05,
+      "loss": 0.7193,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998568360773086e-05,
+      "loss": 0.6937,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998389405869721e-05,
+      "loss": 0.6713,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998210450966357e-05,
+      "loss": 0.6603,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998031496062992e-05,
+      "loss": 0.6642,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997852541159628e-05,
+      "loss": 0.6939,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976735862562633e-05,
+      "loss": 0.6661,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997494631352899e-05,
+      "loss": 0.6748,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997315676449535e-05,
+      "loss": 0.6629,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9971367215461706e-05,
+      "loss": 0.662,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996957766642806e-05,
+      "loss": 0.7149,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996778811739442e-05,
+      "loss": 0.6745,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996599856836078e-05,
+      "loss": 0.6708,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9964209019327127e-05,
+      "loss": 0.6832,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996241947029349e-05,
+      "loss": 0.6034,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9960629921259844e-05,
+      "loss": 0.6366,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99588403722262e-05,
+      "loss": 0.629,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995705082319256e-05,
+      "loss": 0.6794,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9955261274158916e-05,
+      "loss": 0.6434,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995347172512527e-05,
+      "loss": 0.6786,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9951682176091626e-05,
+      "loss": 0.6946,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994989262705798e-05,
+      "loss": 0.6071,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994810307802434e-05,
+      "loss": 0.6894,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99463135289907e-05,
+      "loss": 0.6707,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9944523979957054e-05,
+      "loss": 0.668,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994273443092341e-05,
+      "loss": 0.5982,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9940944881889764e-05,
+      "loss": 0.6333,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9939155332856126e-05,
+      "loss": 0.7821,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993736578382248e-05,
+      "loss": 0.6595,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993557623478884e-05,
+      "loss": 0.6241,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993378668575519e-05,
+      "loss": 0.6841,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993199713672155e-05,
+      "loss": 0.6969,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99302075876879e-05,
+      "loss": 0.6416,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9928418038654264e-05,
+      "loss": 0.713,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992662848962062e-05,
+      "loss": 0.7158,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9924838940586975e-05,
+      "loss": 0.6871,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992304939155334e-05,
+      "loss": 0.6646,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9921259842519685e-05,
+      "loss": 0.6689,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991947029348604e-05,
+      "loss": 0.6946,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99176807444524e-05,
+      "loss": 0.6207,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991589119541876e-05,
+      "loss": 0.6898,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991410164638511e-05,
+      "loss": 0.6499,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9912312097351475e-05,
+      "loss": 0.6236,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991052254831783e-05,
+      "loss": 0.677,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990873299928418e-05,
+      "loss": 0.6627,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990694345025054e-05,
+      "loss": 0.6116,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9905153901216895e-05,
+      "loss": 0.6592,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990336435218325e-05,
+      "loss": 0.6609,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9901574803149606e-05,
+      "loss": 0.6291,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989978525411597e-05,
+      "loss": 0.578,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989799570508232e-05,
+      "loss": 0.7022,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989620615604868e-05,
+      "loss": 0.6714,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989441660701503e-05,
+      "loss": 0.632,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989262705798139e-05,
+      "loss": 0.6574,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9890837508947744e-05,
+      "loss": 0.612,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9889047959914106e-05,
+      "loss": 0.5727,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988725841088046e-05,
+      "loss": 0.6735,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9885468861846816e-05,
+      "loss": 0.6861,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988367931281318e-05,
+      "loss": 0.5993,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988188976377953e-05,
+      "loss": 0.6089,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988010021474588e-05,
+      "loss": 0.6623,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9878310665712244e-05,
+      "loss": 0.6491,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98765211166786e-05,
+      "loss": 0.6589,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9874731567644954e-05,
+      "loss": 0.6489,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9872942018611316e-05,
+      "loss": 0.6839,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.987115246957767e-05,
+      "loss": 0.621,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9869362920544026e-05,
+      "loss": 0.6409,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986757337151038e-05,
+      "loss": 0.6472,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986578382247674e-05,
+      "loss": 0.6968,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986399427344309e-05,
+      "loss": 0.6309,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986220472440945e-05,
+      "loss": 0.6565,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986041517537581e-05,
+      "loss": 0.6765,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9858625626342164e-05,
+      "loss": 0.6211,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985683607730852e-05,
+      "loss": 0.6176,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985504652827488e-05,
+      "loss": 0.5987,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985325697924124e-05,
+      "loss": 0.622,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9851467430207585e-05,
+      "loss": 0.6797,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984967788117395e-05,
+      "loss": 0.7001,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98478883321403e-05,
+      "loss": 0.5792,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984609878310666e-05,
+      "loss": 0.6166,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984430923407302e-05,
+      "loss": 0.6422,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9842519685039375e-05,
+      "loss": 0.6796,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984073013600573e-05,
+      "loss": 0.7139,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9838940586972085e-05,
+      "loss": 0.6105,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983715103793844e-05,
+      "loss": 0.6123,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9835361488904795e-05,
+      "loss": 0.5742,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983357193987116e-05,
+      "loss": 0.5568,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983178239083751e-05,
+      "loss": 0.684,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982999284180387e-05,
+      "loss": 0.6406,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982820329277022e-05,
+      "loss": 0.6723,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9826413743736585e-05,
+      "loss": 0.6316,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982462419470293e-05,
+      "loss": 0.6224,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982283464566929e-05,
+      "loss": 0.5555,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982104509663565e-05,
+      "loss": 0.6196,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9819255547602006e-05,
+      "loss": 0.6634,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981746599856836e-05,
+      "loss": 0.6092,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981567644953472e-05,
+      "loss": 0.5854,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981388690050108e-05,
+      "loss": 0.7289,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981209735146743e-05,
+      "loss": 0.6223,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.981030780243379e-05,
+      "loss": 0.6034,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9808518253400144e-05,
+      "loss": 0.6143,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98067287043665e-05,
+      "loss": 0.7711,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980493915533286e-05,
+      "loss": 0.6737,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9803149606299216e-05,
+      "loss": 0.6942,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980136005726557e-05,
+      "loss": 0.5931,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979957050823193e-05,
+      "loss": 0.5988,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979778095919829e-05,
+      "loss": 0.6568,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979599141016464e-05,
+      "loss": 0.5981,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9794201861131e-05,
+      "loss": 0.6001,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9792412312097354e-05,
+      "loss": 0.6551,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.979062276306371e-05,
+      "loss": 0.6458,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9788833214030064e-05,
+      "loss": 0.5997,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9787043664996426e-05,
+      "loss": 0.6137,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.978525411596278e-05,
+      "loss": 0.6621,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9783464566929137e-05,
+      "loss": 0.5381,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.978167501789549e-05,
+      "loss": 0.6908,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.977988546886185e-05,
+      "loss": 0.7118,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97780959198282e-05,
+      "loss": 0.7153,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9776306370794564e-05,
+      "loss": 0.6565,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.977451682176092e-05,
+      "loss": 0.6231,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9772727272727275e-05,
+      "loss": 0.5716,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9770937723693636e-05,
+      "loss": 0.5812,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.976914817465999e-05,
+      "loss": 0.6775,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.976735862562634e-05,
+      "loss": 0.6186,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97655690765927e-05,
+      "loss": 0.6211,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.976377952755906e-05,
+      "loss": 0.7331,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.976198997852541e-05,
+      "loss": 0.6603,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9760200429491774e-05,
+      "loss": 0.636,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.975841088045813e-05,
+      "loss": 0.5944,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9756621331424485e-05,
+      "loss": 0.6677,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.975483178239084e-05,
+      "loss": 0.6126,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9753042233357195e-05,
+      "loss": 0.6302,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.975125268432355e-05,
+      "loss": 0.6869,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9749463135289906e-05,
+      "loss": 0.6355,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974767358625627e-05,
+      "loss": 0.6909,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974588403722262e-05,
+      "loss": 0.6618,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974409448818898e-05,
+      "loss": 0.5874,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974230493915534e-05,
+      "loss": 0.6568,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.974051539012169e-05,
+      "loss": 0.5909,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9738725841088044e-05,
+      "loss": 0.6612,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9736936292054405e-05,
+      "loss": 0.6388,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973514674302076e-05,
+      "loss": 0.6528,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9733357193987116e-05,
+      "loss": 0.6746,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973156764495348e-05,
+      "loss": 0.5955,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.972977809591983e-05,
+      "loss": 0.6039,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.972798854688619e-05,
+      "loss": 0.5732,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9726198997852543e-05,
+      "loss": 0.6103,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97244094488189e-05,
+      "loss": 0.6184,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9722619899785254e-05,
+      "loss": 0.6364,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9720830350751616e-05,
+      "loss": 0.6555,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971904080171797e-05,
+      "loss": 0.6634,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9717251252684326e-05,
+      "loss": 0.641,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971546170365068e-05,
+      "loss": 0.6183,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971367215461704e-05,
+      "loss": 0.6861,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971188260558339e-05,
+      "loss": 0.5727,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.971009305654975e-05,
+      "loss": 0.6123,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.970830350751611e-05,
+      "loss": 0.6367,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9706513958482464e-05,
+      "loss": 0.5823,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.970472440944882e-05,
+      "loss": 0.5544,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.970293486041518e-05,
+      "loss": 0.5996,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9701145311381536e-05,
+      "loss": 0.5626,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.969935576234789e-05,
+      "loss": 0.6386,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.969756621331425e-05,
+      "loss": 0.7131,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.96957766642806e-05,
+      "loss": 0.5484,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.969398711524696e-05,
+      "loss": 0.6357,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.969219756621332e-05,
+      "loss": 0.6093,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9690408017179674e-05,
+      "loss": 0.6162,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.968861846814603e-05,
+      "loss": 0.5411,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9686828919112385e-05,
+      "loss": 0.6623,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.968503937007874e-05,
+      "loss": 0.5941,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9683249821045095e-05,
+      "loss": 0.6165,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.968146027201146e-05,
+      "loss": 0.6237,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.967967072297781e-05,
+      "loss": 0.6753,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.967788117394417e-05,
+      "loss": 0.7163,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.967609162491052e-05,
+      "loss": 0.6859,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9674302075876885e-05,
+      "loss": 0.7098,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.967251252684324e-05,
+      "loss": 0.6287,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9670722977809595e-05,
+      "loss": 0.6669,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966893342877595e-05,
+      "loss": 0.6327,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9667143879742305e-05,
+      "loss": 0.6022,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966535433070866e-05,
+      "loss": 0.7056,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966356478167502e-05,
+      "loss": 0.5537,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966177523264138e-05,
+      "loss": 0.6103,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.965998568360773e-05,
+      "loss": 0.6153,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9658196134574095e-05,
+      "loss": 0.6563,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.965640658554044e-05,
+      "loss": 0.6178,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.96546170365068e-05,
+      "loss": 0.6738,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.965282748747316e-05,
+      "loss": 0.6011,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9651037938439516e-05,
+      "loss": 0.5758,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964924838940587e-05,
+      "loss": 0.5775,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964745884037223e-05,
+      "loss": 0.6693,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964566929133859e-05,
+      "loss": 0.6214,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.964387974230494e-05,
+      "loss": 0.6645,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.96420901932713e-05,
+      "loss": 0.6448,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9640300644237654e-05,
+      "loss": 0.5853,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963851109520401e-05,
+      "loss": 0.6787,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9636721546170364e-05,
+      "loss": 0.6055,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9634931997136726e-05,
+      "loss": 0.6307,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963314244810308e-05,
+      "loss": 0.5765,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9631352899069436e-05,
+      "loss": 0.6163,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96295633500358e-05,
+      "loss": 0.6067,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962777380100215e-05,
+      "loss": 0.5981,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96259842519685e-05,
+      "loss": 0.5738,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9624194702934864e-05,
+      "loss": 0.5922,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962240515390122e-05,
+      "loss": 0.6263,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9620615604867574e-05,
+      "loss": 0.7036,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9618826055833936e-05,
+      "loss": 0.5983,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.961703650680029e-05,
+      "loss": 0.6333,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.961524695776665e-05,
+      "loss": 0.5967,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9613457408733e-05,
+      "loss": 0.5688,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.961166785969936e-05,
+      "loss": 0.5431,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960987831066571e-05,
+      "loss": 0.601,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9608088761632074e-05,
+      "loss": 0.55,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960629921259843e-05,
+      "loss": 0.612,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9604509663564785e-05,
+      "loss": 0.6823,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960272011453114e-05,
+      "loss": 0.5933,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9600930565497495e-05,
+      "loss": 0.609,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959914101646385e-05,
+      "loss": 0.5306,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9597351467430205e-05,
+      "loss": 0.5925,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959556191839657e-05,
+      "loss": 0.6903,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959377236936292e-05,
+      "loss": 0.6006,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959198282032928e-05,
+      "loss": 0.6002,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.959019327129564e-05,
+      "loss": 0.6269,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9588403722261995e-05,
+      "loss": 0.5891,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.958661417322834e-05,
+      "loss": 0.5483,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9584824624194705e-05,
+      "loss": 0.6536,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.958303507516106e-05,
+      "loss": 0.5589,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9581245526127416e-05,
+      "loss": 0.5735,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957945597709378e-05,
+      "loss": 0.5295,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957766642806013e-05,
+      "loss": 0.5947,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957587687902649e-05,
+      "loss": 0.6426,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957408732999284e-05,
+      "loss": 0.6972,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.95722977809592e-05,
+      "loss": 0.6333,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9570508231925554e-05,
+      "loss": 0.5348,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9568718682891916e-05,
+      "loss": 0.6316,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.956692913385827e-05,
+      "loss": 0.5965,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9565139584824626e-05,
+      "loss": 0.5576,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.956335003579098e-05,
+      "loss": 0.657,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.956156048675734e-05,
+      "loss": 0.5776,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.95597709377237e-05,
+      "loss": 0.6204,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.955798138869005e-05,
+      "loss": 0.6941,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.955619183965641e-05,
+      "loss": 0.6344,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9554402290622764e-05,
+      "loss": 0.5399,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.955261274158912e-05,
+      "loss": 0.6891,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.955082319255548e-05,
+      "loss": 0.628,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9549033643521836e-05,
+      "loss": 0.5711,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.954724409448819e-05,
+      "loss": 0.6191,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9545454545454553e-05,
+      "loss": 0.6102,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.95436649964209e-05,
+      "loss": 0.5795,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.954187544738726e-05,
+      "loss": 0.5303,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.954008589835362e-05,
+      "loss": 0.6408,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9538296349319974e-05,
+      "loss": 0.5289,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953650680028633e-05,
+      "loss": 0.6052,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953471725125269e-05,
+      "loss": 0.6212,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9532927702219047e-05,
+      "loss": 0.6292,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95311381531854e-05,
+      "loss": 0.7407,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952934860415176e-05,
+      "loss": 0.5783,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952755905511811e-05,
+      "loss": 0.5932,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952576950608447e-05,
+      "loss": 0.6269,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952397995705082e-05,
+      "loss": 0.5858,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9522190408017184e-05,
+      "loss": 0.7334,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952040085898354e-05,
+      "loss": 0.5732,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9518611309949895e-05,
+      "loss": 0.6077,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951682176091625e-05,
+      "loss": 0.632,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9515032211882605e-05,
+      "loss": 0.5693,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951324266284896e-05,
+      "loss": 0.6146,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951145311381532e-05,
+      "loss": 0.5379,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950966356478168e-05,
+      "loss": 0.5827,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950787401574803e-05,
+      "loss": 0.6506,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9506084466714395e-05,
+      "loss": 0.646,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950429491768075e-05,
+      "loss": 0.5659,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95025053686471e-05,
+      "loss": 0.5426,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950071581961346e-05,
+      "loss": 0.5884,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9498926270579816e-05,
+      "loss": 0.6459,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949713672154617e-05,
+      "loss": 0.6888,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949534717251253e-05,
+      "loss": 0.6448,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949355762347889e-05,
+      "loss": 0.6125,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949176807444524e-05,
+      "loss": 0.6449,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94899785254116e-05,
+      "loss": 0.542,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9488188976377953e-05,
+      "loss": 0.6363,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948639942734431e-05,
+      "loss": 0.7006,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9484609878310664e-05,
+      "loss": 0.6418,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9482820329277026e-05,
+      "loss": 0.579,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948103078024338e-05,
+      "loss": 0.5805,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9479241231209736e-05,
+      "loss": 0.6813,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94774516821761e-05,
+      "loss": 0.5449,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9475662133142453e-05,
+      "loss": 0.6177,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94738725841088e-05,
+      "loss": 0.6752,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9472083035075164e-05,
+      "loss": 0.5751,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.947029348604152e-05,
+      "loss": 0.571,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9468503937007874e-05,
+      "loss": 0.5864,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9466714387974236e-05,
+      "loss": 0.6054,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.946492483894059e-05,
+      "loss": 0.6791,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9463135289906947e-05,
+      "loss": 0.5739,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94613457408733e-05,
+      "loss": 0.5519,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.945955619183966e-05,
+      "loss": 0.6398,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.945776664280601e-05,
+      "loss": 0.5724,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9455977093772374e-05,
+      "loss": 0.526,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.945418754473873e-05,
+      "loss": 0.5944,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9452397995705084e-05,
+      "loss": 0.5974,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.945060844667144e-05,
+      "loss": 0.6143,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.94488188976378e-05,
+      "loss": 0.7447,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.944702934860415e-05,
+      "loss": 0.8067,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9445239799570505e-05,
+      "loss": 0.5412,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.944345025053687e-05,
+      "loss": 0.6556,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.944166070150322e-05,
+      "loss": 0.6344,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.943987115246958e-05,
+      "loss": 0.6753,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943808160343594e-05,
+      "loss": 0.6223,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9436292054402295e-05,
+      "loss": 0.5596,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943450250536865e-05,
+      "loss": 0.6123,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9432712956335005e-05,
+      "loss": 0.6021,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943092340730136e-05,
+      "loss": 0.7021,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9429133858267716e-05,
+      "loss": 0.5946,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942734430923408e-05,
+      "loss": 0.634,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942555476020043e-05,
+      "loss": 0.6051,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942376521116679e-05,
+      "loss": 0.6454,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942197566213314e-05,
+      "loss": 0.5179,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9420186113099505e-05,
+      "loss": 0.5423,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9418396564065853e-05,
+      "loss": 0.5363,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9416607015032215e-05,
+      "loss": 0.6788,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941481746599857e-05,
+      "loss": 0.5939,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9413027916964926e-05,
+      "loss": 0.5911,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941123836793128e-05,
+      "loss": 0.6029,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940944881889764e-05,
+      "loss": 0.5354,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9407659269864e-05,
+      "loss": 0.5766,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940586972083035e-05,
+      "loss": 0.6177,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940408017179671e-05,
+      "loss": 0.5005,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9402290622763064e-05,
+      "loss": 0.7079,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940050107372942e-05,
+      "loss": 0.5236,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939871152469578e-05,
+      "loss": 0.657,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9396921975662136e-05,
+      "loss": 0.5595,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939513242662849e-05,
+      "loss": 0.532,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939334287759485e-05,
+      "loss": 0.6491,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939155332856121e-05,
+      "loss": 0.5243,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938976377952756e-05,
+      "loss": 0.6047,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938797423049392e-05,
+      "loss": 0.6228,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9386184681460274e-05,
+      "loss": 0.5928,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938439513242663e-05,
+      "loss": 0.6478,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938260558339299e-05,
+      "loss": 0.6554,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9380816034359346e-05,
+      "loss": 0.6298,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93790264853257e-05,
+      "loss": 0.6534,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937723693629206e-05,
+      "loss": 0.541,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937544738725841e-05,
+      "loss": 0.5084,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937365783822477e-05,
+      "loss": 0.5843,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937186828919112e-05,
+      "loss": 0.6001,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9370078740157484e-05,
+      "loss": 0.5555,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936828919112384e-05,
+      "loss": 0.5483,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9366499642090195e-05,
+      "loss": 0.6434,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936471009305656e-05,
+      "loss": 0.5218,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9362920544022905e-05,
+      "loss": 0.5861,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936113099498926e-05,
+      "loss": 0.5985,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935934144595562e-05,
+      "loss": 0.6474,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935755189692198e-05,
+      "loss": 0.565,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935576234788833e-05,
+      "loss": 0.6293,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9353972798854695e-05,
+      "loss": 0.5866,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935218324982105e-05,
+      "loss": 0.6511,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9350393700787405e-05,
+      "loss": 0.5878,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934860415175376e-05,
+      "loss": 0.6128,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9346814602720115e-05,
+      "loss": 0.7021,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934502505368647e-05,
+      "loss": 0.6205,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934323550465283e-05,
+      "loss": 0.6247,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934144595561919e-05,
+      "loss": 0.6016,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933965640658554e-05,
+      "loss": 0.5835,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93378668575519e-05,
+      "loss": 0.5436,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933607730851826e-05,
+      "loss": 0.5846,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933428775948461e-05,
+      "loss": 0.6252,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9332498210450964e-05,
+      "loss": 0.504,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9330708661417326e-05,
+      "loss": 0.6424,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932891911238368e-05,
+      "loss": 0.6456,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9327129563350036e-05,
+      "loss": 0.6239,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93253400143164e-05,
+      "loss": 0.6334,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932355046528275e-05,
+      "loss": 0.586,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932176091624911e-05,
+      "loss": 0.5852,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9319971367215464e-05,
+      "loss": 0.6147,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931818181818182e-05,
+      "loss": 0.5687,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9316392269148174e-05,
+      "loss": 0.4865,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9314602720114536e-05,
+      "loss": 0.594,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.931281317108089e-05,
+      "loss": 0.4935,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9311023622047246e-05,
+      "loss": 0.534,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93092340730136e-05,
+      "loss": 0.6648,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9307444523979963e-05,
+      "loss": 0.6514,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930565497494631e-05,
+      "loss": 0.5339,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9303865425912674e-05,
+      "loss": 0.6418,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930207587687903e-05,
+      "loss": 0.6618,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9300286327845384e-05,
+      "loss": 0.681,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929849677881174e-05,
+      "loss": 0.5561,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92967072297781e-05,
+      "loss": 0.6028,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929491768074446e-05,
+      "loss": 0.5308,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929312813171081e-05,
+      "loss": 0.5666,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929133858267717e-05,
+      "loss": 0.685,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928954903364352e-05,
+      "loss": 0.6629,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928775948460988e-05,
+      "loss": 0.5485,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928596993557624e-05,
+      "loss": 0.6296,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9284180386542595e-05,
+      "loss": 0.5596,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928239083750895e-05,
+      "loss": 0.5768,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928060128847531e-05,
+      "loss": 0.6132,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927881173944166e-05,
+      "loss": 0.545,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9277022190408015e-05,
+      "loss": 0.6138,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927523264137438e-05,
+      "loss": 0.5813,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927344309234073e-05,
+      "loss": 0.6353,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927165354330709e-05,
+      "loss": 0.5538,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926986399427345e-05,
+      "loss": 0.6037,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9268074445239805e-05,
+      "loss": 0.6111,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926628489620616e-05,
+      "loss": 0.6589,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9264495347172515e-05,
+      "loss": 0.6601,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926270579813887e-05,
+      "loss": 0.6186,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9260916249105226e-05,
+      "loss": 0.6206,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925912670007158e-05,
+      "loss": 0.6312,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925733715103794e-05,
+      "loss": 0.6178,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92555476020043e-05,
+      "loss": 0.5719,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925375805297065e-05,
+      "loss": 0.6086,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9251968503937015e-05,
+      "loss": 0.5292,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9250178954903364e-05,
+      "loss": 0.6258,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924838940586972e-05,
+      "loss": 0.6866,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924659985683608e-05,
+      "loss": 0.5804,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9244810307802436e-05,
+      "loss": 0.5836,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924302075876879e-05,
+      "loss": 0.6304,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924123120973515e-05,
+      "loss": 0.5022,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923944166070151e-05,
+      "loss": 0.6673,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9237652111667863e-05,
+      "loss": 0.5277,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923586256263422e-05,
+      "loss": 0.5201,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9234073013600574e-05,
+      "loss": 0.6589,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923228346456693e-05,
+      "loss": 0.6271,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923049391553329e-05,
+      "loss": 0.6417,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9228704366499646e-05,
+      "loss": 0.5726,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9226914817466e-05,
+      "loss": 0.4826,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9225125268432357e-05,
+      "loss": 0.6361,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922333571939872e-05,
+      "loss": 0.5489,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.922154617036507e-05,
+      "loss": 0.5313,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921975662133142e-05,
+      "loss": 0.5059,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9217967072297784e-05,
+      "loss": 0.5893,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921617752326414e-05,
+      "loss": 0.6113,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9214387974230495e-05,
+      "loss": 0.641,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9212598425196856e-05,
+      "loss": 0.5681,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921080887616321e-05,
+      "loss": 0.5583,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920901932712956e-05,
+      "loss": 0.6656,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920722977809592e-05,
+      "loss": 0.6134,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920544022906228e-05,
+      "loss": 0.6841,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920365068002863e-05,
+      "loss": 0.5679,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9201861130994994e-05,
+      "loss": 0.671,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.920007158196135e-05,
+      "loss": 0.6152,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9198282032927705e-05,
+      "loss": 0.5737,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919649248389406e-05,
+      "loss": 0.5029,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9194702934860415e-05,
+      "loss": 0.5658,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919291338582677e-05,
+      "loss": 0.5915,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.919112383679313e-05,
+      "loss": 0.5958,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918933428775949e-05,
+      "loss": 0.5421,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918754473872584e-05,
+      "loss": 0.6464,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91857551896922e-05,
+      "loss": 0.5605,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.918396564065856e-05,
+      "loss": 0.5583,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9182176091624915e-05,
+      "loss": 0.6356,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9180386542591264e-05,
+      "loss": 0.5478,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9178596993557625e-05,
+      "loss": 0.6191,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917680744452398e-05,
+      "loss": 0.6287,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9175017895490336e-05,
+      "loss": 0.578,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91732283464567e-05,
+      "loss": 0.629,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.917143879742305e-05,
+      "loss": 0.5924,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916964924838941e-05,
+      "loss": 0.5319,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916785969935577e-05,
+      "loss": 0.6695,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916607015032212e-05,
+      "loss": 0.5906,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9164280601288474e-05,
+      "loss": 0.6506,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9162491052254836e-05,
+      "loss": 0.5212,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.916070150322119e-05,
+      "loss": 0.557,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9158911954187546e-05,
+      "loss": 0.5464,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915712240515391e-05,
+      "loss": 0.5282,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915533285612026e-05,
+      "loss": 0.6234,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.915354330708662e-05,
+      "loss": 0.5891,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9151753758052974e-05,
+      "loss": 0.6322,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914996420901933e-05,
+      "loss": 0.6493,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9148174659985684e-05,
+      "loss": 0.6333,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914638511095204e-05,
+      "loss": 0.551,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.91445955619184e-05,
+      "loss": 0.6376,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9142806012884756e-05,
+      "loss": 0.5498,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.914101646385111e-05,
+      "loss": 0.5767,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9139226914817474e-05,
+      "loss": 0.6222,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913743736578382e-05,
+      "loss": 0.6581,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913564781675018e-05,
+      "loss": 0.5492,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913385826771654e-05,
+      "loss": 0.6442,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9132068718682894e-05,
+      "loss": 0.5701,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913027916964925e-05,
+      "loss": 0.5992,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912848962061561e-05,
+      "loss": 0.6453,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912670007158197e-05,
+      "loss": 0.7731,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9124910522548315e-05,
+      "loss": 0.5442,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912312097351468e-05,
+      "loss": 0.599,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912133142448103e-05,
+      "loss": 0.6772,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.911954187544739e-05,
+      "loss": 0.5505,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.911775232641375e-05,
+      "loss": 0.5959,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9115962777380105e-05,
+      "loss": 0.5924,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.911417322834646e-05,
+      "loss": 0.4768,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9112383679312815e-05,
+      "loss": 0.6329,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.911059413027917e-05,
+      "loss": 0.6241,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9108804581245525e-05,
+      "loss": 0.5536,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.910701503221188e-05,
+      "loss": 0.536,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.910522548317824e-05,
+      "loss": 0.5721,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91034359341446e-05,
+      "loss": 0.5778,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.910164638511095e-05,
+      "loss": 0.5992,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9099856836077315e-05,
+      "loss": 0.4669,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.909806728704367e-05,
+      "loss": 0.5655,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.909627773801002e-05,
+      "loss": 0.5867,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.909448818897638e-05,
+      "loss": 0.5643,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9092698639942736e-05,
+      "loss": 0.6496,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.909090909090909e-05,
+      "loss": 0.5151,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.908911954187545e-05,
+      "loss": 0.539,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.908732999284181e-05,
+      "loss": 0.514,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.908554044380816e-05,
+      "loss": 0.624,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.908375089477452e-05,
+      "loss": 0.6531,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9081961345740874e-05,
+      "loss": 0.5542,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.908017179670723e-05,
+      "loss": 0.6548,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.907838224767359e-05,
+      "loss": 0.6191,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9076592698639946e-05,
+      "loss": 0.5725,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.90748031496063e-05,
+      "loss": 0.586,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9073013600572656e-05,
+      "loss": 0.6154,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.907122405153902e-05,
+      "loss": 0.5637,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9069434502505374e-05,
+      "loss": 0.5586,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.906764495347172e-05,
+      "loss": 0.5147,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9065855404438084e-05,
+      "loss": 0.5123,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.906406585540444e-05,
+      "loss": 0.4905,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9062276306370794e-05,
+      "loss": 0.5882,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9060486757337156e-05,
+      "loss": 0.6029,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905869720830351e-05,
+      "loss": 0.6318,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905690765926987e-05,
+      "loss": 0.5669,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905511811023622e-05,
+      "loss": 0.6088,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905332856120258e-05,
+      "loss": 0.528,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.905153901216893e-05,
+      "loss": 0.6959,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9049749463135294e-05,
+      "loss": 0.6164,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.904795991410165e-05,
+      "loss": 0.5876,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9046170365068005e-05,
+      "loss": 0.6355,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.904438081603436e-05,
+      "loss": 0.5746,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.904259126700072e-05,
+      "loss": 0.5893,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.904080171796707e-05,
+      "loss": 0.5778,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.903901216893343e-05,
+      "loss": 0.5812,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.903722261989979e-05,
+      "loss": 0.5153,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.903543307086614e-05,
+      "loss": 0.5952,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90336435218325e-05,
+      "loss": 0.6024,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.903185397279886e-05,
+      "loss": 0.5792,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9030064423765215e-05,
+      "loss": 0.5461,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.902827487473157e-05,
+      "loss": 0.5347,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9026485325697925e-05,
+      "loss": 0.6552,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.902469577666428e-05,
+      "loss": 0.5945,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9022906227630636e-05,
+      "loss": 0.555,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9021116678597e-05,
+      "loss": 0.6546,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.901932712956335e-05,
+      "loss": 0.5793,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.901753758052971e-05,
+      "loss": 0.5986,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.901574803149607e-05,
+      "loss": 0.5861,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9013958482462425e-05,
+      "loss": 0.5467,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9012168933428774e-05,
+      "loss": 0.533,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9010379384395136e-05,
+      "loss": 0.6071,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.900858983536149e-05,
+      "loss": 0.5532,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9006800286327846e-05,
+      "loss": 0.5515,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.900501073729421e-05,
+      "loss": 0.595,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.900322118826056e-05,
+      "loss": 0.5893,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.900143163922692e-05,
+      "loss": 0.5668,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8999642090193274e-05,
+      "loss": 0.5903,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.899785254115963e-05,
+      "loss": 0.5652,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8996062992125984e-05,
+      "loss": 0.5023,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.899427344309234e-05,
+      "loss": 0.4702,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.89924838940587e-05,
+      "loss": 0.5913,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8990694345025056e-05,
+      "loss": 0.5799,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.898890479599141e-05,
+      "loss": 0.4942,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8987115246957773e-05,
+      "loss": 0.4773,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.898532569792413e-05,
+      "loss": 0.6069,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.898353614889048e-05,
+      "loss": 0.5894,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.898174659985684e-05,
+      "loss": 0.5172,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8979957050823194e-05,
+      "loss": 0.5236,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.897816750178955e-05,
+      "loss": 0.4774,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.897637795275591e-05,
+      "loss": 0.596,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8974588403722267e-05,
+      "loss": 0.4498,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.897279885468862e-05,
+      "loss": 0.5701,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.897100930565498e-05,
+      "loss": 0.6211,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.896921975662133e-05,
+      "loss": 0.6936,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.896743020758769e-05,
+      "loss": 0.5321,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.896564065855405e-05,
+      "loss": 0.5795,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8963851109520404e-05,
+      "loss": 0.5595,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.896206156048676e-05,
+      "loss": 0.5895,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8960272011453115e-05,
+      "loss": 0.5181,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.895848246241948e-05,
+      "loss": 0.6508,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8956692913385825e-05,
+      "loss": 0.5843,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.895490336435218e-05,
+      "loss": 0.5229,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.895311381531854e-05,
+      "loss": 0.5386,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.89513242662849e-05,
+      "loss": 0.6154,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.894953471725125e-05,
+      "loss": 0.5182,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8947745168217615e-05,
+      "loss": 0.5953,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.894595561918397e-05,
+      "loss": 0.5877,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8944166070150325e-05,
+      "loss": 0.4378,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.894237652111668e-05,
+      "loss": 0.5456,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8940586972083036e-05,
+      "loss": 0.5729,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.893879742304939e-05,
+      "loss": 0.5354,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893700787401575e-05,
+      "loss": 0.6926,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893521832498211e-05,
+      "loss": 0.5891,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893342877594846e-05,
+      "loss": 0.4307,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893163922691482e-05,
+      "loss": 0.4721,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892984967788118e-05,
+      "loss": 0.5102,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892806012884753e-05,
+      "loss": 0.5383,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892627057981389e-05,
+      "loss": 0.7038,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8924481030780246e-05,
+      "loss": 0.5713,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89226914817466e-05,
+      "loss": 0.5116,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8920901932712956e-05,
+      "loss": 0.49,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891911238367932e-05,
+      "loss": 0.5328,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891732283464567e-05,
+      "loss": 0.5372,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891553328561203e-05,
+      "loss": 0.6296,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8913743736578384e-05,
+      "loss": 0.6269,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891195418754474e-05,
+      "loss": 0.5644,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8910164638511094e-05,
+      "loss": 0.5744,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8908375089477456e-05,
+      "loss": 0.5549,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.890658554044381e-05,
+      "loss": 0.5951,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8904795991410167e-05,
+      "loss": 0.6255,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.890300644237653e-05,
+      "loss": 0.5681,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8901216893342884e-05,
+      "loss": 0.6189,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889942734430923e-05,
+      "loss": 0.5992,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8897637795275594e-05,
+      "loss": 0.5542,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889584824624195e-05,
+      "loss": 0.5889,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8894058697208304e-05,
+      "loss": 0.5643,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8892269148174666e-05,
+      "loss": 0.5484,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889047959914102e-05,
+      "loss": 0.5693,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888869005010738e-05,
+      "loss": 0.699,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888690050107373e-05,
+      "loss": 0.5909,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888511095204009e-05,
+      "loss": 0.6606,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.888332140300644e-05,
+      "loss": 0.5964,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88815318539728e-05,
+      "loss": 0.5497,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887974230493916e-05,
+      "loss": 0.6214,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8877952755905515e-05,
+      "loss": 0.5322,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887616320687187e-05,
+      "loss": 0.6234,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887437365783823e-05,
+      "loss": 0.5914,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.887258410880458e-05,
+      "loss": 0.516,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8870794559770936e-05,
+      "loss": 0.5714,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.88690050107373e-05,
+      "loss": 0.5049,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886721546170365e-05,
+      "loss": 0.6675,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886542591267001e-05,
+      "loss": 0.5372,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886363636363637e-05,
+      "loss": 0.5135,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8861846814602725e-05,
+      "loss": 0.609,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886005726556908e-05,
+      "loss": 0.5384,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8858267716535435e-05,
+      "loss": 0.5734,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885647816750179e-05,
+      "loss": 0.4847,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8854688618468146e-05,
+      "loss": 0.6034,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885289906943451e-05,
+      "loss": 0.5371,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.885110952040086e-05,
+      "loss": 0.4453,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884931997136722e-05,
+      "loss": 0.5867,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884753042233357e-05,
+      "loss": 0.556,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8845740873299935e-05,
+      "loss": 0.4945,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8843951324266284e-05,
+      "loss": 0.4746,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884216177523264e-05,
+      "loss": 0.4893,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8840372226199e-05,
+      "loss": 0.623,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8838582677165356e-05,
+      "loss": 0.6195,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883679312813171e-05,
+      "loss": 0.6635,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883500357909807e-05,
+      "loss": 0.5819,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883321403006443e-05,
+      "loss": 0.5322,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8831424481030784e-05,
+      "loss": 0.4707,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882963493199714e-05,
+      "loss": 0.5681,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8827845382963494e-05,
+      "loss": 0.5653,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882605583392985e-05,
+      "loss": 0.6338,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882426628489621e-05,
+      "loss": 0.5875,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8822476735862566e-05,
+      "loss": 0.5711,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882068718682892e-05,
+      "loss": 0.5843,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881889763779528e-05,
+      "loss": 0.6388,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881710808876163e-05,
+      "loss": 0.5779,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881531853972799e-05,
+      "loss": 0.5409,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881352899069435e-05,
+      "loss": 0.6128,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8811739441660704e-05,
+      "loss": 0.6705,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880994989262706e-05,
+      "loss": 0.5288,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8808160343593415e-05,
+      "loss": 0.5617,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880637079455978e-05,
+      "loss": 0.5819,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880458124552613e-05,
+      "loss": 0.5418,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880279169649248e-05,
+      "loss": 0.5196,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880100214745884e-05,
+      "loss": 0.5471,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.87992125984252e-05,
+      "loss": 0.5968,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879742304939155e-05,
+      "loss": 0.5142,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8795633500357915e-05,
+      "loss": 0.5457,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879384395132427e-05,
+      "loss": 0.595,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8792054402290625e-05,
+      "loss": 0.6114,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.879026485325699e-05,
+      "loss": 0.6143,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8788475304223335e-05,
+      "loss": 0.5825,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878668575518969e-05,
+      "loss": 0.5482,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878489620615605e-05,
+      "loss": 0.4642,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878310665712241e-05,
+      "loss": 0.5632,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.878131710808876e-05,
+      "loss": 0.6821,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8779527559055125e-05,
+      "loss": 0.5736,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877773801002148e-05,
+      "loss": 0.5698,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8775948460987835e-05,
+      "loss": 0.4762,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877415891195419e-05,
+      "loss": 0.536,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8772369362920546e-05,
+      "loss": 0.4884,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.87705798138869e-05,
+      "loss": 0.5674,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8768790264853256e-05,
+      "loss": 0.577,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876700071581962e-05,
+      "loss": 0.5552,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876521116678597e-05,
+      "loss": 0.5646,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876342161775233e-05,
+      "loss": 0.5605,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876163206871869e-05,
+      "loss": 0.5045,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875984251968504e-05,
+      "loss": 0.5866,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8758052970651394e-05,
+      "loss": 0.5872,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8756263421617756e-05,
+      "loss": 0.7006,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875447387258411e-05,
+      "loss": 0.6149,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8752684323550466e-05,
+      "loss": 0.5548,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875089477451683e-05,
+      "loss": 0.5503,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8749105225483183e-05,
+      "loss": 0.6051,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874731567644954e-05,
+      "loss": 0.5729,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8745526127415894e-05,
+      "loss": 0.5812,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874373657838225e-05,
+      "loss": 0.5761,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8741947029348604e-05,
+      "loss": 0.5346,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8740157480314966e-05,
+      "loss": 0.5299,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873836793128132e-05,
+      "loss": 0.5117,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.873657838224768e-05,
+      "loss": 0.6163,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.873478883321403e-05,
+      "loss": 0.5129,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.873299928418039e-05,
+      "loss": 0.5737,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.873120973514674e-05,
+      "loss": 0.6153,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.87294201861131e-05,
+      "loss": 0.521,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872763063707946e-05,
+      "loss": 0.638,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8725841088045815e-05,
+      "loss": 0.469,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872405153901217e-05,
+      "loss": 0.4991,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872226198997853e-05,
+      "loss": 0.5256,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872047244094489e-05,
+      "loss": 0.6055,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8718682891911235e-05,
+      "loss": 0.5101,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.87168933428776e-05,
+      "loss": 0.4652,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.871510379384395e-05,
+      "loss": 0.5492,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.871331424481031e-05,
+      "loss": 0.6136,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.871152469577667e-05,
+      "loss": 0.8429,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8709735146743025e-05,
+      "loss": 0.6496,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.870794559770938e-05,
+      "loss": 0.5157,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8706156048675735e-05,
+      "loss": 0.4854,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.870436649964209e-05,
+      "loss": 0.5483,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8702576950608446e-05,
+      "loss": 0.6708,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.870078740157481e-05,
+      "loss": 0.501,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.869899785254116e-05,
+      "loss": 0.6487,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.869720830350752e-05,
+      "loss": 0.4944,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.869541875447387e-05,
+      "loss": 0.4996,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8693629205440235e-05,
+      "loss": 0.5739,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.869183965640659e-05,
+      "loss": 0.5575,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.869005010737294e-05,
+      "loss": 0.6005,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.86882605583393e-05,
+      "loss": 0.5562,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8686471009305656e-05,
+      "loss": 0.6586,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.868468146027201e-05,
+      "loss": 0.5277,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.868289191123837e-05,
+      "loss": 0.6003,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.868110236220473e-05,
+      "loss": 0.5808,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8679312813171083e-05,
+      "loss": 0.5351,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8677523264137445e-05,
+      "loss": 0.5633,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8675733715103794e-05,
+      "loss": 0.7113,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.867394416607015e-05,
+      "loss": 0.5642,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.867215461703651e-05,
+      "loss": 0.6183,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8670365068002866e-05,
+      "loss": 0.4477,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.866857551896922e-05,
+      "loss": 0.6188,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8666785969935577e-05,
+      "loss": 0.5872,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.866499642090194e-05,
+      "loss": 0.5466,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.866320687186829e-05,
+      "loss": 0.4644,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.866141732283465e-05,
+      "loss": 0.6194,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8659627773801004e-05,
+      "loss": 0.5556,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.865783822476736e-05,
+      "loss": 0.4813,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8656048675733715e-05,
+      "loss": 0.6344,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8654259126700076e-05,
+      "loss": 0.5906,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.865246957766643e-05,
+      "loss": 0.577,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.865068002863279e-05,
+      "loss": 0.5525,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.864889047959914e-05,
+      "loss": 0.5762,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.86471009305655e-05,
+      "loss": 0.52,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.864531138153185e-05,
+      "loss": 0.474,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8643521832498214e-05,
+      "loss": 0.5725,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.864173228346457e-05,
+      "loss": 0.5036,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8639942734430925e-05,
+      "loss": 0.6416,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.863815318539729e-05,
+      "loss": 0.6321,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.863636363636364e-05,
+      "loss": 0.5014,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.863457408732999e-05,
+      "loss": 0.5026,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.863278453829635e-05,
+      "loss": 0.4747,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.863099498926271e-05,
+      "loss": 0.5787,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862920544022906e-05,
+      "loss": 0.6561,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8627415891195425e-05,
+      "loss": 0.5084,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862562634216178e-05,
+      "loss": 0.5674,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8623836793128135e-05,
+      "loss": 0.4923,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862204724409449e-05,
+      "loss": 0.526,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8620257695060845e-05,
+      "loss": 0.5529,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86184681460272e-05,
+      "loss": 0.5202,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8616678596993556e-05,
+      "loss": 0.4974,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.861488904795992e-05,
+      "loss": 0.4192,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.861309949892627e-05,
+      "loss": 0.5038,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.861130994989263e-05,
+      "loss": 0.5416,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.860952040085899e-05,
+      "loss": 0.5553,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8607730851825345e-05,
+      "loss": 0.5116,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8605941302791694e-05,
+      "loss": 0.6079,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8604151753758056e-05,
+      "loss": 0.5392,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.860236220472441e-05,
+      "loss": 0.6267,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8600572655690766e-05,
+      "loss": 0.5176,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.859878310665713e-05,
+      "loss": 0.5321,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.859699355762348e-05,
+      "loss": 0.486,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.859520400858984e-05,
+      "loss": 0.5574,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8593414459556194e-05,
+      "loss": 0.5851,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.859162491052255e-05,
+      "loss": 0.6097,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8589835361488904e-05,
+      "loss": 0.5087,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8588045812455266e-05,
+      "loss": 0.574,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.858625626342162e-05,
+      "loss": 0.5348,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8584466714387976e-05,
+      "loss": 0.5368,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.858267716535433e-05,
+      "loss": 0.5785,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8580887616320694e-05,
+      "loss": 0.5315,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.857909806728704e-05,
+      "loss": 0.6185,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.85773085182534e-05,
+      "loss": 0.6347,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.857551896921976e-05,
+      "loss": 0.5453,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8573729420186114e-05,
+      "loss": 0.5489,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.857193987115247e-05,
+      "loss": 0.5666,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.857015032211883e-05,
+      "loss": 0.5243,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.856836077308519e-05,
+      "loss": 0.5147,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.856657122405154e-05,
+      "loss": 0.5671,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.85647816750179e-05,
+      "loss": 0.5323,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.856299212598425e-05,
+      "loss": 0.5099,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.856120257695061e-05,
+      "loss": 0.4947,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.855941302791697e-05,
+      "loss": 0.4188,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8557623478883325e-05,
+      "loss": 0.5525,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.855583392984968e-05,
+      "loss": 0.6093,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8554044380816035e-05,
+      "loss": 0.4675,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.85522548317824e-05,
+      "loss": 0.4412,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8550465282748745e-05,
+      "loss": 0.546,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.854867573371511e-05,
+      "loss": 0.5072,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.854688618468146e-05,
+      "loss": 0.5655,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.854509663564782e-05,
+      "loss": 0.5442,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.854330708661417e-05,
+      "loss": 0.5973,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8541517537580535e-05,
+      "loss": 0.6073,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.853972798854689e-05,
+      "loss": 0.4313,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8537938439513245e-05,
+      "loss": 0.4454,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.85361488904796e-05,
+      "loss": 0.4963,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8534359341445956e-05,
+      "loss": 0.5041,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853256979241231e-05,
+      "loss": 0.4751,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853078024337867e-05,
+      "loss": 0.542,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852899069434503e-05,
+      "loss": 0.6056,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852720114531138e-05,
+      "loss": 0.5869,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8525411596277745e-05,
+      "loss": 0.6209,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.85236220472441e-05,
+      "loss": 0.5188,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852183249821045e-05,
+      "loss": 0.4475,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852004294917681e-05,
+      "loss": 0.399,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8518253400143166e-05,
+      "loss": 0.5876,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.851646385110952e-05,
+      "loss": 0.5433,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.851467430207588e-05,
+      "loss": 0.4632,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.851288475304224e-05,
+      "loss": 0.4674,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8511095204008594e-05,
+      "loss": 0.5673,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.850930565497495e-05,
+      "loss": 0.5182,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8507516105941304e-05,
+      "loss": 0.5405,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.850572655690766e-05,
+      "loss": 0.4724,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8503937007874014e-05,
+      "loss": 0.6276,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8502147458840376e-05,
+      "loss": 0.5203,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.850035790980673e-05,
+      "loss": 0.5832,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.849856836077309e-05,
+      "loss": 0.5189,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.849677881173945e-05,
+      "loss": 0.6331,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.84949892627058e-05,
+      "loss": 0.6208,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.849319971367215e-05,
+      "loss": 0.5636,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8491410164638514e-05,
+      "loss": 0.4987,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.848962061560487e-05,
+      "loss": 0.5316,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8487831066571225e-05,
+      "loss": 0.5458,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8486041517537587e-05,
+      "loss": 0.5025,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.848425196850394e-05,
+      "loss": 0.5064,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.84824624194703e-05,
+      "loss": 0.5741,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.848067287043665e-05,
+      "loss": 0.5984,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.847888332140301e-05,
+      "loss": 0.5538,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.847709377236936e-05,
+      "loss": 0.4879,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8475304223335725e-05,
+      "loss": 0.578,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.847351467430208e-05,
+      "loss": 0.4789,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8471725125268435e-05,
+      "loss": 0.4522,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.846993557623479e-05,
+      "loss": 0.5315,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.846814602720115e-05,
+      "loss": 0.5204,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.84663564781675e-05,
+      "loss": 0.4821,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8464566929133856e-05,
+      "loss": 0.4817,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.846277738010022e-05,
+      "loss": 0.5806,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.846098783106657e-05,
+      "loss": 0.5952,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.845919828203293e-05,
+      "loss": 0.6106,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.845740873299929e-05,
+      "loss": 0.5106,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8455619183965645e-05,
+      "loss": 0.4529,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8453829634932e-05,
+      "loss": 0.4622,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8452040085898356e-05,
+      "loss": 0.5778,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.845025053686471e-05,
+      "loss": 0.5149,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8448460987831066e-05,
+      "loss": 0.4181,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.844667143879743e-05,
+      "loss": 0.61,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.844488188976378e-05,
+      "loss": 0.7086,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.844309234073014e-05,
+      "loss": 0.4768,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8441302791696494e-05,
+      "loss": 0.5456,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8439513242662855e-05,
+      "loss": 0.5123,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8437723693629204e-05,
+      "loss": 0.5852,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8435934144595566e-05,
+      "loss": 0.489,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.843414459556192e-05,
+      "loss": 0.5008,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8432355046528276e-05,
+      "loss": 0.5107,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.843056549749463e-05,
+      "loss": 0.6336,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8428775948460993e-05,
+      "loss": 0.6792,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842698639942735e-05,
+      "loss": 0.4528,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.84251968503937e-05,
+      "loss": 0.5331,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842340730136006e-05,
+      "loss": 0.5104,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8421617752326414e-05,
+      "loss": 0.5891,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.841982820329277e-05,
+      "loss": 0.5303,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.841803865425913e-05,
+      "loss": 0.5942,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8416249105225487e-05,
+      "loss": 0.4785,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.841445955619184e-05,
+      "loss": 0.4349,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8412670007158204e-05,
+      "loss": 0.5421,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.841088045812455e-05,
+      "loss": 0.5356,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.840909090909091e-05,
+      "loss": 0.5483,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.840730136005727e-05,
+      "loss": 0.4941,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8405511811023624e-05,
+      "loss": 0.672,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.840372226198998e-05,
+      "loss": 0.4787,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8401932712956335e-05,
+      "loss": 0.5013,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.84001431639227e-05,
+      "loss": 0.5368,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839835361488905e-05,
+      "loss": 0.4813,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839656406585541e-05,
+      "loss": 0.5133,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839477451682176e-05,
+      "loss": 0.6288,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839298496778812e-05,
+      "loss": 0.4487,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.839119541875447e-05,
+      "loss": 0.4515,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8389405869720835e-05,
+      "loss": 0.475,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.838761632068719e-05,
+      "loss": 0.5395,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8385826771653545e-05,
+      "loss": 0.5844,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.838403722261991e-05,
+      "loss": 0.6566,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8382247673586256e-05,
+      "loss": 0.4816,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.838045812455261e-05,
+      "loss": 0.5485,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.837866857551897e-05,
+      "loss": 0.4506,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.837687902648533e-05,
+      "loss": 0.712,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.837508947745168e-05,
+      "loss": 0.4882,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8373299928418045e-05,
+      "loss": 0.549,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.83715103793844e-05,
+      "loss": 0.4754,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8369720830350755e-05,
+      "loss": 0.5291,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.836793128131711e-05,
+      "loss": 0.5376,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8366141732283466e-05,
+      "loss": 0.5156,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.836435218324982e-05,
+      "loss": 0.458,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.836256263421618e-05,
+      "loss": 0.6112,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.836077308518254e-05,
+      "loss": 0.5847,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.835898353614889e-05,
+      "loss": 0.4952,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.835719398711525e-05,
+      "loss": 0.4915,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.835540443808161e-05,
+      "loss": 0.4809,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.835361488904796e-05,
+      "loss": 0.4601,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8351825340014314e-05,
+      "loss": 0.5697,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8350035790980676e-05,
+      "loss": 0.6256,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.834824624194703e-05,
+      "loss": 0.5936,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8346456692913387e-05,
+      "loss": 0.5451,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.834466714387975e-05,
+      "loss": 0.4791,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8342877594846104e-05,
+      "loss": 0.5206,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.834108804581245e-05,
+      "loss": 0.4562,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8339298496778814e-05,
+      "loss": 0.5944,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.833750894774517e-05,
+      "loss": 0.5154,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8335719398711524e-05,
+      "loss": 0.5053,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8333929849677886e-05,
+      "loss": 0.6067,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.833214030064424e-05,
+      "loss": 0.5499,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83303507516106e-05,
+      "loss": 0.4767,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832856120257695e-05,
+      "loss": 0.4396,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832677165354331e-05,
+      "loss": 0.5733,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832498210450966e-05,
+      "loss": 0.5367,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8323192555476024e-05,
+      "loss": 0.5276,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832140300644238e-05,
+      "loss": 0.4728,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8319613457408735e-05,
+      "loss": 0.602,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.831782390837509e-05,
+      "loss": 0.5216,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.831603435934145e-05,
+      "loss": 0.7115,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.831424481030781e-05,
+      "loss": 0.5601,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8312455261274155e-05,
+      "loss": 0.4888,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.831066571224052e-05,
+      "loss": 0.4876,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.830887616320687e-05,
+      "loss": 0.526,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.830708661417323e-05,
+      "loss": 0.4896,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.830529706513959e-05,
+      "loss": 0.5071,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8303507516105945e-05,
+      "loss": 0.5447,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83017179670723e-05,
+      "loss": 0.496,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829992841803866e-05,
+      "loss": 0.4888,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829813886900501e-05,
+      "loss": 0.5709,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8296349319971366e-05,
+      "loss": 0.6177,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829455977093773e-05,
+      "loss": 0.4885,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829277022190408e-05,
+      "loss": 0.4315,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.829098067287044e-05,
+      "loss": 0.5065,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.828919112383679e-05,
+      "loss": 0.54,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8287401574803155e-05,
+      "loss": 0.5237,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.828561202576951e-05,
+      "loss": 0.664,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8283822476735866e-05,
+      "loss": 0.5158,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.828203292770222e-05,
+      "loss": 0.6478,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8280243378668576e-05,
+      "loss": 0.5843,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827845382963493e-05,
+      "loss": 0.5358,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827666428060129e-05,
+      "loss": 0.5282,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827487473156765e-05,
+      "loss": 0.5624,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8273085182534004e-05,
+      "loss": 0.5845,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8271295633500366e-05,
+      "loss": 0.5219,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8269506084466714e-05,
+      "loss": 0.5396,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.826771653543307e-05,
+      "loss": 0.5693,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.826592698639943e-05,
+      "loss": 0.4865,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8264137437365786e-05,
+      "loss": 0.552,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.826234788833214e-05,
+      "loss": 0.5612,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8260558339298504e-05,
+      "loss": 0.4883,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.825876879026486e-05,
+      "loss": 0.5689,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.825697924123121e-05,
+      "loss": 0.5429,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.825518969219757e-05,
+      "loss": 0.5738,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8253400143163924e-05,
+      "loss": 0.4977,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.825161059413028e-05,
+      "loss": 0.494,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824982104509664e-05,
+      "loss": 0.482,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8248031496063e-05,
+      "loss": 0.4906,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824624194702935e-05,
+      "loss": 0.464,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824445239799571e-05,
+      "loss": 0.4985,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824266284896206e-05,
+      "loss": 0.5437,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.824087329992842e-05,
+      "loss": 0.5297,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.823908375089477e-05,
+      "loss": 0.5187,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8237294201861135e-05,
+      "loss": 0.5274,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.823550465282749e-05,
+      "loss": 0.5394,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8233715103793845e-05,
+      "loss": 0.5807,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.823192555476021e-05,
+      "loss": 0.526,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.823013600572656e-05,
+      "loss": 0.6635,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.822834645669291e-05,
+      "loss": 0.6424,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.822655690765927e-05,
+      "loss": 0.431,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.822476735862563e-05,
+      "loss": 0.5121,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.822297780959198e-05,
+      "loss": 0.5562,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8221188260558345e-05,
+      "loss": 0.5897,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.82193987115247e-05,
+      "loss": 0.6119,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8217609162491055e-05,
+      "loss": 0.52,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.821581961345741e-05,
+      "loss": 0.5208,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8214030064423766e-05,
+      "loss": 0.6034,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.821224051539012e-05,
+      "loss": 0.5446,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.821045096635648e-05,
+      "loss": 0.4406,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.820866141732284e-05,
+      "loss": 0.5305,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.820687186828919e-05,
+      "loss": 0.5711,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.820508231925555e-05,
+      "loss": 0.4575,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.820329277022191e-05,
+      "loss": 0.493,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8201503221188266e-05,
+      "loss": 0.5879,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8199713672154614e-05,
+      "loss": 0.58,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8197924123120976e-05,
+      "loss": 0.431,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.819613457408733e-05,
+      "loss": 0.6367,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8194345025053686e-05,
+      "loss": 0.5771,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.819255547602005e-05,
+      "loss": 0.5043,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8190765926986403e-05,
+      "loss": 0.5738,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.818897637795276e-05,
+      "loss": 0.5096,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8187186828919114e-05,
+      "loss": 0.5689,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.818539727988547e-05,
+      "loss": 0.5478,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8183607730851824e-05,
+      "loss": 0.612,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8181818181818186e-05,
+      "loss": 0.5325,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.818002863278454e-05,
+      "loss": 0.5784,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8178239083750897e-05,
+      "loss": 0.5571,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.817644953471725e-05,
+      "loss": 0.5149,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8174659985683614e-05,
+      "loss": 0.5005,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.817287043664996e-05,
+      "loss": 0.506,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8171080887616324e-05,
+      "loss": 0.4377,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816929133858268e-05,
+      "loss": 0.591,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8167501789549035e-05,
+      "loss": 0.6381,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816571224051539e-05,
+      "loss": 0.5201,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816392269148175e-05,
+      "loss": 0.6053,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816213314244811e-05,
+      "loss": 0.4816,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.816034359341446e-05,
+      "loss": 0.542,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.815855404438082e-05,
+      "loss": 0.4755,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.815676449534717e-05,
+      "loss": 0.5988,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.815497494631353e-05,
+      "loss": 0.5006,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.815318539727989e-05,
+      "loss": 0.4575,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8151395848246245e-05,
+      "loss": 0.4186,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.81496062992126e-05,
+      "loss": 0.5062,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814781675017896e-05,
+      "loss": 0.4399,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814602720114532e-05,
+      "loss": 0.4904,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8144237652111666e-05,
+      "loss": 0.6252,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814244810307803e-05,
+      "loss": 0.5589,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.814065855404438e-05,
+      "loss": 0.6159,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.813886900501074e-05,
+      "loss": 0.4688,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.81370794559771e-05,
+      "loss": 0.5628,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8135289906943455e-05,
+      "loss": 0.3952,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.813350035790981e-05,
+      "loss": 0.5677,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8131710808876166e-05,
+      "loss": 0.603,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.812992125984252e-05,
+      "loss": 0.5236,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8128131710808876e-05,
+      "loss": 0.5264,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.812634216177523e-05,
+      "loss": 0.6027,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.812455261274159e-05,
+      "loss": 0.4909,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.812276306370795e-05,
+      "loss": 0.4925,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8120973514674303e-05,
+      "loss": 0.5091,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8119183965640665e-05,
+      "loss": 0.535,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.811739441660702e-05,
+      "loss": 0.6085,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.811560486757337e-05,
+      "loss": 0.4773,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.811381531853973e-05,
+      "loss": 0.5478,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8112025769506086e-05,
+      "loss": 0.406,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.811023622047244e-05,
+      "loss": 0.4407,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.81084466714388e-05,
+      "loss": 0.4684,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.810665712240516e-05,
+      "loss": 0.5322,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8104867573371514e-05,
+      "loss": 0.545,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.810307802433787e-05,
+      "loss": 0.5186,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8101288475304224e-05,
+      "loss": 0.5968,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.809949892627058e-05,
+      "loss": 0.5174,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.809770937723694e-05,
+      "loss": 0.425,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8095919828203296e-05,
+      "loss": 0.5364,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.809413027916965e-05,
+      "loss": 0.4737,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.809234073013601e-05,
+      "loss": 0.4392,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.809055118110237e-05,
+      "loss": 0.5237,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.808876163206872e-05,
+      "loss": 0.5345,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.808697208303507e-05,
+      "loss": 0.5201,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8085182534001434e-05,
+      "loss": 0.4748,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.808339298496779e-05,
+      "loss": 0.6037,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8081603435934145e-05,
+      "loss": 0.6075,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807981388690051e-05,
+      "loss": 0.512,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807802433786686e-05,
+      "loss": 0.5115,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807623478883322e-05,
+      "loss": 0.6081,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807444523979957e-05,
+      "loss": 0.5731,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807265569076593e-05,
+      "loss": 0.4563,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.807086614173228e-05,
+      "loss": 0.4643,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8069076592698645e-05,
+      "loss": 0.5516,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8067287043665e-05,
+      "loss": 0.6091,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8065497494631355e-05,
+      "loss": 0.4687,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.806370794559771e-05,
+      "loss": 0.4796,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.806191839656407e-05,
+      "loss": 0.4535,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.806012884753042e-05,
+      "loss": 0.659,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805833929849678e-05,
+      "loss": 0.4991,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805654974946314e-05,
+      "loss": 0.4794,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805476020042949e-05,
+      "loss": 0.5785,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805297065139585e-05,
+      "loss": 0.5618,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.805118110236221e-05,
+      "loss": 0.5387,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8049391553328565e-05,
+      "loss": 0.5682,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.804760200429492e-05,
+      "loss": 0.4393,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8045812455261276e-05,
+      "loss": 0.5165,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.804402290622763e-05,
+      "loss": 0.4957,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8042233357193986e-05,
+      "loss": 0.6705,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.804044380816035e-05,
+      "loss": 0.553,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.80386542591267e-05,
+      "loss": 0.5028,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.803686471009306e-05,
+      "loss": 0.4463,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.803507516105942e-05,
+      "loss": 0.4615,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.803328561202577e-05,
+      "loss": 0.5112,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8031496062992124e-05,
+      "loss": 0.4642,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8029706513958486e-05,
+      "loss": 0.4085,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802791696492484e-05,
+      "loss": 0.5503,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8026127415891196e-05,
+      "loss": 0.5822,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802433786685755e-05,
+      "loss": 0.5021,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8022548317823914e-05,
+      "loss": 0.4369,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802075876879027e-05,
+      "loss": 0.5576,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8018969219756624e-05,
+      "loss": 0.4916,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801717967072298e-05,
+      "loss": 0.5289,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8015390121689334e-05,
+      "loss": 0.5332,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801360057265569e-05,
+      "loss": 0.5775,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801181102362205e-05,
+      "loss": 0.5316,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.801002147458841e-05,
+      "loss": 0.4943,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.800823192555476e-05,
+      "loss": 0.504,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8006442376521124e-05,
+      "loss": 0.4614,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.800465282748747e-05,
+      "loss": 0.5189,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.800286327845383e-05,
+      "loss": 0.4444,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.800107372942019e-05,
+      "loss": 0.4438,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7999284180386545e-05,
+      "loss": 0.4854,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.79974946313529e-05,
+      "loss": 0.4951,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.799570508231926e-05,
+      "loss": 0.4412,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.799391553328562e-05,
+      "loss": 0.4619,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.799212598425197e-05,
+      "loss": 0.4341,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.799033643521833e-05,
+      "loss": 0.556,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.798854688618468e-05,
+      "loss": 0.5919,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.798675733715104e-05,
+      "loss": 0.4885,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.79849677881174e-05,
+      "loss": 0.5812,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7983178239083755e-05,
+      "loss": 0.4434,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.798138869005011e-05,
+      "loss": 0.446,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7979599141016465e-05,
+      "loss": 0.5162,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.797780959198283e-05,
+      "loss": 0.5217,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7976020042949176e-05,
+      "loss": 0.5916,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.797423049391553e-05,
+      "loss": 0.3941,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.797244094488189e-05,
+      "loss": 0.5248,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.797065139584825e-05,
+      "loss": 0.54,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.79688618468146e-05,
+      "loss": 0.6733,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7967072297780965e-05,
+      "loss": 0.4795,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.796528274874732e-05,
+      "loss": 0.5129,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7963493199713676e-05,
+      "loss": 0.4824,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.796170365068003e-05,
+      "loss": 0.5325,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7959914101646386e-05,
+      "loss": 0.4739,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.795812455261274e-05,
+      "loss": 0.5647,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.79563350035791e-05,
+      "loss": 0.6651,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.795454545454546e-05,
+      "loss": 0.5071,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7952755905511814e-05,
+      "loss": 0.5843,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.795096635647817e-05,
+      "loss": 0.5499,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7949176807444524e-05,
+      "loss": 0.5464,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.794738725841088e-05,
+      "loss": 0.4942,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.794559770937724e-05,
+      "loss": 0.4334,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7943808160343596e-05,
+      "loss": 0.6122,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.794201861130995e-05,
+      "loss": 0.4452,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.794022906227631e-05,
+      "loss": 0.5357,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.793843951324267e-05,
+      "loss": 0.466,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7936649964209024e-05,
+      "loss": 0.5194,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793486041517537e-05,
+      "loss": 0.4217,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7933070866141734e-05,
+      "loss": 0.4902,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793128131710809e-05,
+      "loss": 0.5383,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7929491768074445e-05,
+      "loss": 0.4684,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7927702219040807e-05,
+      "loss": 0.5714,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792591267000716e-05,
+      "loss": 0.4761,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792412312097352e-05,
+      "loss": 0.3971,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792233357193988e-05,
+      "loss": 0.4163,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792054402290623e-05,
+      "loss": 0.4237,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.791875447387258e-05,
+      "loss": 0.4353,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7916964924838945e-05,
+      "loss": 0.4994,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.79151753758053e-05,
+      "loss": 0.5664,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7913385826771655e-05,
+      "loss": 0.4843,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.791159627773801e-05,
+      "loss": 0.5436,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790980672870437e-05,
+      "loss": 0.4343,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790801717967073e-05,
+      "loss": 0.4866,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790622763063708e-05,
+      "loss": 0.4738,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790443808160344e-05,
+      "loss": 0.5958,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790264853256979e-05,
+      "loss": 0.5108,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790085898353615e-05,
+      "loss": 0.5355,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789906943450251e-05,
+      "loss": 0.4436,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7897279885468865e-05,
+      "loss": 0.4762,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789549033643522e-05,
+      "loss": 0.4301,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789370078740158e-05,
+      "loss": 0.5608,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.789191123836793e-05,
+      "loss": 0.5292,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7890121689334286e-05,
+      "loss": 0.4604,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.788833214030065e-05,
+      "loss": 0.4549,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7886542591267e-05,
+      "loss": 0.433,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.788475304223336e-05,
+      "loss": 0.4965,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.788296349319972e-05,
+      "loss": 0.5337,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7881173944166075e-05,
+      "loss": 0.4515,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787938439513243e-05,
+      "loss": 0.5068,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7877594846098786e-05,
+      "loss": 0.5562,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787580529706514e-05,
+      "loss": 0.4226,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7874015748031496e-05,
+      "loss": 0.4542,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787222619899786e-05,
+      "loss": 0.4463,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7870436649964213e-05,
+      "loss": 0.4763,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786864710093057e-05,
+      "loss": 0.4835,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7866857551896924e-05,
+      "loss": 0.5623,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786506800286328e-05,
+      "loss": 0.3894,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7863278453829634e-05,
+      "loss": 0.5391,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.786148890479599e-05,
+      "loss": 0.4883,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785969935576235e-05,
+      "loss": 0.6655,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7857909806728707e-05,
+      "loss": 0.4802,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785612025769506e-05,
+      "loss": 0.5441,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7854330708661424e-05,
+      "loss": 0.6904,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785254115962778e-05,
+      "loss": 0.5007,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.785075161059413e-05,
+      "loss": 0.4683,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784896206156049e-05,
+      "loss": 0.5246,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7847172512526844e-05,
+      "loss": 0.4843,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.78453829634932e-05,
+      "loss": 0.4265,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784359341445956e-05,
+      "loss": 0.5118,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784180386542592e-05,
+      "loss": 0.5505,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784001431639227e-05,
+      "loss": 0.5677,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783822476735863e-05,
+      "loss": 0.5488,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.783643521832498e-05,
+      "loss": 0.4416,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.783464566929134e-05,
+      "loss": 0.4606,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.78328561202577e-05,
+      "loss": 0.4129,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7831066571224055e-05,
+      "loss": 0.5802,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782927702219041e-05,
+      "loss": 0.4251,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7827487473156765e-05,
+      "loss": 0.5427,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782569792412313e-05,
+      "loss": 0.4121,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782390837508948e-05,
+      "loss": 0.5548,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782211882605583e-05,
+      "loss": 0.4054,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782032927702219e-05,
+      "loss": 0.6091,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.781853972798855e-05,
+      "loss": 0.6266,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.78167501789549e-05,
+      "loss": 0.6551,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7814960629921265e-05,
+      "loss": 0.5375,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.781317108088762e-05,
+      "loss": 0.5957,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7811381531853975e-05,
+      "loss": 0.526,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.780959198282034e-05,
+      "loss": 0.5754,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7807802433786686e-05,
+      "loss": 0.4915,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.780601288475304e-05,
+      "loss": 0.4642,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.78042233357194e-05,
+      "loss": 0.4423,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.780243378668576e-05,
+      "loss": 0.5079,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.780064423765211e-05,
+      "loss": 0.4894,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.779885468861847e-05,
+      "loss": 0.5372,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.779706513958483e-05,
+      "loss": 0.4904,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.779527559055118e-05,
+      "loss": 0.5638,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.779348604151754e-05,
+      "loss": 0.4539,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7791696492483896e-05,
+      "loss": 0.4303,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.778990694345025e-05,
+      "loss": 0.4779,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7788117394416606e-05,
+      "loss": 0.5284,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.778632784538297e-05,
+      "loss": 0.4747,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7784538296349324e-05,
+      "loss": 0.4369,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.778274874731568e-05,
+      "loss": 0.5973,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7780959198282034e-05,
+      "loss": 0.4834,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.777916964924839e-05,
+      "loss": 0.5044,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7777380100214744e-05,
+      "loss": 0.4618,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7775590551181106e-05,
+      "loss": 0.4964,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.777380100214746e-05,
+      "loss": 0.6015,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.777201145311382e-05,
+      "loss": 0.5037,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.777022190408018e-05,
+      "loss": 0.4448,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7768432355046534e-05,
+      "loss": 0.4084,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.776664280601288e-05,
+      "loss": 0.4918,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7764853256979244e-05,
+      "loss": 0.5273,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.77630637079456e-05,
+      "loss": 0.4213,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7761274158911955e-05,
+      "loss": 0.4705,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775948460987832e-05,
+      "loss": 0.4143,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775769506084467e-05,
+      "loss": 0.4644,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775590551181103e-05,
+      "loss": 0.4269,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775411596277738e-05,
+      "loss": 0.569,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775232641374374e-05,
+      "loss": 0.5326,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.775053686471009e-05,
+      "loss": 0.5146,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.774874731567645e-05,
+      "loss": 0.4239,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.774695776664281e-05,
+      "loss": 0.5092,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7745168217609165e-05,
+      "loss": 0.4862,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.774337866857552e-05,
+      "loss": 0.4773,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.774158911954188e-05,
+      "loss": 0.5068,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.773979957050824e-05,
+      "loss": 0.422,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7738010021474586e-05,
+      "loss": 0.5645,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.773622047244095e-05,
+      "loss": 0.5331,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.77344309234073e-05,
+      "loss": 0.6338,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.773264137437366e-05,
+      "loss": 0.4336,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.773085182534002e-05,
+      "loss": 0.5474,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7729062276306375e-05,
+      "loss": 0.5209,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.5909,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7725483178239086e-05,
+      "loss": 0.5247,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.772369362920544e-05,
+      "loss": 0.566,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7721904080171796e-05,
+      "loss": 0.5413,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.772011453113816e-05,
+      "loss": 0.4823,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.771832498210451e-05,
+      "loss": 0.48,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.771653543307087e-05,
+      "loss": 0.5807,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7714745884037224e-05,
+      "loss": 0.5271,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7712956335003586e-05,
+      "loss": 0.5165,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7711166785969934e-05,
+      "loss": 0.4644,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.770937723693629e-05,
+      "loss": 0.4955,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.770758768790265e-05,
+      "loss": 0.5026,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7705798138869006e-05,
+      "loss": 0.5867,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.770400858983536e-05,
+      "loss": 0.6385,
+      "step": 1333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7702219040801724e-05,
+      "loss": 0.4424,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.770042949176808e-05,
+      "loss": 0.5114,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7698639942734434e-05,
+      "loss": 0.6074,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.769685039370079e-05,
+      "loss": 0.4945,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7695060844667144e-05,
+      "loss": 0.4332,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.76932712956335e-05,
+      "loss": 0.4747,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.769148174659986e-05,
+      "loss": 0.5186,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.768969219756622e-05,
+      "loss": 0.4168,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.768790264853257e-05,
+      "loss": 0.5306,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.768611309949893e-05,
+      "loss": 0.4603,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.768432355046529e-05,
+      "loss": 0.4189,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.768253400143164e-05,
+      "loss": 0.5123,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7680744452398e-05,
+      "loss": 0.5521,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7678954903364355e-05,
+      "loss": 0.4633,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.767716535433071e-05,
+      "loss": 0.4569,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7675375805297065e-05,
+      "loss": 0.4772,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.767358625626343e-05,
+      "loss": 0.4589,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.767179670722978e-05,
+      "loss": 0.4639,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.767000715819614e-05,
+      "loss": 0.4554,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.766821760916249e-05,
+      "loss": 0.7226,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.766642806012885e-05,
+      "loss": 0.5015,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.76646385110952e-05,
+      "loss": 0.5118,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7662848962061565e-05,
+      "loss": 0.4108,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.766105941302792e-05,
+      "loss": 0.445,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7659269863994275e-05,
+      "loss": 0.4999,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.765748031496064e-05,
+      "loss": 0.5311,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.765569076592699e-05,
+      "loss": 0.5749,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.765390121689334e-05,
+      "loss": 0.4596,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.76521116678597e-05,
+      "loss": 0.6357,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.765032211882606e-05,
+      "loss": 0.5207,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.764853256979241e-05,
+      "loss": 0.4517,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.764674302075877e-05,
+      "loss": 0.5432,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.764495347172513e-05,
+      "loss": 0.5609,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7643163922691486e-05,
+      "loss": 0.4663,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.764137437365784e-05,
+      "loss": 0.5308,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7639584824624196e-05,
+      "loss": 0.6382,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.763779527559055e-05,
+      "loss": 0.4829,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7636005726556906e-05,
+      "loss": 0.4951,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.763421617752327e-05,
+      "loss": 0.5043,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7632426628489623e-05,
+      "loss": 0.6393,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.763063707945598e-05,
+      "loss": 0.4728,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762884753042234e-05,
+      "loss": 0.4883,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762705798138869e-05,
+      "loss": 0.4938,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7625268432355044e-05,
+      "loss": 0.4446,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7623478883321406e-05,
+      "loss": 0.4973,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762168933428776e-05,
+      "loss": 0.4861,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7619899785254117e-05,
+      "loss": 0.5271,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.761811023622048e-05,
+      "loss": 0.5066,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7616320687186834e-05,
+      "loss": 0.4533,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.761453113815319e-05,
+      "loss": 0.432,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7612741589119544e-05,
+      "loss": 0.5196,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.76109520400859e-05,
+      "loss": 0.5792,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7609162491052255e-05,
+      "loss": 0.5572,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7607372942018617e-05,
+      "loss": 0.3989,
+      "step": 1387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.760558339298497e-05,
+      "loss": 0.4623,
+      "step": 1388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.760379384395133e-05,
+      "loss": 0.4982,
+      "step": 1389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.760200429491768e-05,
+      "loss": 0.5088,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7600214745884044e-05,
+      "loss": 0.5527,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.759842519685039e-05,
+      "loss": 0.4135,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.759663564781675e-05,
+      "loss": 0.496,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.759484609878311e-05,
+      "loss": 0.4636,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7593056549749465e-05,
+      "loss": 0.4485,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.759126700071582e-05,
+      "loss": 0.4045,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.758947745168218e-05,
+      "loss": 0.4404,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.758768790264854e-05,
+      "loss": 0.4326,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.758589835361489e-05,
+      "loss": 0.4941,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.758410880458125e-05,
+      "loss": 0.4043,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.75823192555476e-05,
+      "loss": 0.394,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.758052970651396e-05,
+      "loss": 0.4966,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.757874015748032e-05,
+      "loss": 0.632,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7576950608446675e-05,
+      "loss": 0.4906,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.757516105941303e-05,
+      "loss": 0.4513,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7573371510379385e-05,
+      "loss": 0.54,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.757158196134575e-05,
+      "loss": 0.5149,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7569792412312096e-05,
+      "loss": 0.4913,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.756800286327846e-05,
+      "loss": 0.5214,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.756621331424481e-05,
+      "loss": 0.5066,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.756442376521117e-05,
+      "loss": 0.491,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7562634216177523e-05,
+      "loss": 0.498,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7560844667143885e-05,
+      "loss": 0.5305,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.755905511811024e-05,
+      "loss": 0.482,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.755726556907659e-05,
+      "loss": 0.4827,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.755547602004295e-05,
+      "loss": 0.4022,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7553686471009306e-05,
+      "loss": 0.493,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.755189692197566e-05,
+      "loss": 0.5006,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.755010737294202e-05,
+      "loss": 0.4282,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.754831782390838e-05,
+      "loss": 0.4896,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7546528274874734e-05,
+      "loss": 0.5169,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7544738725841096e-05,
+      "loss": 0.4645,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7542949176807444e-05,
+      "loss": 0.412,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.75411596277738e-05,
+      "loss": 0.4872,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.753937007874016e-05,
+      "loss": 0.4975,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7537580529706516e-05,
+      "loss": 0.6036,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.753579098067287e-05,
+      "loss": 0.4431,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.753400143163923e-05,
+      "loss": 0.5654,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.753221188260559e-05,
+      "loss": 0.4534,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7530422333571944e-05,
+      "loss": 0.4746,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75286327845383e-05,
+      "loss": 0.4173,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7526843235504654e-05,
+      "loss": 0.3933,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.752505368647101e-05,
+      "loss": 0.555,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7523264137437365e-05,
+      "loss": 0.6298,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.752147458840373e-05,
+      "loss": 0.4824,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.751968503937008e-05,
+      "loss": 0.5049,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.751789549033644e-05,
+      "loss": 0.4866,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75161059413028e-05,
+      "loss": 0.4875,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.751431639226915e-05,
+      "loss": 0.3826,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75125268432355e-05,
+      "loss": 0.5093,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7510737294201865e-05,
+      "loss": 0.4691,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.750894774516822e-05,
+      "loss": 0.5769,
+      "step": 1442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7507158196134575e-05,
+      "loss": 0.5009,
+      "step": 1443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.750536864710094e-05,
+      "loss": 0.4514,
+      "step": 1444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.750357909806729e-05,
+      "loss": 0.4177,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.750178954903365e-05,
+      "loss": 0.5134,
+      "step": 1446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75e-05,
+      "loss": 0.4405,
+      "step": 1447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.749821045096636e-05,
+      "loss": 0.5424,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.749642090193271e-05,
+      "loss": 0.4793,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7494631352899075e-05,
+      "loss": 0.4187,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.749284180386543e-05,
+      "loss": 0.6511,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7491052254831785e-05,
+      "loss": 0.4586,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.748926270579814e-05,
+      "loss": 0.4403,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.74874731567645e-05,
+      "loss": 0.3903,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.748568360773085e-05,
+      "loss": 0.5123,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7483894058697206e-05,
+      "loss": 0.4226,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.748210450966357e-05,
+      "loss": 0.5993,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.748031496062992e-05,
+      "loss": 0.4873,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.747852541159628e-05,
+      "loss": 0.5316,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.747673586256264e-05,
+      "loss": 0.6687,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7474946313528996e-05,
+      "loss": 0.5466,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7473156764495344e-05,
+      "loss": 0.3647,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7471367215461706e-05,
+      "loss": 0.37,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.746957766642806e-05,
+      "loss": 0.5537,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7467788117394416e-05,
+      "loss": 0.5462,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.746599856836078e-05,
+      "loss": 0.557,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7464209019327134e-05,
+      "loss": 0.4532,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.746241947029349e-05,
+      "loss": 0.5475,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7460629921259844e-05,
+      "loss": 0.4429,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.74588403722262e-05,
+      "loss": 0.5187,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7457050823192554e-05,
+      "loss": 0.3851,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7455261274158916e-05,
+      "loss": 0.491,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.745347172512527e-05,
+      "loss": 0.5646,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.745168217609163e-05,
+      "loss": 0.5217,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.744989262705798e-05,
+      "loss": 0.4912,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7448103078024344e-05,
+      "loss": 0.4586,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.74463135289907e-05,
+      "loss": 0.3871,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.744452397995705e-05,
+      "loss": 0.4132,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.744273443092341e-05,
+      "loss": 0.4102,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7440944881889765e-05,
+      "loss": 0.3761,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.743915533285612e-05,
+      "loss": 0.5806,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.743736578382248e-05,
+      "loss": 0.4257,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.743557623478884e-05,
+      "loss": 0.5475,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.743378668575519e-05,
+      "loss": 0.4342,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7431997136721554e-05,
+      "loss": 0.4759,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.74302075876879e-05,
+      "loss": 0.5484,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742841803865426e-05,
+      "loss": 0.4772,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742662848962062e-05,
+      "loss": 0.4783,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7424838940586975e-05,
+      "loss": 0.4834,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742304939155333e-05,
+      "loss": 0.4098,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7421259842519685e-05,
+      "loss": 0.5516,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.741947029348605e-05,
+      "loss": 0.5777,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.74176807444524e-05,
+      "loss": 0.4886,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.741589119541876e-05,
+      "loss": 0.3835,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.741410164638511e-05,
+      "loss": 0.4642,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.741231209735147e-05,
+      "loss": 0.5148,
+      "step": 1496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.741052254831782e-05,
+      "loss": 0.6704,
+      "step": 1497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7408732999284185e-05,
+      "loss": 0.5962,
+      "step": 1498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.740694345025054e-05,
+      "loss": 0.6174,
+      "step": 1499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7405153901216896e-05,
+      "loss": 0.4762,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.740336435218325e-05,
+      "loss": 0.4805,
+      "step": 1501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7401574803149606e-05,
+      "loss": 0.4226,
+      "step": 1502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.739978525411596e-05,
+      "loss": 0.5533,
+      "step": 1503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.739799570508232e-05,
+      "loss": 0.3823,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.739620615604868e-05,
+      "loss": 0.4772,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7394416607015034e-05,
+      "loss": 0.5068,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7392627057981396e-05,
+      "loss": 0.4402,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.739083750894775e-05,
+      "loss": 0.3879,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73890479599141e-05,
+      "loss": 0.5369,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.738725841088046e-05,
+      "loss": 0.5295,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7385468861846816e-05,
+      "loss": 0.4753,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.738367931281317e-05,
+      "loss": 0.6305,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.738188976377953e-05,
+      "loss": 0.5315,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.738010021474589e-05,
+      "loss": 0.4923,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7378310665712244e-05,
+      "loss": 0.3972,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73765211166786e-05,
+      "loss": 0.3151,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7374731567644954e-05,
+      "loss": 0.4407,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.737294201861131e-05,
+      "loss": 0.371,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7371152469577665e-05,
+      "loss": 0.4954,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7369362920544027e-05,
+      "loss": 0.4155,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.736757337151038e-05,
+      "loss": 0.3298,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.736578382247674e-05,
+      "loss": 0.5401,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73639942734431e-05,
+      "loss": 0.4122,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7362204724409454e-05,
+      "loss": 0.5657,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73604151753758e-05,
+      "loss": 0.4776,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7358625626342164e-05,
+      "loss": 0.6674,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.735683607730852e-05,
+      "loss": 0.5749,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7355046528274875e-05,
+      "loss": 0.5224,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.735325697924124e-05,
+      "loss": 0.4414,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.735146743020759e-05,
+      "loss": 0.4905,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.734967788117395e-05,
+      "loss": 0.482,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73478883321403e-05,
+      "loss": 0.4738,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.734609878310666e-05,
+      "loss": 0.4644,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.734430923407301e-05,
+      "loss": 0.5749,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7342519685039375e-05,
+      "loss": 0.501,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.734073013600573e-05,
+      "loss": 0.5591,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7338940586972085e-05,
+      "loss": 0.4625,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.733715103793844e-05,
+      "loss": 0.3778,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.73353614889048e-05,
+      "loss": 0.5684,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.733357193987116e-05,
+      "loss": 0.4921,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7331782390837506e-05,
+      "loss": 0.546,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732999284180387e-05,
+      "loss": 0.4381,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732820329277022e-05,
+      "loss": 0.6232,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732641374373658e-05,
+      "loss": 0.5507,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732462419470294e-05,
+      "loss": 0.4278,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7322834645669295e-05,
+      "loss": 0.465,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732104509663565e-05,
+      "loss": 0.5068,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7319255547602006e-05,
+      "loss": 0.5064,
+      "step": 1548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.731746599856836e-05,
+      "loss": 0.5585,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7315676449534716e-05,
+      "loss": 0.4739,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.731388690050108e-05,
+      "loss": 0.4168,
+      "step": 1551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7312097351467433e-05,
+      "loss": 0.4426,
+      "step": 1552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.731030780243379e-05,
+      "loss": 0.5469,
+      "step": 1553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7308518253400144e-05,
+      "loss": 0.41,
+      "step": 1554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7306728704366506e-05,
+      "loss": 0.4333,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7304939155332854e-05,
+      "loss": 0.4324,
+      "step": 1556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7303149606299216e-05,
+      "loss": 0.5102,
+      "step": 1557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.730136005726557e-05,
+      "loss": 0.4893,
+      "step": 1558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7299570508231927e-05,
+      "loss": 0.4472,
+      "step": 1559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.729778095919828e-05,
+      "loss": 0.5446,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7295991410164644e-05,
+      "loss": 0.5041,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7294201861131e-05,
+      "loss": 0.3791,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7292412312097354e-05,
+      "loss": 0.4376,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.729062276306371e-05,
+      "loss": 0.4625,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7288833214030064e-05,
+      "loss": 0.5044,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.728704366499642e-05,
+      "loss": 0.5241,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.728525411596278e-05,
+      "loss": 0.4204,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.728346456692914e-05,
+      "loss": 0.4909,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.728167501789549e-05,
+      "loss": 0.4131,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7279885468861854e-05,
+      "loss": 0.4614,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.727809591982821e-05,
+      "loss": 0.6767,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.727630637079456e-05,
+      "loss": 0.5309,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.727451682176092e-05,
+      "loss": 0.3976,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7272727272727275e-05,
+      "loss": 0.4952,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.727093772369363e-05,
+      "loss": 0.4372,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7269148174659985e-05,
+      "loss": 0.4987,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.726735862562635e-05,
+      "loss": 0.5065,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.72655690765927e-05,
+      "loss": 0.4154,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.726377952755906e-05,
+      "loss": 0.4744,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.726198997852541e-05,
+      "loss": 0.52,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.726020042949177e-05,
+      "loss": 0.4879,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.725841088045812e-05,
+      "loss": 0.5286,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7256621331424485e-05,
+      "loss": 0.5496,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.725483178239084e-05,
+      "loss": 0.4947,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7253042233357195e-05,
+      "loss": 0.4439,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.725125268432356e-05,
+      "loss": 0.4367,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.724946313528991e-05,
+      "loss": 0.4405,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.724767358625626e-05,
+      "loss": 0.4879,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.724588403722262e-05,
+      "loss": 0.4692,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.724409448818898e-05,
+      "loss": 0.4457,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.724230493915533e-05,
+      "loss": 0.5445,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7240515390121695e-05,
+      "loss": 0.4384,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.723872584108805e-05,
+      "loss": 0.469,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7236936292054406e-05,
+      "loss": 0.4586,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.723514674302076e-05,
+      "loss": 0.5382,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7233357193987116e-05,
+      "loss": 0.4341,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.723156764495347e-05,
+      "loss": 0.5838,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722977809591983e-05,
+      "loss": 0.5477,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722798854688619e-05,
+      "loss": 0.4611,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7226198997852544e-05,
+      "loss": 0.4114,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.72244094488189e-05,
+      "loss": 0.4659,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722261989978526e-05,
+      "loss": 0.5321,
+      "step": 1602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722083035075161e-05,
+      "loss": 0.4553,
+      "step": 1603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7219040801717964e-05,
+      "loss": 0.4215,
+      "step": 1604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7217251252684326e-05,
+      "loss": 0.5293,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.721546170365068e-05,
+      "loss": 0.5074,
+      "step": 1606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.721367215461704e-05,
+      "loss": 0.5054,
+      "step": 1607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.72118826055834e-05,
+      "loss": 0.5168,
+      "step": 1608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7210093056549754e-05,
+      "loss": 0.3963,
+      "step": 1609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.720830350751611e-05,
+      "loss": 0.603,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7206513958482464e-05,
+      "loss": 0.4643,
+      "step": 1611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.720472440944882e-05,
+      "loss": 0.4761,
+      "step": 1612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7202934860415175e-05,
+      "loss": 0.3814,
+      "step": 1613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.720114531138154e-05,
+      "loss": 0.5538,
+      "step": 1614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.719935576234789e-05,
+      "loss": 0.5354,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.719756621331425e-05,
+      "loss": 0.6058,
+      "step": 1616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.71957766642806e-05,
+      "loss": 0.5105,
+      "step": 1617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7193987115246964e-05,
+      "loss": 0.5286,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.719219756621331e-05,
+      "loss": 0.4904,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7190408017179675e-05,
+      "loss": 0.3816,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.718861846814603e-05,
+      "loss": 0.5178,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7186828919112385e-05,
+      "loss": 0.5218,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.718503937007874e-05,
+      "loss": 0.5069,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.71832498210451e-05,
+      "loss": 0.4283,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.718146027201146e-05,
+      "loss": 0.5516,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.717967072297781e-05,
+      "loss": 0.4712,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.717788117394417e-05,
+      "loss": 0.5789,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.717609162491052e-05,
+      "loss": 0.5707,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.717430207587688e-05,
+      "loss": 0.3388,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.717251252684324e-05,
+      "loss": 0.5596,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7170722977809595e-05,
+      "loss": 0.4041,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.716893342877595e-05,
+      "loss": 0.5388,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.716714387974231e-05,
+      "loss": 0.3864,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.716535433070866e-05,
+      "loss": 0.4463,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7163564781675016e-05,
+      "loss": 0.4673,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.716177523264138e-05,
+      "loss": 0.4944,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.715998568360773e-05,
+      "loss": 0.4142,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.715819613457409e-05,
+      "loss": 0.6557,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7156406585540444e-05,
+      "loss": 0.572,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7154617036506806e-05,
+      "loss": 0.6073,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.715282748747316e-05,
+      "loss": 0.4866,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7151037938439516e-05,
+      "loss": 0.5698,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.714924838940587e-05,
+      "loss": 0.6726,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7147458840372226e-05,
+      "loss": 0.3855,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.714566929133858e-05,
+      "loss": 0.4895,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7143879742304944e-05,
+      "loss": 0.4232,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.71420901932713e-05,
+      "loss": 0.4864,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7140300644237654e-05,
+      "loss": 0.3703,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7138511095204016e-05,
+      "loss": 0.419,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7136721546170364e-05,
+      "loss": 0.4223,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.713493199713672e-05,
+      "loss": 0.4624,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.713314244810308e-05,
+      "loss": 0.4847,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.713135289906944e-05,
+      "loss": 0.4164,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.712956335003579e-05,
+      "loss": 0.4592,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7127773801002154e-05,
+      "loss": 0.4907,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.712598425196851e-05,
+      "loss": 0.4489,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7124194702934864e-05,
+      "loss": 0.4113,
+      "step": 1657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.712240515390122e-05,
+      "loss": 0.5055,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7120615604867575e-05,
+      "loss": 0.4161,
+      "step": 1659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.711882605583393e-05,
+      "loss": 0.493,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.711703650680029e-05,
+      "loss": 0.4644,
+      "step": 1661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.711524695776665e-05,
+      "loss": 0.4296,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7113457408733e-05,
+      "loss": 0.5014,
+      "step": 1663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.711166785969936e-05,
+      "loss": 0.5184,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.710987831066572e-05,
+      "loss": 0.3555,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.710808876163207e-05,
+      "loss": 0.4201,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.710629921259842e-05,
+      "loss": 0.4598,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7104509663564785e-05,
+      "loss": 0.3592,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.710272011453114e-05,
+      "loss": 0.3436,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7100930565497495e-05,
+      "loss": 0.4346,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709914101646386e-05,
+      "loss": 0.4383,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709735146743021e-05,
+      "loss": 0.5636,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709556191839657e-05,
+      "loss": 0.4258,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709377236936292e-05,
+      "loss": 0.3561,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709198282032928e-05,
+      "loss": 0.3875,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.709019327129563e-05,
+      "loss": 0.4929,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7088403722261995e-05,
+      "loss": 0.547,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.708661417322835e-05,
+      "loss": 0.4483,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7084824624194706e-05,
+      "loss": 0.4729,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.708303507516106e-05,
+      "loss": 0.474,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7081245526127416e-05,
+      "loss": 0.4512,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.707945597709377e-05,
+      "loss": 0.449,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.707766642806013e-05,
+      "loss": 0.4228,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.707587687902649e-05,
+      "loss": 0.3963,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7074087329992843e-05,
+      "loss": 0.5637,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.70722977809592e-05,
+      "loss": 0.4629,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.707050823192556e-05,
+      "loss": 0.5759,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7068718682891916e-05,
+      "loss": 0.4872,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7066929133858264e-05,
+      "loss": 0.5001,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7065139584824626e-05,
+      "loss": 0.431,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.706335003579098e-05,
+      "loss": 0.4877,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7061560486757337e-05,
+      "loss": 0.4762,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.70597709377237e-05,
+      "loss": 0.4186,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7057981388690054e-05,
+      "loss": 0.4592,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.705619183965641e-05,
+      "loss": 0.5921,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.705440229062277e-05,
+      "loss": 0.528,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.705261274158912e-05,
+      "loss": 0.4938,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7050823192555475e-05,
+      "loss": 0.5354,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7049033643521836e-05,
+      "loss": 0.5158,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.704724409448819e-05,
+      "loss": 0.4342,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.704545454545455e-05,
+      "loss": 0.4791,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.70436649964209e-05,
+      "loss": 0.4028,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7041875447387264e-05,
+      "loss": 0.4436,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.704008589835362e-05,
+      "loss": 0.4496,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7038296349319974e-05,
+      "loss": 0.4794,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.703650680028633e-05,
+      "loss": 0.4803,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7034717251252685e-05,
+      "loss": 0.4306,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.703292770221904e-05,
+      "loss": 0.5393,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.70311381531854e-05,
+      "loss": 0.442,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702934860415176e-05,
+      "loss": 0.3698,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702755905511811e-05,
+      "loss": 0.4788,
+      "step": 1711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7025769506084474e-05,
+      "loss": 0.4272,
+      "step": 1712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702397995705082e-05,
+      "loss": 0.4969,
+      "step": 1713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702219040801718e-05,
+      "loss": 0.4317,
+      "step": 1714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702040085898354e-05,
+      "loss": 0.3679,
+      "step": 1715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7018611309949895e-05,
+      "loss": 0.4562,
+      "step": 1716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.701682176091625e-05,
+      "loss": 0.3669,
+      "step": 1717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.701503221188261e-05,
+      "loss": 0.4631,
+      "step": 1718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.701324266284897e-05,
+      "loss": 0.5514,
+      "step": 1719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7011453113815316e-05,
+      "loss": 0.4076,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.700966356478168e-05,
+      "loss": 0.4328,
+      "step": 1721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.700787401574803e-05,
+      "loss": 0.5024,
+      "step": 1722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.700608446671439e-05,
+      "loss": 0.4587,
+      "step": 1723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7004294917680743e-05,
+      "loss": 0.5085,
+      "step": 1724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7002505368647105e-05,
+      "loss": 0.4766,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.700071581961346e-05,
+      "loss": 0.4874,
+      "step": 1726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6998926270579816e-05,
+      "loss": 0.4561,
+      "step": 1727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.699713672154617e-05,
+      "loss": 0.3705,
+      "step": 1728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6995347172512526e-05,
+      "loss": 0.5941,
+      "step": 1729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.699355762347888e-05,
+      "loss": 0.4657,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.699176807444524e-05,
+      "loss": 0.4154,
+      "step": 1731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.69899785254116e-05,
+      "loss": 0.5225,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6988188976377954e-05,
+      "loss": 0.4698,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6986399427344316e-05,
+      "loss": 0.3568,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.698460987831067e-05,
+      "loss": 0.4472,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.698282032927702e-05,
+      "loss": 0.4964,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.698103078024338e-05,
+      "loss": 0.4014,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6979241231209736e-05,
+      "loss": 0.5212,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.697745168217609e-05,
+      "loss": 0.6608,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6975662133142454e-05,
+      "loss": 0.4687,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.697387258410881e-05,
+      "loss": 0.6724,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6972083035075164e-05,
+      "loss": 0.4962,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.697029348604152e-05,
+      "loss": 0.5131,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6968503937007874e-05,
+      "loss": 0.5008,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.696671438797423e-05,
+      "loss": 0.3893,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.696492483894059e-05,
+      "loss": 0.5144,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.696313528990695e-05,
+      "loss": 0.4567,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.69613457408733e-05,
+      "loss": 0.5469,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.695955619183966e-05,
+      "loss": 0.4776,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.695776664280602e-05,
+      "loss": 0.5137,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6955977093772374e-05,
+      "loss": 0.4276,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.695418754473872e-05,
+      "loss": 0.4147,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6952397995705085e-05,
+      "loss": 0.5456,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.695060844667144e-05,
+      "loss": 0.391,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6948818897637795e-05,
+      "loss": 0.4127,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694702934860416e-05,
+      "loss": 0.4962,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694523979957051e-05,
+      "loss": 0.4731,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694345025053687e-05,
+      "loss": 0.6023,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.694166070150323e-05,
+      "loss": 0.3917,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.693987115246958e-05,
+      "loss": 0.4172,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.693808160343593e-05,
+      "loss": 0.4279,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6936292054402295e-05,
+      "loss": 0.4564,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.693450250536865e-05,
+      "loss": 0.4247,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6932712956335005e-05,
+      "loss": 0.4933,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.693092340730136e-05,
+      "loss": 0.5343,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692913385826772e-05,
+      "loss": 0.4351,
+      "step": 1766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692734430923407e-05,
+      "loss": 0.4429,
+      "step": 1767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692555476020043e-05,
+      "loss": 0.4843,
+      "step": 1768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692376521116679e-05,
+      "loss": 0.3621,
+      "step": 1769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692197566213314e-05,
+      "loss": 0.4129,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.69201861130995e-05,
+      "loss": 0.3653,
+      "step": 1771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.691839656406586e-05,
+      "loss": 0.5111,
+      "step": 1772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6916607015032216e-05,
+      "loss": 0.4327,
+      "step": 1773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.691481746599857e-05,
+      "loss": 0.5729,
+      "step": 1774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6913027916964926e-05,
+      "loss": 0.4868,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.691123836793128e-05,
+      "loss": 0.3809,
+      "step": 1776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6909448818897636e-05,
+      "loss": 0.5277,
+      "step": 1777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6907659269864e-05,
+      "loss": 0.5259,
+      "step": 1778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6905869720830354e-05,
+      "loss": 0.4554,
+      "step": 1779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.690408017179671e-05,
+      "loss": 0.5519,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.690229062276307e-05,
+      "loss": 0.4177,
+      "step": 1781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6900501073729426e-05,
+      "loss": 0.401,
+      "step": 1782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6898711524695774e-05,
+      "loss": 0.4055,
+      "step": 1783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6896921975662136e-05,
+      "loss": 0.5207,
+      "step": 1784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.689513242662849e-05,
+      "loss": 0.3354,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.689334287759485e-05,
+      "loss": 0.3424,
+      "step": 1786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.68915533285612e-05,
+      "loss": 0.4054,
+      "step": 1787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6889763779527564e-05,
+      "loss": 0.4705,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.688797423049392e-05,
+      "loss": 0.3358,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6886184681460274e-05,
+      "loss": 0.3452,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.688439513242663e-05,
+      "loss": 0.3876,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6882605583392985e-05,
+      "loss": 0.6017,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.688081603435934e-05,
+      "loss": 0.4993,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.68790264853257e-05,
+      "loss": 0.4388,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.687723693629206e-05,
+      "loss": 0.5727,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.687544738725841e-05,
+      "loss": 0.3732,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6873657838224774e-05,
+      "loss": 0.4201,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.687186828919113e-05,
+      "loss": 0.4268,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.687007874015748e-05,
+      "loss": 0.4547,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.686828919112384e-05,
+      "loss": 0.3757,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6866499642090195e-05,
+      "loss": 0.4525,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.686471009305655e-05,
+      "loss": 0.4365,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.686292054402291e-05,
+      "loss": 0.3744,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.686113099498927e-05,
+      "loss": 0.5491,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685934144595562e-05,
+      "loss": 0.2928,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685755189692198e-05,
+      "loss": 0.5323,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685576234788833e-05,
+      "loss": 0.44,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685397279885469e-05,
+      "loss": 0.4754,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.685218324982105e-05,
+      "loss": 0.496,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6850393700787405e-05,
+      "loss": 0.4406,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.684860415175376e-05,
+      "loss": 0.4789,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6846814602720116e-05,
+      "loss": 0.5233,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.684502505368648e-05,
+      "loss": 0.509,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6843235504652826e-05,
+      "loss": 0.5533,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.684144595561918e-05,
+      "loss": 0.4276,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.683965640658554e-05,
+      "loss": 0.6295,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.68378668575519e-05,
+      "loss": 0.4617,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6836077308518254e-05,
+      "loss": 0.3498,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6834287759484615e-05,
+      "loss": 0.4873,
+      "step": 1819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.683249821045097e-05,
+      "loss": 0.4583,
+      "step": 1820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6830708661417326e-05,
+      "loss": 0.419,
+      "step": 1821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682891911238368e-05,
+      "loss": 0.4898,
+      "step": 1822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6827129563350036e-05,
+      "loss": 0.4168,
+      "step": 1823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682534001431639e-05,
+      "loss": 0.362,
+      "step": 1824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6823550465282753e-05,
+      "loss": 0.541,
+      "step": 1825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682176091624911e-05,
+      "loss": 0.4223,
+      "step": 1826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6819971367215464e-05,
+      "loss": 0.4438,
+      "step": 1827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681818181818182e-05,
+      "loss": 0.4573,
+      "step": 1828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681639226914818e-05,
+      "loss": 0.4706,
+      "step": 1829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681460272011453e-05,
+      "loss": 0.457,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681281317108089e-05,
+      "loss": 0.641,
+      "step": 1831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6811023622047247e-05,
+      "loss": 0.6306,
+      "step": 1832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.68092340730136e-05,
+      "loss": 0.3676,
+      "step": 1833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.680744452397996e-05,
+      "loss": 0.4798,
+      "step": 1834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.680565497494632e-05,
+      "loss": 0.4798,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6803865425912674e-05,
+      "loss": 0.4234,
+      "step": 1836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.680207587687903e-05,
+      "loss": 0.4363,
+      "step": 1837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6800286327845384e-05,
+      "loss": 0.4684,
+      "step": 1838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.679849677881174e-05,
+      "loss": 0.4681,
+      "step": 1839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6796707229778095e-05,
+      "loss": 0.3774,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.679491768074446e-05,
+      "loss": 0.599,
+      "step": 1841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.679312813171081e-05,
+      "loss": 0.4966,
+      "step": 1842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.679133858267717e-05,
+      "loss": 0.4219,
+      "step": 1843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.678954903364353e-05,
+      "loss": 0.5172,
+      "step": 1844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6787759484609884e-05,
+      "loss": 0.5085,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.678596993557623e-05,
+      "loss": 0.4338,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6784180386542595e-05,
+      "loss": 0.3619,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.678239083750895e-05,
+      "loss": 0.5385,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6780601288475305e-05,
+      "loss": 0.4743,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677881173944166e-05,
+      "loss": 0.4157,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677702219040802e-05,
+      "loss": 0.506,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677523264137438e-05,
+      "loss": 0.4588,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677344309234073e-05,
+      "loss": 0.3525,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677165354330709e-05,
+      "loss": 0.4096,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.676986399427344e-05,
+      "loss": 0.4524,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.67680744452398e-05,
+      "loss": 0.4111,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.676628489620616e-05,
+      "loss": 0.3493,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6764495347172515e-05,
+      "loss": 0.3758,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.676270579813887e-05,
+      "loss": 0.4693,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.676091624910523e-05,
+      "loss": 0.447,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.675912670007158e-05,
+      "loss": 0.5493,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6757337151037936e-05,
+      "loss": 0.4195,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.67555476020043e-05,
+      "loss": 0.4166,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.675375805297065e-05,
+      "loss": 0.4569,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.675196850393701e-05,
+      "loss": 0.3173,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.675017895490337e-05,
+      "loss": 0.4641,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6748389405869726e-05,
+      "loss": 0.4221,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.674659985683608e-05,
+      "loss": 0.4466,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6744810307802436e-05,
+      "loss": 0.4832,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.674302075876879e-05,
+      "loss": 0.508,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6741231209735147e-05,
+      "loss": 0.4715,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.673944166070151e-05,
+      "loss": 0.3687,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6737652111667864e-05,
+      "loss": 0.3843,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.673586256263422e-05,
+      "loss": 0.3906,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6734073013600574e-05,
+      "loss": 0.3928,
+      "step": 1875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6732283464566936e-05,
+      "loss": 0.3393,
+      "step": 1876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6730493915533284e-05,
+      "loss": 0.6036,
+      "step": 1877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.672870436649964e-05,
+      "loss": 0.5735,
+      "step": 1878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6726914817466e-05,
+      "loss": 0.4678,
+      "step": 1879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.672512526843236e-05,
+      "loss": 0.3323,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.672333571939871e-05,
+      "loss": 0.6017,
+      "step": 1881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6721546170365074e-05,
+      "loss": 0.4293,
+      "step": 1882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.671975662133143e-05,
+      "loss": 0.4068,
+      "step": 1883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6717967072297784e-05,
+      "loss": 0.3394,
+      "step": 1884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.671617752326414e-05,
+      "loss": 0.3714,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6714387974230495e-05,
+      "loss": 0.4503,
+      "step": 1886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.671259842519685e-05,
+      "loss": 0.3976,
+      "step": 1887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.671080887616321e-05,
+      "loss": 0.3974,
+      "step": 1888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670901932712957e-05,
+      "loss": 0.5122,
+      "step": 1889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670722977809592e-05,
+      "loss": 0.5474,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670544022906228e-05,
+      "loss": 0.3983,
+      "step": 1891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670365068002864e-05,
+      "loss": 0.4181,
+      "step": 1892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670186113099499e-05,
+      "loss": 0.4748,
+      "step": 1893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.670007158196135e-05,
+      "loss": 0.4116,
+      "step": 1894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6698282032927705e-05,
+      "loss": 0.4337,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.669649248389406e-05,
+      "loss": 0.3432,
+      "step": 1896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6694702934860415e-05,
+      "loss": 0.4561,
+      "step": 1897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.669291338582678e-05,
+      "loss": 0.4868,
+      "step": 1898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.669112383679313e-05,
+      "loss": 0.4824,
+      "step": 1899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.668933428775948e-05,
+      "loss": 0.4707,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.668754473872584e-05,
+      "loss": 0.4323,
+      "step": 1901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.66857551896922e-05,
+      "loss": 0.3194,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.668396564065855e-05,
+      "loss": 0.5659,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6682176091624915e-05,
+      "loss": 0.4768,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.668038654259127e-05,
+      "loss": 0.4044,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6678596993557626e-05,
+      "loss": 0.4126,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.667680744452399e-05,
+      "loss": 0.3783,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6675017895490336e-05,
+      "loss": 0.5372,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.667322834645669e-05,
+      "loss": 0.4578,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.667143879742305e-05,
+      "loss": 0.5206,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.666964924838941e-05,
+      "loss": 0.4888,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6667859699355764e-05,
+      "loss": 0.5268,
+      "step": 1912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.666607015032212e-05,
+      "loss": 0.3664,
+      "step": 1913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.666428060128848e-05,
+      "loss": 0.5097,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6662491052254836e-05,
+      "loss": 0.3751,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.666070150322119e-05,
+      "loss": 0.3972,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6658911954187546e-05,
+      "loss": 0.4782,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.66571224051539e-05,
+      "loss": 0.5891,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.665533285612026e-05,
+      "loss": 0.3345,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.665354330708662e-05,
+      "loss": 0.5161,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6651753758052974e-05,
+      "loss": 0.5349,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664996420901933e-05,
+      "loss": 0.5231,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664817465998569e-05,
+      "loss": 0.4486,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664638511095204e-05,
+      "loss": 0.5055,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6644595561918395e-05,
+      "loss": 0.5403,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664280601288476e-05,
+      "loss": 0.5437,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.664101646385111e-05,
+      "loss": 0.449,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.663922691481747e-05,
+      "loss": 0.4466,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.663743736578383e-05,
+      "loss": 0.4298,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6635647816750184e-05,
+      "loss": 0.4622,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.663385826771654e-05,
+      "loss": 0.4053,
+      "step": 1931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6632068718682895e-05,
+      "loss": 0.5192,
+      "step": 1932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.663027916964925e-05,
+      "loss": 0.569,
+      "step": 1933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6628489620615605e-05,
+      "loss": 0.4636,
+      "step": 1934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662670007158196e-05,
+      "loss": 0.4219,
+      "step": 1935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662491052254832e-05,
+      "loss": 0.4774,
+      "step": 1936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662312097351468e-05,
+      "loss": 0.4681,
+      "step": 1937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662133142448103e-05,
+      "loss": 0.5571,
+      "step": 1938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6619541875447395e-05,
+      "loss": 0.4313,
+      "step": 1939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.661775232641374e-05,
+      "loss": 0.5028,
+      "step": 1940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.66159627773801e-05,
+      "loss": 0.5671,
+      "step": 1941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.661417322834646e-05,
+      "loss": 0.4253,
+      "step": 1942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6612383679312815e-05,
+      "loss": 0.552,
+      "step": 1943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.661059413027917e-05,
+      "loss": 0.4771,
+      "step": 1944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.660880458124553e-05,
+      "loss": 0.3204,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.660701503221189e-05,
+      "loss": 0.3408,
+      "step": 1946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6605225483178236e-05,
+      "loss": 0.5764,
+      "step": 1947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.66034359341446e-05,
+      "loss": 0.4577,
+      "step": 1948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.660164638511095e-05,
+      "loss": 0.4442,
+      "step": 1949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.659985683607731e-05,
+      "loss": 0.5518,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.659806728704367e-05,
+      "loss": 0.3916,
+      "step": 1951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6596277738010026e-05,
+      "loss": 0.5751,
+      "step": 1952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.659448818897638e-05,
+      "loss": 0.5137,
+      "step": 1953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6592698639942736e-05,
+      "loss": 0.4715,
+      "step": 1954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.659090909090909e-05,
+      "loss": 0.4662,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6589119541875446e-05,
+      "loss": 0.451,
+      "step": 1956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.658732999284181e-05,
+      "loss": 0.4078,
+      "step": 1957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6585540443808163e-05,
+      "loss": 0.4336,
+      "step": 1958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.658375089477452e-05,
+      "loss": 0.5329,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6581961345740874e-05,
+      "loss": 0.4384,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6580171796707236e-05,
+      "loss": 0.3971,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.657838224767359e-05,
+      "loss": 0.4152,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.657659269863994e-05,
+      "loss": 0.4048,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.65748031496063e-05,
+      "loss": 0.3347,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.657301360057266e-05,
+      "loss": 0.4016,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.657122405153901e-05,
+      "loss": 0.4502,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6569434502505374e-05,
+      "loss": 0.4309,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.656764495347173e-05,
+      "loss": 0.4321,
+      "step": 1968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6565855404438084e-05,
+      "loss": 0.6234,
+      "step": 1969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6564065855404446e-05,
+      "loss": 0.4659,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6562276306370795e-05,
+      "loss": 0.4536,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.656048675733715e-05,
+      "loss": 0.3099,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.655869720830351e-05,
+      "loss": 0.4562,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.655690765926987e-05,
+      "loss": 0.3775,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.655511811023622e-05,
+      "loss": 0.4414,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.655332856120258e-05,
+      "loss": 0.3706,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.655153901216894e-05,
+      "loss": 0.4981,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6549749463135294e-05,
+      "loss": 0.4064,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.654795991410165e-05,
+      "loss": 0.4441,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6546170365068005e-05,
+      "loss": 0.3786,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.654438081603436e-05,
+      "loss": 0.5374,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6542591267000715e-05,
+      "loss": 0.4843,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.654080171796708e-05,
+      "loss": 0.4526,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.653901216893343e-05,
+      "loss": 0.6648,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.653722261989979e-05,
+      "loss": 0.3462,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.653543307086614e-05,
+      "loss": 0.5834,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.65336435218325e-05,
+      "loss": 0.3565,
+      "step": 1987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.653185397279885e-05,
+      "loss": 0.4574,
+      "step": 1988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6530064423765215e-05,
+      "loss": 0.4483,
+      "step": 1989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.652827487473157e-05,
+      "loss": 0.5197,
+      "step": 1990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6526485325697926e-05,
+      "loss": 0.4346,
+      "step": 1991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.652469577666429e-05,
+      "loss": 0.4329,
+      "step": 1992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.652290622763064e-05,
+      "loss": 0.4939,
+      "step": 1993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.652111667859699e-05,
+      "loss": 0.5883,
+      "step": 1994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.651932712956335e-05,
+      "loss": 0.5301,
+      "step": 1995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.651753758052971e-05,
+      "loss": 0.3613,
+      "step": 1996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6515748031496063e-05,
+      "loss": 0.3999,
+      "step": 1997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.651395848246242e-05,
+      "loss": 0.4513,
+      "step": 1998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.651216893342878e-05,
+      "loss": 0.4135,
+      "step": 1999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6510379384395136e-05,
+      "loss": 0.4479,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.650858983536149e-05,
+      "loss": 0.574,
+      "step": 2001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6506800286327846e-05,
+      "loss": 0.367,
+      "step": 2002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.65050107372942e-05,
+      "loss": 0.3436,
+      "step": 2003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6503221188260557e-05,
+      "loss": 0.3696,
+      "step": 2004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.650143163922692e-05,
+      "loss": 0.4648,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6499642090193274e-05,
+      "loss": 0.38,
+      "step": 2006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.649785254115963e-05,
+      "loss": 0.4373,
+      "step": 2007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.649606299212599e-05,
+      "loss": 0.3892,
+      "step": 2008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6494273443092346e-05,
+      "loss": 0.4413,
+      "step": 2009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6492483894058695e-05,
+      "loss": 0.5,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6490694345025056e-05,
+      "loss": 0.4993,
+      "step": 2011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.648890479599141e-05,
+      "loss": 0.4404,
+      "step": 2012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.648711524695777e-05,
+      "loss": 0.4226,
+      "step": 2013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.648532569792413e-05,
+      "loss": 0.4596,
+      "step": 2014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6483536148890484e-05,
+      "loss": 0.5288,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.648174659985684e-05,
+      "loss": 0.4294,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6479957050823194e-05,
+      "loss": 0.4431,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.647816750178955e-05,
+      "loss": 0.5144,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6476377952755905e-05,
+      "loss": 0.4116,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.647458840372227e-05,
+      "loss": 0.5141,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.647279885468862e-05,
+      "loss": 0.4308,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.647100930565498e-05,
+      "loss": 0.2838,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.646921975662133e-05,
+      "loss": 0.5549,
+      "step": 2023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6467430207587694e-05,
+      "loss": 0.462,
+      "step": 2024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.646564065855405e-05,
+      "loss": 0.4995,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.64638511095204e-05,
+      "loss": 0.4406,
+      "step": 2026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.646206156048676e-05,
+      "loss": 0.4836,
+      "step": 2027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6460272011453115e-05,
+      "loss": 0.408,
+      "step": 2028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.645848246241947e-05,
+      "loss": 0.5539,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.645669291338583e-05,
+      "loss": 0.513,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.645490336435219e-05,
+      "loss": 0.463,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.645311381531854e-05,
+      "loss": 0.409,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.64513242662849e-05,
+      "loss": 0.3539,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.644953471725125e-05,
+      "loss": 0.4164,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.644774516821761e-05,
+      "loss": 0.3913,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.644595561918397e-05,
+      "loss": 0.3442,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6444166070150325e-05,
+      "loss": 0.4821,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.644237652111668e-05,
+      "loss": 0.4969,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6440586972083036e-05,
+      "loss": 0.5464,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.64387974230494e-05,
+      "loss": 0.5041,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6437007874015746e-05,
+      "loss": 0.3204,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.643521832498211e-05,
+      "loss": 0.4732,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.643342877594846e-05,
+      "loss": 0.4719,
+      "step": 2043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.643163922691482e-05,
+      "loss": 0.3984,
+      "step": 2044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6429849677881174e-05,
+      "loss": 0.4709,
+      "step": 2045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6428060128847536e-05,
+      "loss": 0.6398,
+      "step": 2046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.642627057981389e-05,
+      "loss": 0.5426,
+      "step": 2047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6424481030780246e-05,
+      "loss": 0.4855,
+      "step": 2048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.64226914817466e-05,
+      "loss": 0.464,
+      "step": 2049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6420901932712956e-05,
+      "loss": 0.4432,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.641911238367931e-05,
+      "loss": 0.4962,
+      "step": 2051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6417322834645674e-05,
+      "loss": 0.4826,
+      "step": 2052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.641553328561203e-05,
+      "loss": 0.5075,
+      "step": 2053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6413743736578384e-05,
+      "loss": 0.5408,
+      "step": 2054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6411954187544746e-05,
+      "loss": 0.435,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.64101646385111e-05,
+      "loss": 0.4263,
+      "step": 2056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.640837508947745e-05,
+      "loss": 0.439,
+      "step": 2057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.640658554044381e-05,
+      "loss": 0.5001,
+      "step": 2058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.640479599141017e-05,
+      "loss": 0.4712,
+      "step": 2059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.640300644237652e-05,
+      "loss": 0.4072,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.640121689334288e-05,
+      "loss": 0.4323,
+      "step": 2061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.639942734430924e-05,
+      "loss": 0.4889,
+      "step": 2062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6397637795275594e-05,
+      "loss": 0.4558,
+      "step": 2063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.639584824624195e-05,
+      "loss": 0.4873,
+      "step": 2064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6394058697208305e-05,
+      "loss": 0.4005,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.639226914817466e-05,
+      "loss": 0.4855,
+      "step": 2066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6390479599141015e-05,
+      "loss": 0.3997,
+      "step": 2067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.638869005010738e-05,
+      "loss": 0.4063,
+      "step": 2068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.638690050107373e-05,
+      "loss": 0.4066,
+      "step": 2069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.638511095204009e-05,
+      "loss": 0.3882,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.638332140300645e-05,
+      "loss": 0.4329,
+      "step": 2071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.63815318539728e-05,
+      "loss": 0.5095,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.637974230493915e-05,
+      "loss": 0.3495,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6377952755905515e-05,
+      "loss": 0.2819,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.637616320687187e-05,
+      "loss": 0.5792,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6374373657838225e-05,
+      "loss": 0.4799,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.637258410880459e-05,
+      "loss": 0.5302,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.637079455977094e-05,
+      "loss": 0.4835,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.63690050107373e-05,
+      "loss": 0.4145,
+      "step": 2079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636721546170365e-05,
+      "loss": 0.6175,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636542591267001e-05,
+      "loss": 0.4743,
+      "step": 2081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636363636363636e-05,
+      "loss": 0.433,
+      "step": 2082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636184681460272e-05,
+      "loss": 0.3656,
+      "step": 2083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636005726556908e-05,
+      "loss": 0.3819,
+      "step": 2084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6358267716535436e-05,
+      "loss": 0.5353,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.635647816750179e-05,
+      "loss": 0.3474,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.635468861846815e-05,
+      "loss": 0.3383,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.63528990694345e-05,
+      "loss": 0.3116,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6351109520400856e-05,
+      "loss": 0.4779,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.634931997136722e-05,
+      "loss": 0.5248,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6347530422333574e-05,
+      "loss": 0.4181,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.634574087329993e-05,
+      "loss": 0.3921,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.634395132426629e-05,
+      "loss": 0.3766,
+      "step": 2093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6342161775232646e-05,
+      "loss": 0.2181,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6340372226199e-05,
+      "loss": 0.4117,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6338582677165356e-05,
+      "loss": 0.5055,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.633679312813171e-05,
+      "loss": 0.4885,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.633500357909807e-05,
+      "loss": 0.4872,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.633321403006443e-05,
+      "loss": 0.4759,
+      "step": 2099
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6331424481030784e-05,
+      "loss": 0.4185,
+      "step": 2100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.632963493199714e-05,
+      "loss": 0.3455,
+      "step": 2101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6327845382963494e-05,
+      "loss": 0.451,
+      "step": 2102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6326055833929856e-05,
+      "loss": 0.3212,
+      "step": 2103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6324266284896205e-05,
+      "loss": 0.4123,
+      "step": 2104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6322476735862567e-05,
+      "loss": 0.4497,
+      "step": 2105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.632068718682892e-05,
+      "loss": 0.4697,
+      "step": 2106
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.631889763779528e-05,
+      "loss": 0.3279,
+      "step": 2107
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.631710808876163e-05,
+      "loss": 0.386,
+      "step": 2108
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6315318539727994e-05,
+      "loss": 0.4244,
+      "step": 2109
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.631352899069435e-05,
+      "loss": 0.3302,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6311739441660705e-05,
+      "loss": 0.4124,
+      "step": 2111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.630994989262706e-05,
+      "loss": 0.46,
+      "step": 2112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6308160343593415e-05,
+      "loss": 0.4412,
+      "step": 2113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.630637079455977e-05,
+      "loss": 0.4103,
+      "step": 2114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.630458124552613e-05,
+      "loss": 0.4066,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.630279169649249e-05,
+      "loss": 0.5225,
+      "step": 2116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.630100214745884e-05,
+      "loss": 0.3567,
+      "step": 2117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6299212598425204e-05,
+      "loss": 0.3966,
+      "step": 2118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.629742304939155e-05,
+      "loss": 0.3284,
+      "step": 2119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.629563350035791e-05,
+      "loss": 0.5138,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.629384395132427e-05,
+      "loss": 0.4335,
+      "step": 2121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6292054402290625e-05,
+      "loss": 0.4414,
+      "step": 2122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.629026485325698e-05,
+      "loss": 0.3926,
+      "step": 2123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6288475304223336e-05,
+      "loss": 0.4249,
+      "step": 2124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.62866857551897e-05,
+      "loss": 0.4455,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.628489620615605e-05,
+      "loss": 0.3899,
+      "step": 2126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.628310665712241e-05,
+      "loss": 0.392,
+      "step": 2127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.628131710808876e-05,
+      "loss": 0.4313,
+      "step": 2128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.627952755905512e-05,
+      "loss": 0.3937,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6277738010021474e-05,
+      "loss": 0.6101,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6275948460987835e-05,
+      "loss": 0.4722,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.627415891195419e-05,
+      "loss": 0.3803,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6272369362920546e-05,
+      "loss": 0.4984,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.627057981388691e-05,
+      "loss": 0.4852,
+      "step": 2134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6268790264853256e-05,
+      "loss": 0.3619,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.626700071581961e-05,
+      "loss": 0.5283,
+      "step": 2136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6265211166785973e-05,
+      "loss": 0.3814,
+      "step": 2137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.626342161775233e-05,
+      "loss": 0.4261,
+      "step": 2138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6261632068718684e-05,
+      "loss": 0.4261,
+      "step": 2139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6259842519685046e-05,
+      "loss": 0.3759,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.62580529706514e-05,
+      "loss": 0.3373,
+      "step": 2141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6256263421617756e-05,
+      "loss": 0.3635,
+      "step": 2142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.625447387258411e-05,
+      "loss": 0.4342,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6252684323550467e-05,
+      "loss": 0.4193,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.625089477451682e-05,
+      "loss": 0.2752,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624910522548318e-05,
+      "loss": 0.4789,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624731567644954e-05,
+      "loss": 0.423,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6245526127415894e-05,
+      "loss": 0.4844,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624373657838225e-05,
+      "loss": 0.4593,
+      "step": 2149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624194702934861e-05,
+      "loss": 0.3382,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.624015748031496e-05,
+      "loss": 0.4194,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6238367931281315e-05,
+      "loss": 0.3937,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.623657838224768e-05,
+      "loss": 0.474,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.623478883321403e-05,
+      "loss": 0.4391,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.623299928418039e-05,
+      "loss": 0.5247,
+      "step": 2155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.623120973514675e-05,
+      "loss": 0.5967,
+      "step": 2156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6229420186113104e-05,
+      "loss": 0.247,
+      "step": 2157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.622763063707946e-05,
+      "loss": 0.5156,
+      "step": 2158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6225841088045815e-05,
+      "loss": 0.4224,
+      "step": 2159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.622405153901217e-05,
+      "loss": 0.5323,
+      "step": 2160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6222261989978525e-05,
+      "loss": 0.3973,
+      "step": 2161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.622047244094489e-05,
+      "loss": 0.343,
+      "step": 2162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.621868289191124e-05,
+      "loss": 0.4943,
+      "step": 2163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.62168933428776e-05,
+      "loss": 0.3943,
+      "step": 2164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.621510379384395e-05,
+      "loss": 0.3474,
+      "step": 2165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.621331424481031e-05,
+      "loss": 0.4245,
+      "step": 2166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.621152469577666e-05,
+      "loss": 0.4483,
+      "step": 2167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6209735146743025e-05,
+      "loss": 0.3911,
+      "step": 2168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.620794559770938e-05,
+      "loss": 0.4167,
+      "step": 2169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6206156048675735e-05,
+      "loss": 0.4619,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.620436649964209e-05,
+      "loss": 0.4006,
+      "step": 2171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.620257695060845e-05,
+      "loss": 0.4614,
+      "step": 2172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.620078740157481e-05,
+      "loss": 0.3205,
+      "step": 2173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6198997852541156e-05,
+      "loss": 0.5373,
+      "step": 2174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.619720830350752e-05,
+      "loss": 0.48,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.619541875447387e-05,
+      "loss": 0.3697,
+      "step": 2176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.619362920544023e-05,
+      "loss": 0.3554,
+      "step": 2177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.619183965640659e-05,
+      "loss": 0.4277,
+      "step": 2178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6190050107372946e-05,
+      "loss": 0.5493,
+      "step": 2179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.61882605583393e-05,
+      "loss": 0.5007,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.618647100930566e-05,
+      "loss": 0.413,
+      "step": 2181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.618468146027201e-05,
+      "loss": 0.4595,
+      "step": 2182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6182891911238367e-05,
+      "loss": 0.5295,
+      "step": 2183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.618110236220473e-05,
+      "loss": 0.4885,
+      "step": 2184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6179312813171084e-05,
+      "loss": 0.4706,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.617752326413744e-05,
+      "loss": 0.4278,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6175733715103794e-05,
+      "loss": 0.3964,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6173944166070156e-05,
+      "loss": 0.3976,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.617215461703651e-05,
+      "loss": 0.4013,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6170365068002866e-05,
+      "loss": 0.4456,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.616857551896922e-05,
+      "loss": 0.4168,
+      "step": 2191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.616678596993558e-05,
+      "loss": 0.3907,
+      "step": 2192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.616499642090193e-05,
+      "loss": 0.4888,
+      "step": 2193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6163206871868294e-05,
+      "loss": 0.448,
+      "step": 2194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.616141732283465e-05,
+      "loss": 0.4035,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6159627773801004e-05,
+      "loss": 0.4257,
+      "step": 2196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6157838224767366e-05,
+      "loss": 0.5166,
+      "step": 2197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6156048675733715e-05,
+      "loss": 0.4574,
+      "step": 2198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.615425912670007e-05,
+      "loss": 0.4187,
+      "step": 2199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.615246957766643e-05,
+      "loss": 0.4793,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.615068002863279e-05,
+      "loss": 0.3962,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.614889047959914e-05,
+      "loss": 0.4154,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6147100930565504e-05,
+      "loss": 0.5266,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.614531138153186e-05,
+      "loss": 0.3989,
+      "step": 2204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.614352183249821e-05,
+      "loss": 0.394,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.614173228346457e-05,
+      "loss": 0.35,
+      "step": 2206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6139942734430925e-05,
+      "loss": 0.4504,
+      "step": 2207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.613815318539728e-05,
+      "loss": 0.4244,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6136363636363635e-05,
+      "loss": 0.402,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.613457408733e-05,
+      "loss": 0.43,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.613278453829635e-05,
+      "loss": 0.374,
+      "step": 2211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.613099498926271e-05,
+      "loss": 0.3896,
+      "step": 2212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.612920544022906e-05,
+      "loss": 0.4654,
+      "step": 2213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.612741589119542e-05,
+      "loss": 0.3679,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.612562634216177e-05,
+      "loss": 0.4598,
+      "step": 2215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6123836793128135e-05,
+      "loss": 0.3397,
+      "step": 2216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.612204724409449e-05,
+      "loss": 0.459,
+      "step": 2217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6120257695060846e-05,
+      "loss": 0.4038,
+      "step": 2218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611846814602721e-05,
+      "loss": 0.3819,
+      "step": 2219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611667859699356e-05,
+      "loss": 0.3934,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611488904795991e-05,
+      "loss": 0.3803,
+      "step": 2221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611309949892627e-05,
+      "loss": 0.4869,
+      "step": 2222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611130994989263e-05,
+      "loss": 0.4844,
+      "step": 2223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6109520400858984e-05,
+      "loss": 0.3971,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6107730851825346e-05,
+      "loss": 0.4847,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.61059413027917e-05,
+      "loss": 0.3751,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6104151753758056e-05,
+      "loss": 0.4398,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.610236220472441e-05,
+      "loss": 0.3907,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6100572655690766e-05,
+      "loss": 0.3971,
+      "step": 2229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.609878310665712e-05,
+      "loss": 0.3823,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6096993557623484e-05,
+      "loss": 0.3678,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.609520400858984e-05,
+      "loss": 0.3449,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6093414459556194e-05,
+      "loss": 0.4633,
+      "step": 2233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.609162491052255e-05,
+      "loss": 0.4197,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.608983536148891e-05,
+      "loss": 0.4561,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6088045812455266e-05,
+      "loss": 0.4116,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6086256263421615e-05,
+      "loss": 0.4416,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.608446671438798e-05,
+      "loss": 0.4359,
+      "step": 2238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.608267716535433e-05,
+      "loss": 0.4057,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.608088761632069e-05,
+      "loss": 0.3905,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.607909806728705e-05,
+      "loss": 0.3586,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6077308518253404e-05,
+      "loss": 0.3829,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.607551896921976e-05,
+      "loss": 0.3292,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.607372942018612e-05,
+      "loss": 0.3975,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.607193987115247e-05,
+      "loss": 0.3803,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6070150322118825e-05,
+      "loss": 0.4755,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.606836077308519e-05,
+      "loss": 0.3836,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.606657122405154e-05,
+      "loss": 0.5108,
+      "step": 2248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.60647816750179e-05,
+      "loss": 0.4783,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.606299212598425e-05,
+      "loss": 0.4063,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6061202576950614e-05,
+      "loss": 0.3174,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.605941302791696e-05,
+      "loss": 0.3969,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6057623478883325e-05,
+      "loss": 0.4313,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.605583392984968e-05,
+      "loss": 0.4301,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6054044380816035e-05,
+      "loss": 0.3878,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.605225483178239e-05,
+      "loss": 0.3694,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.605046528274875e-05,
+      "loss": 0.5871,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604867573371511e-05,
+      "loss": 0.3667,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604688618468146e-05,
+      "loss": 0.419,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604509663564782e-05,
+      "loss": 0.4376,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604330708661417e-05,
+      "loss": 0.4412,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.604151753758053e-05,
+      "loss": 0.4093,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.603972798854689e-05,
+      "loss": 0.3148,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6037938439513246e-05,
+      "loss": 0.4245,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.60361488904796e-05,
+      "loss": 0.556,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.603435934144596e-05,
+      "loss": 0.4825,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.603256979241232e-05,
+      "loss": 0.424,
+      "step": 2267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6030780243378666e-05,
+      "loss": 0.4991,
+      "step": 2268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.602899069434503e-05,
+      "loss": 0.4318,
+      "step": 2269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6027201145311383e-05,
+      "loss": 0.6386,
+      "step": 2270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.602541159627774e-05,
+      "loss": 0.4914,
+      "step": 2271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6023622047244094e-05,
+      "loss": 0.4152,
+      "step": 2272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6021832498210456e-05,
+      "loss": 0.5071,
+      "step": 2273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.602004294917681e-05,
+      "loss": 0.4001,
+      "step": 2274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6018253400143166e-05,
+      "loss": 0.3503,
+      "step": 2275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.601646385110952e-05,
+      "loss": 0.3588,
+      "step": 2276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.601467430207588e-05,
+      "loss": 0.3346,
+      "step": 2277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.601288475304223e-05,
+      "loss": 0.3618,
+      "step": 2278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6011095204008594e-05,
+      "loss": 0.3226,
+      "step": 2279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.600930565497495e-05,
+      "loss": 0.5224,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6007516105941304e-05,
+      "loss": 0.4004,
+      "step": 2281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6005726556907666e-05,
+      "loss": 0.2867,
+      "step": 2282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.600393700787402e-05,
+      "loss": 0.3499,
+      "step": 2283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.600214745884037e-05,
+      "loss": 0.6337,
+      "step": 2284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.600035790980673e-05,
+      "loss": 0.4048,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.599856836077309e-05,
+      "loss": 0.3508,
+      "step": 2286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.599677881173944e-05,
+      "loss": 0.5047,
+      "step": 2287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5994989262705804e-05,
+      "loss": 0.5195,
+      "step": 2288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.599319971367216e-05,
+      "loss": 0.4288,
+      "step": 2289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5991410164638514e-05,
+      "loss": 0.5256,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.598962061560487e-05,
+      "loss": 0.4676,
+      "step": 2291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5987831066571225e-05,
+      "loss": 0.5409,
+      "step": 2292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.598604151753758e-05,
+      "loss": 0.3398,
+      "step": 2293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5984251968503935e-05,
+      "loss": 0.4567,
+      "step": 2294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.59824624194703e-05,
+      "loss": 0.4422,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.598067287043665e-05,
+      "loss": 0.4264,
+      "step": 2296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.597888332140301e-05,
+      "loss": 0.4817,
+      "step": 2297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.597709377236937e-05,
+      "loss": 0.4042,
+      "step": 2298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.597530422333572e-05,
+      "loss": 0.3681,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.597351467430207e-05,
+      "loss": 0.3884,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5971725125268435e-05,
+      "loss": 0.4932,
+      "step": 2301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.596993557623479e-05,
+      "loss": 0.4337,
+      "step": 2302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5968146027201146e-05,
+      "loss": 0.4492,
+      "step": 2303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.596635647816751e-05,
+      "loss": 0.5461,
+      "step": 2304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.596456692913386e-05,
+      "loss": 0.3335,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.596277738010022e-05,
+      "loss": 0.4399,
+      "step": 2306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.596098783106657e-05,
+      "loss": 0.4128,
+      "step": 2307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.595919828203293e-05,
+      "loss": 0.3453,
+      "step": 2308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5957408732999283e-05,
+      "loss": 0.3445,
+      "step": 2309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5955619183965645e-05,
+      "loss": 0.4545,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5953829634932e-05,
+      "loss": 0.4644,
+      "step": 2311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5952040085898356e-05,
+      "loss": 0.43,
+      "step": 2312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.595025053686471e-05,
+      "loss": 0.2699,
+      "step": 2313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.594846098783107e-05,
+      "loss": 0.4561,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.594667143879742e-05,
+      "loss": 0.4164,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.594488188976378e-05,
+      "loss": 0.4771,
+      "step": 2316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.594309234073014e-05,
+      "loss": 0.4289,
+      "step": 2317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5941302791696494e-05,
+      "loss": 0.3187,
+      "step": 2318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.593951324266285e-05,
+      "loss": 0.408,
+      "step": 2319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.593772369362921e-05,
+      "loss": 0.424,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5935934144595566e-05,
+      "loss": 0.4191,
+      "step": 2321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.593414459556192e-05,
+      "loss": 0.3804,
+      "step": 2322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5932355046528276e-05,
+      "loss": 0.5384,
+      "step": 2323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.593056549749463e-05,
+      "loss": 0.5783,
+      "step": 2324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592877594846099e-05,
+      "loss": 0.3585,
+      "step": 2325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592698639942735e-05,
+      "loss": 0.4538,
+      "step": 2326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5925196850393704e-05,
+      "loss": 0.4594,
+      "step": 2327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592340730136006e-05,
+      "loss": 0.4149,
+      "step": 2328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592161775232642e-05,
+      "loss": 0.4274,
+      "step": 2329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5919828203292776e-05,
+      "loss": 0.3896,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5918038654259125e-05,
+      "loss": 0.5689,
+      "step": 2331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.591624910522549e-05,
+      "loss": 0.4792,
+      "step": 2332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.591445955619184e-05,
+      "loss": 0.3702,
+      "step": 2333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.59126700071582e-05,
+      "loss": 0.3115,
+      "step": 2334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.591088045812455e-05,
+      "loss": 0.3653,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5909090909090914e-05,
+      "loss": 0.4323,
+      "step": 2336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.590730136005727e-05,
+      "loss": 0.3323,
+      "step": 2337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5905511811023625e-05,
+      "loss": 0.2788,
+      "step": 2338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.590372226198998e-05,
+      "loss": 0.4791,
+      "step": 2339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5901932712956335e-05,
+      "loss": 0.3761,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.590014316392269e-05,
+      "loss": 0.5279,
+      "step": 2341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.589835361488905e-05,
+      "loss": 0.404,
+      "step": 2342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.589656406585541e-05,
+      "loss": 0.401,
+      "step": 2343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.589477451682176e-05,
+      "loss": 0.4249,
+      "step": 2344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5892984967788125e-05,
+      "loss": 0.4515,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.589119541875447e-05,
+      "loss": 0.4805,
+      "step": 2346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.588940586972083e-05,
+      "loss": 0.3561,
+      "step": 2347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.588761632068719e-05,
+      "loss": 0.432,
+      "step": 2348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5885826771653545e-05,
+      "loss": 0.4418,
+      "step": 2349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.58840372226199e-05,
+      "loss": 0.5067,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.588224767358626e-05,
+      "loss": 0.4106,
+      "step": 2351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.588045812455262e-05,
+      "loss": 0.3486,
+      "step": 2352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.587866857551897e-05,
+      "loss": 0.4364,
+      "step": 2353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.587687902648533e-05,
+      "loss": 0.4099,
+      "step": 2354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.587508947745168e-05,
+      "loss": 0.3149,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.587329992841804e-05,
+      "loss": 0.4615,
+      "step": 2356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5871510379384394e-05,
+      "loss": 0.5267,
+      "step": 2357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5869720830350756e-05,
+      "loss": 0.3572,
+      "step": 2358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.586793128131711e-05,
+      "loss": 0.4272,
+      "step": 2359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5866141732283466e-05,
+      "loss": 0.5594,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.586435218324983e-05,
+      "loss": 0.4805,
+      "step": 2361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5862562634216176e-05,
+      "loss": 0.5082,
+      "step": 2362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.586077308518253e-05,
+      "loss": 0.3911,
+      "step": 2363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5858983536148894e-05,
+      "loss": 0.3711,
+      "step": 2364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.585719398711525e-05,
+      "loss": 0.5012,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5855404438081604e-05,
+      "loss": 0.3906,
+      "step": 2366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5853614889047966e-05,
+      "loss": 0.3981,
+      "step": 2367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.585182534001432e-05,
+      "loss": 0.5861,
+      "step": 2368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5850035790980676e-05,
+      "loss": 0.4169,
+      "step": 2369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.584824624194703e-05,
+      "loss": 0.4862,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.584645669291339e-05,
+      "loss": 0.4618,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.584466714387974e-05,
+      "loss": 0.2979,
+      "step": 2372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5842877594846104e-05,
+      "loss": 0.3861,
+      "step": 2373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.584108804581246e-05,
+      "loss": 0.4986,
+      "step": 2374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5839298496778814e-05,
+      "loss": 0.2792,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.583750894774517e-05,
+      "loss": 0.3287,
+      "step": 2376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.583571939871153e-05,
+      "loss": 0.4,
+      "step": 2377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.583392984967788e-05,
+      "loss": 0.2977,
+      "step": 2378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.583214030064424e-05,
+      "loss": 0.4028,
+      "step": 2379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.58303507516106e-05,
+      "loss": 0.4435,
+      "step": 2380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582856120257695e-05,
+      "loss": 0.4477,
+      "step": 2381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582677165354331e-05,
+      "loss": 0.4752,
+      "step": 2382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582498210450967e-05,
+      "loss": 0.2808,
+      "step": 2383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5823192555476025e-05,
+      "loss": 0.4783,
+      "step": 2384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582140300644237e-05,
+      "loss": 0.4824,
+      "step": 2385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5819613457408735e-05,
+      "loss": 0.3817,
+      "step": 2386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.581782390837509e-05,
+      "loss": 0.3949,
+      "step": 2387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5816034359341445e-05,
+      "loss": 0.5081,
+      "step": 2388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.581424481030781e-05,
+      "loss": 0.3696,
+      "step": 2389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.581245526127416e-05,
+      "loss": 0.3261,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.581066571224052e-05,
+      "loss": 0.4621,
+      "step": 2391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.580887616320688e-05,
+      "loss": 0.3205,
+      "step": 2392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.580708661417323e-05,
+      "loss": 0.3787,
+      "step": 2393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.580529706513958e-05,
+      "loss": 0.4739,
+      "step": 2394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5803507516105945e-05,
+      "loss": 0.456,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.58017179670723e-05,
+      "loss": 0.5077,
+      "step": 2396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5799928418038656e-05,
+      "loss": 0.4523,
+      "step": 2397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.579813886900501e-05,
+      "loss": 0.3388,
+      "step": 2398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.579634931997137e-05,
+      "loss": 0.4647,
+      "step": 2399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.579455977093773e-05,
+      "loss": 0.3879,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.579277022190408e-05,
+      "loss": 0.4284,
+      "step": 2401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.579098067287044e-05,
+      "loss": 0.4814,
+      "step": 2402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5789191123836794e-05,
+      "loss": 0.4566,
+      "step": 2403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.578740157480315e-05,
+      "loss": 0.428,
+      "step": 2404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.578561202576951e-05,
+      "loss": 0.3408,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5783822476735866e-05,
+      "loss": 0.3837,
+      "step": 2406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.578203292770222e-05,
+      "loss": 0.5349,
+      "step": 2407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.578024337866858e-05,
+      "loss": 0.3547,
+      "step": 2408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.577845382963493e-05,
+      "loss": 0.4696,
+      "step": 2409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.577666428060129e-05,
+      "loss": 0.4227,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.577487473156765e-05,
+      "loss": 0.3374,
+      "step": 2411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5773085182534004e-05,
+      "loss": 0.3824,
+      "step": 2412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.577129563350036e-05,
+      "loss": 0.4231,
+      "step": 2413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.576950608446672e-05,
+      "loss": 0.3683,
+      "step": 2414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5767716535433076e-05,
+      "loss": 0.4174,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.576592698639943e-05,
+      "loss": 0.316,
+      "step": 2416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5764137437365787e-05,
+      "loss": 0.4108,
+      "step": 2417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.576234788833214e-05,
+      "loss": 0.4362,
+      "step": 2418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.57605583392985e-05,
+      "loss": 0.3925,
+      "step": 2419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.575876879026485e-05,
+      "loss": 0.3505,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5756979241231214e-05,
+      "loss": 0.3857,
+      "step": 2421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.575518969219757e-05,
+      "loss": 0.3784,
+      "step": 2422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5753400143163925e-05,
+      "loss": 0.3413,
+      "step": 2423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.575161059413028e-05,
+      "loss": 0.361,
+      "step": 2424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5749821045096635e-05,
+      "loss": 0.417,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.574803149606299e-05,
+      "loss": 0.514,
+      "step": 2426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.574624194702935e-05,
+      "loss": 0.2708,
+      "step": 2427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.574445239799571e-05,
+      "loss": 0.4567,
+      "step": 2428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.574266284896206e-05,
+      "loss": 0.4985,
+      "step": 2429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5740873299928424e-05,
+      "loss": 0.4029,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.573908375089478e-05,
+      "loss": 0.341,
+      "step": 2431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.573729420186113e-05,
+      "loss": 0.3971,
+      "step": 2432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.573550465282749e-05,
+      "loss": 0.3316,
+      "step": 2433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5733715103793845e-05,
+      "loss": 0.3673,
+      "step": 2434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.57319255547602e-05,
+      "loss": 0.3445,
+      "step": 2435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.573013600572656e-05,
+      "loss": 0.4358,
+      "step": 2436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572834645669292e-05,
+      "loss": 0.2593,
+      "step": 2437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572655690765927e-05,
+      "loss": 0.4307,
+      "step": 2438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572476735862563e-05,
+      "loss": 0.5008,
+      "step": 2439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572297780959198e-05,
+      "loss": 0.4049,
+      "step": 2440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572118826055834e-05,
+      "loss": 0.4102,
+      "step": 2441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5719398711524694e-05,
+      "loss": 0.4669,
+      "step": 2442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5717609162491055e-05,
+      "loss": 0.522,
+      "step": 2443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.571581961345741e-05,
+      "loss": 0.4408,
+      "step": 2444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5714030064423766e-05,
+      "loss": 0.305,
+      "step": 2445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.571224051539013e-05,
+      "loss": 0.4006,
+      "step": 2446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.571045096635648e-05,
+      "loss": 0.4089,
+      "step": 2447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.570866141732283e-05,
+      "loss": 0.39,
+      "step": 2448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5706871868289193e-05,
+      "loss": 0.4228,
+      "step": 2449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.570508231925555e-05,
+      "loss": 0.3797,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5703292770221904e-05,
+      "loss": 0.6725,
+      "step": 2451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5701503221188266e-05,
+      "loss": 0.4146,
+      "step": 2452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.569971367215462e-05,
+      "loss": 0.4919,
+      "step": 2453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5697924123120976e-05,
+      "loss": 0.3327,
+      "step": 2454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.569613457408734e-05,
+      "loss": 0.4052,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5694345025053687e-05,
+      "loss": 0.3943,
+      "step": 2456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.569255547602004e-05,
+      "loss": 0.5493,
+      "step": 2457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5690765926986404e-05,
+      "loss": 0.5672,
+      "step": 2458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.568897637795276e-05,
+      "loss": 0.3954,
+      "step": 2459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5687186828919114e-05,
+      "loss": 0.4646,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.568539727988547e-05,
+      "loss": 0.3747,
+      "step": 2461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.568360773085183e-05,
+      "loss": 0.5341,
+      "step": 2462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5681818181818186e-05,
+      "loss": 0.4392,
+      "step": 2463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.568002863278454e-05,
+      "loss": 0.4383,
+      "step": 2464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.56782390837509e-05,
+      "loss": 0.5017,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.567644953471725e-05,
+      "loss": 0.3343,
+      "step": 2466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.567465998568361e-05,
+      "loss": 0.3519,
+      "step": 2467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.567287043664997e-05,
+      "loss": 0.2999,
+      "step": 2468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5671080887616324e-05,
+      "loss": 0.3916,
+      "step": 2469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.566929133858268e-05,
+      "loss": 0.3517,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5667501789549035e-05,
+      "loss": 0.3995,
+      "step": 2471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.566571224051539e-05,
+      "loss": 0.4609,
+      "step": 2472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5663922691481745e-05,
+      "loss": 0.4115,
+      "step": 2473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.566213314244811e-05,
+      "loss": 0.5407,
+      "step": 2474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.566034359341446e-05,
+      "loss": 0.5157,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.565855404438082e-05,
+      "loss": 0.4407,
+      "step": 2476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.565676449534718e-05,
+      "loss": 0.36,
+      "step": 2477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5654974946313535e-05,
+      "loss": 0.3725,
+      "step": 2478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.565318539727988e-05,
+      "loss": 0.5062,
+      "step": 2479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5651395848246245e-05,
+      "loss": 0.3386,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.56496062992126e-05,
+      "loss": 0.4886,
+      "step": 2481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5647816750178955e-05,
+      "loss": 0.608,
+      "step": 2482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.564602720114531e-05,
+      "loss": 0.4533,
+      "step": 2483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.564423765211167e-05,
+      "loss": 0.5377,
+      "step": 2484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.564244810307803e-05,
+      "loss": 0.3993,
+      "step": 2485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.564065855404438e-05,
+      "loss": 0.4954,
+      "step": 2486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.563886900501074e-05,
+      "loss": 0.3835,
+      "step": 2487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.563707945597709e-05,
+      "loss": 0.367,
+      "step": 2488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.563528990694345e-05,
+      "loss": 0.2797,
+      "step": 2489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.563350035790981e-05,
+      "loss": 0.3865,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5631710808876166e-05,
+      "loss": 0.4951,
+      "step": 2491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562992125984252e-05,
+      "loss": 0.3727,
+      "step": 2492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562813171080888e-05,
+      "loss": 0.3738,
+      "step": 2493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562634216177524e-05,
+      "loss": 0.4585,
+      "step": 2494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5624552612741586e-05,
+      "loss": 0.3722,
+      "step": 2495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562276306370795e-05,
+      "loss": 0.3568,
+      "step": 2496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5620973514674304e-05,
+      "loss": 0.3657,
+      "step": 2497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.561918396564066e-05,
+      "loss": 0.3178,
+      "step": 2498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.561739441660702e-05,
+      "loss": 0.3901,
+      "step": 2499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5615604867573376e-05,
+      "loss": 0.4429,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.561381531853973e-05,
+      "loss": 0.4295,
+      "step": 2501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5612025769506086e-05,
+      "loss": 0.4048,
+      "step": 2502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.561023622047244e-05,
+      "loss": 0.4028,
+      "step": 2503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.56084466714388e-05,
+      "loss": 0.4262,
+      "step": 2504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.560665712240515e-05,
+      "loss": 0.4161,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5604867573371514e-05,
+      "loss": 0.4422,
+      "step": 2506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.560307802433787e-05,
+      "loss": 0.3587,
+      "step": 2507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5601288475304224e-05,
+      "loss": 0.5134,
+      "step": 2508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5599498926270586e-05,
+      "loss": 0.3378,
+      "step": 2509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.559770937723694e-05,
+      "loss": 0.3559,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.559591982820329e-05,
+      "loss": 0.3733,
+      "step": 2511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.559413027916965e-05,
+      "loss": 0.427,
+      "step": 2512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.559234073013601e-05,
+      "loss": 0.3042,
+      "step": 2513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.559055118110236e-05,
+      "loss": 0.3836,
+      "step": 2514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5588761632068724e-05,
+      "loss": 0.3585,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.558697208303508e-05,
+      "loss": 0.4026,
+      "step": 2516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5585182534001435e-05,
+      "loss": 0.5136,
+      "step": 2517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.558339298496779e-05,
+      "loss": 0.5408,
+      "step": 2518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5581603435934145e-05,
+      "loss": 0.3732,
+      "step": 2519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.55798138869005e-05,
+      "loss": 0.3823,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557802433786686e-05,
+      "loss": 0.4008,
+      "step": 2521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557623478883322e-05,
+      "loss": 0.2934,
+      "step": 2522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557444523979957e-05,
+      "loss": 0.2713,
+      "step": 2523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557265569076593e-05,
+      "loss": 0.4741,
+      "step": 2524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557086614173229e-05,
+      "loss": 0.3814,
+      "step": 2525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.556907659269864e-05,
+      "loss": 0.4904,
+      "step": 2526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5567287043665e-05,
+      "loss": 0.377,
+      "step": 2527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5565497494631355e-05,
+      "loss": 0.4471,
+      "step": 2528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.556370794559771e-05,
+      "loss": 0.4245,
+      "step": 2529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5561918396564066e-05,
+      "loss": 0.3724,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.556012884753043e-05,
+      "loss": 0.2243,
+      "step": 2531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.555833929849678e-05,
+      "loss": 0.3577,
+      "step": 2532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.555654974946314e-05,
+      "loss": 0.4166,
+      "step": 2533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.555476020042949e-05,
+      "loss": 0.271,
+      "step": 2534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.555297065139585e-05,
+      "loss": 0.3721,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5551181102362204e-05,
+      "loss": 0.3266,
+      "step": 2536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5549391553328566e-05,
+      "loss": 0.395,
+      "step": 2537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.554760200429492e-05,
+      "loss": 0.3518,
+      "step": 2538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5545812455261276e-05,
+      "loss": 0.4824,
+      "step": 2539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.554402290622764e-05,
+      "loss": 0.4035,
+      "step": 2540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.554223335719399e-05,
+      "loss": 0.5883,
+      "step": 2541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.554044380816034e-05,
+      "loss": 0.3998,
+      "step": 2542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5538654259126704e-05,
+      "loss": 0.3653,
+      "step": 2543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.553686471009306e-05,
+      "loss": 0.4301,
+      "step": 2544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5535075161059414e-05,
+      "loss": 0.3806,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.553328561202577e-05,
+      "loss": 0.4127,
+      "step": 2546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.553149606299213e-05,
+      "loss": 0.2611,
+      "step": 2547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5529706513958486e-05,
+      "loss": 0.4094,
+      "step": 2548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.552791696492484e-05,
+      "loss": 0.3937,
+      "step": 2549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.55261274158912e-05,
+      "loss": 0.3343,
+      "step": 2550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.552433786685755e-05,
+      "loss": 0.4519,
+      "step": 2551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.552254831782391e-05,
+      "loss": 0.3562,
+      "step": 2552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.552075876879027e-05,
+      "loss": 0.45,
+      "step": 2553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5518969219756624e-05,
+      "loss": 0.5027,
+      "step": 2554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.551717967072298e-05,
+      "loss": 0.3517,
+      "step": 2555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.551539012168934e-05,
+      "loss": 0.3676,
+      "step": 2556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.551360057265569e-05,
+      "loss": 0.3706,
+      "step": 2557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5511811023622045e-05,
+      "loss": 0.5355,
+      "step": 2558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.551002147458841e-05,
+      "loss": 0.4309,
+      "step": 2559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.550823192555476e-05,
+      "loss": 0.3513,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.550644237652112e-05,
+      "loss": 0.4108,
+      "step": 2561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.550465282748748e-05,
+      "loss": 0.4327,
+      "step": 2562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5502863278453834e-05,
+      "loss": 0.3886,
+      "step": 2563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.550107372942019e-05,
+      "loss": 0.3635,
+      "step": 2564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5499284180386545e-05,
+      "loss": 0.4184,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.54974946313529e-05,
+      "loss": 0.4219,
+      "step": 2566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5495705082319255e-05,
+      "loss": 0.42,
+      "step": 2567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.549391553328561e-05,
+      "loss": 0.4407,
+      "step": 2568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.549212598425197e-05,
+      "loss": 0.3211,
+      "step": 2569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.549033643521833e-05,
+      "loss": 0.293,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.548854688618468e-05,
+      "loss": 0.3888,
+      "step": 2571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5486757337151045e-05,
+      "loss": 0.3142,
+      "step": 2572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.548496778811739e-05,
+      "loss": 0.4859,
+      "step": 2573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.548317823908375e-05,
+      "loss": 0.3324,
+      "step": 2574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.548138869005011e-05,
+      "loss": 0.3314,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5479599141016466e-05,
+      "loss": 0.4168,
+      "step": 2576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.547780959198282e-05,
+      "loss": 0.4905,
+      "step": 2577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.547602004294918e-05,
+      "loss": 0.6039,
+      "step": 2578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.547423049391554e-05,
+      "loss": 0.4923,
+      "step": 2579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.547244094488189e-05,
+      "loss": 0.4506,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.547065139584825e-05,
+      "loss": 0.6172,
+      "step": 2581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5468861846814603e-05,
+      "loss": 0.5414,
+      "step": 2582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.546707229778096e-05,
+      "loss": 0.4361,
+      "step": 2583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.546528274874732e-05,
+      "loss": 0.4687,
+      "step": 2584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5463493199713676e-05,
+      "loss": 0.5087,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.546170365068003e-05,
+      "loss": 0.4106,
+      "step": 2586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5459914101646386e-05,
+      "loss": 0.4644,
+      "step": 2587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.545812455261275e-05,
+      "loss": 0.4732,
+      "step": 2588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5456335003579097e-05,
+      "loss": 0.3846,
+      "step": 2589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.5132,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5452755905511814e-05,
+      "loss": 0.3977,
+      "step": 2591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.545096635647817e-05,
+      "loss": 0.4402,
+      "step": 2592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5449176807444524e-05,
+      "loss": 0.4092,
+      "step": 2593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5447387258410886e-05,
+      "loss": 0.4317,
+      "step": 2594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.544559770937724e-05,
+      "loss": 0.3746,
+      "step": 2595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5443808160343597e-05,
+      "loss": 0.3906,
+      "step": 2596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.544201861130995e-05,
+      "loss": 0.4165,
+      "step": 2597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.544022906227631e-05,
+      "loss": 0.3941,
+      "step": 2598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.543843951324266e-05,
+      "loss": 0.4406,
+      "step": 2599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5436649964209024e-05,
+      "loss": 0.5107,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.543486041517538e-05,
+      "loss": 0.3833,
+      "step": 2601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5433070866141734e-05,
+      "loss": 0.3582,
+      "step": 2602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5431281317108096e-05,
+      "loss": 0.3318,
+      "step": 2603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5429491768074445e-05,
+      "loss": 0.4566,
+      "step": 2604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.54277022190408e-05,
+      "loss": 0.2805,
+      "step": 2605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542591267000716e-05,
+      "loss": 0.4462,
+      "step": 2606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542412312097352e-05,
+      "loss": 0.3176,
+      "step": 2607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542233357193987e-05,
+      "loss": 0.4147,
+      "step": 2608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542054402290623e-05,
+      "loss": 0.3669,
+      "step": 2609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.541875447387259e-05,
+      "loss": 0.4846,
+      "step": 2610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5416964924838945e-05,
+      "loss": 0.3471,
+      "step": 2611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.54151753758053e-05,
+      "loss": 0.3104,
+      "step": 2612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5413385826771655e-05,
+      "loss": 0.47,
+      "step": 2613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.541159627773801e-05,
+      "loss": 0.3348,
+      "step": 2614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5409806728704365e-05,
+      "loss": 0.4059,
+      "step": 2615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.540801717967073e-05,
+      "loss": 0.386,
+      "step": 2616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.540622763063708e-05,
+      "loss": 0.4923,
+      "step": 2617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.540443808160344e-05,
+      "loss": 0.3757,
+      "step": 2618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.54026485325698e-05,
+      "loss": 0.3859,
+      "step": 2619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.540085898353615e-05,
+      "loss": 0.4634,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5399069434502503e-05,
+      "loss": 0.3795,
+      "step": 2621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5397279885468865e-05,
+      "loss": 0.4415,
+      "step": 2622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.539549033643522e-05,
+      "loss": 0.5455,
+      "step": 2623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5393700787401576e-05,
+      "loss": 0.4718,
+      "step": 2624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.539191123836794e-05,
+      "loss": 0.3573,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.539012168933429e-05,
+      "loss": 0.3507,
+      "step": 2626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.538833214030065e-05,
+      "loss": 0.3187,
+      "step": 2627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5386542591267e-05,
+      "loss": 0.2837,
+      "step": 2628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.538475304223336e-05,
+      "loss": 0.417,
+      "step": 2629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5382963493199714e-05,
+      "loss": 0.4068,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.538117394416607e-05,
+      "loss": 0.3583,
+      "step": 2631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.537938439513243e-05,
+      "loss": 0.3458,
+      "step": 2632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5377594846098786e-05,
+      "loss": 0.4623,
+      "step": 2633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.537580529706514e-05,
+      "loss": 0.3836,
+      "step": 2634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.53740157480315e-05,
+      "loss": 0.3515,
+      "step": 2635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.537222619899785e-05,
+      "loss": 0.4484,
+      "step": 2636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.537043664996421e-05,
+      "loss": 0.4017,
+      "step": 2637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.536864710093057e-05,
+      "loss": 0.3362,
+      "step": 2638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5366857551896924e-05,
+      "loss": 0.3651,
+      "step": 2639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.536506800286328e-05,
+      "loss": 0.3901,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.536327845382964e-05,
+      "loss": 0.3861,
+      "step": 2641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5361488904795996e-05,
+      "loss": 0.3675,
+      "step": 2642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5359699355762345e-05,
+      "loss": 0.4092,
+      "step": 2643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.535790980672871e-05,
+      "loss": 0.3408,
+      "step": 2644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.535612025769506e-05,
+      "loss": 0.3872,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.535433070866142e-05,
+      "loss": 0.5711,
+      "step": 2646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.535254115962778e-05,
+      "loss": 0.5312,
+      "step": 2647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5350751610594134e-05,
+      "loss": 0.4382,
+      "step": 2648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.534896206156049e-05,
+      "loss": 0.4153,
+      "step": 2649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5347172512526845e-05,
+      "loss": 0.3912,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.53453829634932e-05,
+      "loss": 0.4542,
+      "step": 2651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5343593414459555e-05,
+      "loss": 0.5025,
+      "step": 2652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.534180386542591e-05,
+      "loss": 0.3908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.534001431639227e-05,
+      "loss": 0.3866,
+      "step": 2654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.533822476735863e-05,
+      "loss": 0.3415,
+      "step": 2655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.533643521832498e-05,
+      "loss": 0.4874,
+      "step": 2656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5334645669291345e-05,
+      "loss": 0.5265,
+      "step": 2657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.53328561202577e-05,
+      "loss": 0.491,
+      "step": 2658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.533106657122405e-05,
+      "loss": 0.3801,
+      "step": 2659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532927702219041e-05,
+      "loss": 0.4322,
+      "step": 2660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5327487473156765e-05,
+      "loss": 0.4251,
+      "step": 2661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532569792412312e-05,
+      "loss": 0.3858,
+      "step": 2662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532390837508948e-05,
+      "loss": 0.3483,
+      "step": 2663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532211882605584e-05,
+      "loss": 0.3714,
+      "step": 2664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532032927702219e-05,
+      "loss": 0.3935,
+      "step": 2665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5318539727988555e-05,
+      "loss": 0.4366,
+      "step": 2666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.53167501789549e-05,
+      "loss": 0.3731,
+      "step": 2667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.531496062992126e-05,
+      "loss": 0.3288,
+      "step": 2668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.531317108088762e-05,
+      "loss": 0.3347,
+      "step": 2669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5311381531853976e-05,
+      "loss": 0.4292,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.530959198282033e-05,
+      "loss": 0.4827,
+      "step": 2671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5307802433786686e-05,
+      "loss": 0.3915,
+      "step": 2672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.530601288475305e-05,
+      "loss": 0.3172,
+      "step": 2673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.53042233357194e-05,
+      "loss": 0.4163,
+      "step": 2674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.530243378668576e-05,
+      "loss": 0.198,
+      "step": 2675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5300644237652114e-05,
+      "loss": 0.4864,
+      "step": 2676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.529885468861847e-05,
+      "loss": 0.3843,
+      "step": 2677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5297065139584824e-05,
+      "loss": 0.3207,
+      "step": 2678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5295275590551186e-05,
+      "loss": 0.4839,
+      "step": 2679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.529348604151754e-05,
+      "loss": 0.3734,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5291696492483896e-05,
+      "loss": 0.4763,
+      "step": 2681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.528990694345026e-05,
+      "loss": 0.2684,
+      "step": 2682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.528811739441661e-05,
+      "loss": 0.5075,
+      "step": 2683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.528632784538296e-05,
+      "loss": 0.431,
+      "step": 2684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5284538296349324e-05,
+      "loss": 0.4222,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.528274874731568e-05,
+      "loss": 0.3151,
+      "step": 2686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5280959198282034e-05,
+      "loss": 0.3769,
+      "step": 2687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5279169649248396e-05,
+      "loss": 0.55,
+      "step": 2688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.527738010021475e-05,
+      "loss": 0.3087,
+      "step": 2689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.52755905511811e-05,
+      "loss": 0.3239,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.527380100214746e-05,
+      "loss": 0.3638,
+      "step": 2691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.527201145311382e-05,
+      "loss": 0.4185,
+      "step": 2692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.527022190408017e-05,
+      "loss": 0.3961,
+      "step": 2693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.526843235504653e-05,
+      "loss": 0.3489,
+      "step": 2694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.526664280601289e-05,
+      "loss": 0.3303,
+      "step": 2695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5264853256979245e-05,
+      "loss": 0.439,
+      "step": 2696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.52630637079456e-05,
+      "loss": 0.332,
+      "step": 2697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5261274158911955e-05,
+      "loss": 0.4003,
+      "step": 2698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.525948460987831e-05,
+      "loss": 0.371,
+      "step": 2699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5257695060844665e-05,
+      "loss": 0.4568,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.525590551181103e-05,
+      "loss": 0.3986,
+      "step": 2701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.525411596277738e-05,
+      "loss": 0.469,
+      "step": 2702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.525232641374374e-05,
+      "loss": 0.3394,
+      "step": 2703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.52505368647101e-05,
+      "loss": 0.3741,
+      "step": 2704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5248747315676455e-05,
+      "loss": 0.3774,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.52469577666428e-05,
+      "loss": 0.3856,
+      "step": 2706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5245168217609165e-05,
+      "loss": 0.4212,
+      "step": 2707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.524337866857552e-05,
+      "loss": 0.3466,
+      "step": 2708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5241589119541876e-05,
+      "loss": 0.3816,
+      "step": 2709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.523979957050824e-05,
+      "loss": 0.5048,
+      "step": 2710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.523801002147459e-05,
+      "loss": 0.2686,
+      "step": 2711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.523622047244095e-05,
+      "loss": 0.5635,
+      "step": 2712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.52344309234073e-05,
+      "loss": 0.4647,
+      "step": 2713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.523264137437366e-05,
+      "loss": 0.4218,
+      "step": 2714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5230851825340014e-05,
+      "loss": 0.3508,
+      "step": 2715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522906227630637e-05,
+      "loss": 0.5445,
+      "step": 2716
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522727272727273e-05,
+      "loss": 0.5334,
+      "step": 2717
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5225483178239086e-05,
+      "loss": 0.4949,
+      "step": 2718
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522369362920544e-05,
+      "loss": 0.4004,
+      "step": 2719
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.52219040801718e-05,
+      "loss": 0.3631,
+      "step": 2720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522011453113816e-05,
+      "loss": 0.3688,
+      "step": 2721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.521832498210451e-05,
+      "loss": 0.2897,
+      "step": 2722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.521653543307087e-05,
+      "loss": 0.3553,
+      "step": 2723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5214745884037224e-05,
+      "loss": 0.3661,
+      "step": 2724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.521295633500358e-05,
+      "loss": 0.4253,
+      "step": 2725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.521116678596994e-05,
+      "loss": 0.4099,
+      "step": 2726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5209377236936296e-05,
+      "loss": 0.3027,
+      "step": 2727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.520758768790265e-05,
+      "loss": 0.3811,
+      "step": 2728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5205798138869007e-05,
+      "loss": 0.4026,
+      "step": 2729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.520400858983536e-05,
+      "loss": 0.3352,
+      "step": 2730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.520221904080172e-05,
+      "loss": 0.2674,
+      "step": 2731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.520042949176808e-05,
+      "loss": 0.403,
+      "step": 2732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5198639942734434e-05,
+      "loss": 0.3371,
+      "step": 2733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.519685039370079e-05,
+      "loss": 0.3136,
+      "step": 2734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5195060844667145e-05,
+      "loss": 0.4126,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5193271295633506e-05,
+      "loss": 0.3219,
+      "step": 2736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5191481746599855e-05,
+      "loss": 0.3256,
+      "step": 2737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.518969219756622e-05,
+      "loss": 0.3035,
+      "step": 2738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.518790264853257e-05,
+      "loss": 0.4872,
+      "step": 2739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.518611309949893e-05,
+      "loss": 0.4069,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.518432355046528e-05,
+      "loss": 0.4011,
+      "step": 2741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5182534001431644e-05,
+      "loss": 0.4949,
+      "step": 2742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5180744452398e-05,
+      "loss": 0.4476,
+      "step": 2743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5178954903364355e-05,
+      "loss": 0.384,
+      "step": 2744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.517716535433071e-05,
+      "loss": 0.6214,
+      "step": 2745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5175375805297065e-05,
+      "loss": 0.3029,
+      "step": 2746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.517358625626342e-05,
+      "loss": 0.4716,
+      "step": 2747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.517179670722978e-05,
+      "loss": 0.5508,
+      "step": 2748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.517000715819614e-05,
+      "loss": 0.5214,
+      "step": 2749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.516821760916249e-05,
+      "loss": 0.3277,
+      "step": 2750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5166428060128855e-05,
+      "loss": 0.3116,
+      "step": 2751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.516463851109521e-05,
+      "loss": 0.3932,
+      "step": 2752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.516284896206156e-05,
+      "loss": 0.4957,
+      "step": 2753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.516105941302792e-05,
+      "loss": 0.4028,
+      "step": 2754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5159269863994275e-05,
+      "loss": 0.3365,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.515748031496063e-05,
+      "loss": 0.3835,
+      "step": 2756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5155690765926986e-05,
+      "loss": 0.3445,
+      "step": 2757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.515390121689335e-05,
+      "loss": 0.3075,
+      "step": 2758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.51521116678597e-05,
+      "loss": 0.2817,
+      "step": 2759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.515032211882606e-05,
+      "loss": 0.4465,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5148532569792413e-05,
+      "loss": 0.3476,
+      "step": 2761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.514674302075877e-05,
+      "loss": 0.4091,
+      "step": 2762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5144953471725124e-05,
+      "loss": 0.4009,
+      "step": 2763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5143163922691486e-05,
+      "loss": 0.3195,
+      "step": 2764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.514137437365784e-05,
+      "loss": 0.3294,
+      "step": 2765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5139584824624196e-05,
+      "loss": 0.3972,
+      "step": 2766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.513779527559056e-05,
+      "loss": 0.4296,
+      "step": 2767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.513600572655691e-05,
+      "loss": 0.3938,
+      "step": 2768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.513421617752326e-05,
+      "loss": 0.3308,
+      "step": 2769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5132426628489624e-05,
+      "loss": 0.3529,
+      "step": 2770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.513063707945598e-05,
+      "loss": 0.3835,
+      "step": 2771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5128847530422334e-05,
+      "loss": 0.334,
+      "step": 2772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5127057981388696e-05,
+      "loss": 0.3523,
+      "step": 2773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.512526843235505e-05,
+      "loss": 0.4154,
+      "step": 2774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5123478883321406e-05,
+      "loss": 0.4722,
+      "step": 2775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.512168933428776e-05,
+      "loss": 0.4781,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511989978525412e-05,
+      "loss": 0.4112,
+      "step": 2777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511811023622047e-05,
+      "loss": 0.4073,
+      "step": 2778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511632068718683e-05,
+      "loss": 0.4131,
+      "step": 2779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511453113815319e-05,
+      "loss": 0.3589,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5112741589119544e-05,
+      "loss": 0.3671,
+      "step": 2781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.51109520400859e-05,
+      "loss": 0.3699,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.510916249105226e-05,
+      "loss": 0.3531,
+      "step": 2783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.510737294201861e-05,
+      "loss": 0.549,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5105583392984965e-05,
+      "loss": 0.3108,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.510379384395133e-05,
+      "loss": 0.3459,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.510200429491768e-05,
+      "loss": 0.4734,
+      "step": 2787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.510021474588404e-05,
+      "loss": 0.4184,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.50984251968504e-05,
+      "loss": 0.3897,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5096635647816755e-05,
+      "loss": 0.3352,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.509484609878311e-05,
+      "loss": 0.3335,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5093056549749465e-05,
+      "loss": 0.4095,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.509126700071582e-05,
+      "loss": 0.4641,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5089477451682175e-05,
+      "loss": 0.2898,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.508768790264854e-05,
+      "loss": 0.466,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.508589835361489e-05,
+      "loss": 0.4956,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.508410880458125e-05,
+      "loss": 0.4997,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.50823192555476e-05,
+      "loss": 0.4559,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5080529706513965e-05,
+      "loss": 0.3948,
+      "step": 2799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.507874015748031e-05,
+      "loss": 0.3795,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5076950608446675e-05,
+      "loss": 0.4937,
+      "step": 2801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.507516105941303e-05,
+      "loss": 0.3835,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5073371510379386e-05,
+      "loss": 0.4281,
+      "step": 2803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.507158196134574e-05,
+      "loss": 0.4023,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.50697924123121e-05,
+      "loss": 0.4392,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.506800286327846e-05,
+      "loss": 0.3898,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.506621331424481e-05,
+      "loss": 0.4017,
+      "step": 2807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.506442376521117e-05,
+      "loss": 0.3762,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5062634216177524e-05,
+      "loss": 0.3244,
+      "step": 2809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.506084466714388e-05,
+      "loss": 0.3713,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.505905511811024e-05,
+      "loss": 0.3495,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5057265569076596e-05,
+      "loss": 0.4393,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.505547602004295e-05,
+      "loss": 0.4361,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.505368647100931e-05,
+      "loss": 0.4675,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.505189692197567e-05,
+      "loss": 0.4193,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.505010737294202e-05,
+      "loss": 0.4094,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.504831782390838e-05,
+      "loss": 0.356,
+      "step": 2817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5046528274874734e-05,
+      "loss": 0.4516,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.504473872584109e-05,
+      "loss": 0.2934,
+      "step": 2819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5042949176807444e-05,
+      "loss": 0.3785,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5041159627773806e-05,
+      "loss": 0.348,
+      "step": 2821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503937007874016e-05,
+      "loss": 0.4462,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503758052970652e-05,
+      "loss": 0.4047,
+      "step": 2823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503579098067287e-05,
+      "loss": 0.2177,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503400143163923e-05,
+      "loss": 0.3146,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.503221188260558e-05,
+      "loss": 0.3512,
+      "step": 2826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5030422333571944e-05,
+      "loss": 0.3002,
+      "step": 2827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.50286327845383e-05,
+      "loss": 0.349,
+      "step": 2828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5026843235504655e-05,
+      "loss": 0.4347,
+      "step": 2829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5025053686471017e-05,
+      "loss": 0.3785,
+      "step": 2830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5023264137437365e-05,
+      "loss": 0.4784,
+      "step": 2831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.502147458840372e-05,
+      "loss": 0.3106,
+      "step": 2832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501968503937008e-05,
+      "loss": 0.3358,
+      "step": 2833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501789549033644e-05,
+      "loss": 0.3213,
+      "step": 2834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501610594130279e-05,
+      "loss": 0.6042,
+      "step": 2835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5014316392269155e-05,
+      "loss": 0.352,
+      "step": 2836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501252684323551e-05,
+      "loss": 0.3327,
+      "step": 2837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5010737294201865e-05,
+      "loss": 0.2675,
+      "step": 2838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.500894774516822e-05,
+      "loss": 0.3109,
+      "step": 2839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5007158196134575e-05,
+      "loss": 0.2856,
+      "step": 2840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.500536864710093e-05,
+      "loss": 0.2084,
+      "step": 2841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5003579098067286e-05,
+      "loss": 0.3271,
+      "step": 2842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.500178954903365e-05,
+      "loss": 0.501,
+      "step": 2843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5e-05,
+      "loss": 0.2895,
+      "step": 2844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.499821045096636e-05,
+      "loss": 0.352,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.499642090193272e-05,
+      "loss": 0.5982,
+      "step": 2846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.499463135289907e-05,
+      "loss": 0.2925,
+      "step": 2847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4992841803865424e-05,
+      "loss": 0.531,
+      "step": 2848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4991052254831786e-05,
+      "loss": 0.347,
+      "step": 2849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.498926270579814e-05,
+      "loss": 0.3357,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4987473156764496e-05,
+      "loss": 0.2889,
+      "step": 2851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.498568360773086e-05,
+      "loss": 0.49,
+      "step": 2852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.498389405869721e-05,
+      "loss": 0.4032,
+      "step": 2853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.498210450966357e-05,
+      "loss": 0.3581,
+      "step": 2854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4980314960629924e-05,
+      "loss": 0.3208,
+      "step": 2855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.497852541159628e-05,
+      "loss": 0.2274,
+      "step": 2856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4976735862562634e-05,
+      "loss": 0.4771,
+      "step": 2857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4974946313528996e-05,
+      "loss": 0.285,
+      "step": 2858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.497315676449535e-05,
+      "loss": 0.3448,
+      "step": 2859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4971367215461706e-05,
+      "loss": 0.3969,
+      "step": 2860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.496957766642806e-05,
+      "loss": 0.4349,
+      "step": 2861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.496778811739442e-05,
+      "loss": 0.407,
+      "step": 2862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.496599856836077e-05,
+      "loss": 0.4764,
+      "step": 2863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.496420901932713e-05,
+      "loss": 0.4102,
+      "step": 2864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.496241947029349e-05,
+      "loss": 0.454,
+      "step": 2865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4960629921259844e-05,
+      "loss": 0.2659,
+      "step": 2866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.49588403722262e-05,
+      "loss": 0.4842,
+      "step": 2867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.495705082319256e-05,
+      "loss": 0.46,
+      "step": 2868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4955261274158917e-05,
+      "loss": 0.3208,
+      "step": 2869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4953471725125265e-05,
+      "loss": 0.4795,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.495168217609163e-05,
+      "loss": 0.4221,
+      "step": 2871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.494989262705798e-05,
+      "loss": 0.3144,
+      "step": 2872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.494810307802434e-05,
+      "loss": 0.3728,
+      "step": 2873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.49463135289907e-05,
+      "loss": 0.494,
+      "step": 2874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4944523979957054e-05,
+      "loss": 0.4646,
+      "step": 2875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.494273443092341e-05,
+      "loss": 0.3716,
+      "step": 2876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.494094488188977e-05,
+      "loss": 0.2803,
+      "step": 2877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.493915533285612e-05,
+      "loss": 0.5691,
+      "step": 2878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4937365783822475e-05,
+      "loss": 0.2801,
+      "step": 2879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.493557623478884e-05,
+      "loss": 0.3544,
+      "step": 2880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.493378668575519e-05,
+      "loss": 0.3347,
+      "step": 2881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.493199713672155e-05,
+      "loss": 0.4181,
+      "step": 2882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.49302075876879e-05,
+      "loss": 0.4437,
+      "step": 2883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4928418038654265e-05,
+      "loss": 0.3057,
+      "step": 2884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.492662848962062e-05,
+      "loss": 0.3757,
+      "step": 2885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4924838940586975e-05,
+      "loss": 0.2974,
+      "step": 2886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.492304939155333e-05,
+      "loss": 0.3557,
+      "step": 2887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4921259842519686e-05,
+      "loss": 0.3735,
+      "step": 2888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.491947029348604e-05,
+      "loss": 0.3345,
+      "step": 2889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.49176807444524e-05,
+      "loss": 0.3241,
+      "step": 2890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.491589119541876e-05,
+      "loss": 0.3086,
+      "step": 2891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.491410164638511e-05,
+      "loss": 0.3075,
+      "step": 2892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4912312097351475e-05,
+      "loss": 0.3103,
+      "step": 2893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4910522548317823e-05,
+      "loss": 0.3847,
+      "step": 2894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.490873299928418e-05,
+      "loss": 0.3401,
+      "step": 2895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.490694345025054e-05,
+      "loss": 0.5129,
+      "step": 2896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4905153901216896e-05,
+      "loss": 0.4071,
+      "step": 2897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.490336435218325e-05,
+      "loss": 0.3751,
+      "step": 2898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.490157480314961e-05,
+      "loss": 0.36,
+      "step": 2899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.489978525411597e-05,
+      "loss": 0.3614,
+      "step": 2900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.489799570508232e-05,
+      "loss": 0.3752,
+      "step": 2901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.489620615604868e-05,
+      "loss": 0.3852,
+      "step": 2902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4894416607015034e-05,
+      "loss": 0.5601,
+      "step": 2903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.489262705798139e-05,
+      "loss": 0.3875,
+      "step": 2904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4890837508947744e-05,
+      "loss": 0.3394,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4889047959914106e-05,
+      "loss": 0.3356,
+      "step": 2906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.488725841088046e-05,
+      "loss": 0.4465,
+      "step": 2907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4885468861846816e-05,
+      "loss": 0.4683,
+      "step": 2908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.488367931281317e-05,
+      "loss": 0.4747,
+      "step": 2909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.488188976377953e-05,
+      "loss": 0.3381,
+      "step": 2910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.488010021474588e-05,
+      "loss": 0.3055,
+      "step": 2911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4878310665712244e-05,
+      "loss": 0.364,
+      "step": 2912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.48765211166786e-05,
+      "loss": 0.2746,
+      "step": 2913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4874731567644954e-05,
+      "loss": 0.4255,
+      "step": 2914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4872942018611316e-05,
+      "loss": 0.3374,
+      "step": 2915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.487115246957767e-05,
+      "loss": 0.4513,
+      "step": 2916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.486936292054402e-05,
+      "loss": 0.4933,
+      "step": 2917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.486757337151038e-05,
+      "loss": 0.314,
+      "step": 2918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.486578382247674e-05,
+      "loss": 0.2278,
+      "step": 2919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.486399427344309e-05,
+      "loss": 0.4333,
+      "step": 2920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4862204724409454e-05,
+      "loss": 0.474,
+      "step": 2921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.486041517537581e-05,
+      "loss": 0.4674,
+      "step": 2922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4858625626342165e-05,
+      "loss": 0.3859,
+      "step": 2923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.485683607730852e-05,
+      "loss": 0.4534,
+      "step": 2924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4855046528274875e-05,
+      "loss": 0.4061,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.485325697924123e-05,
+      "loss": 0.3227,
+      "step": 2926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4851467430207585e-05,
+      "loss": 0.4336,
+      "step": 2927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.484967788117395e-05,
+      "loss": 0.4174,
+      "step": 2928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.48478883321403e-05,
+      "loss": 0.302,
+      "step": 2929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.484609878310666e-05,
+      "loss": 0.4126,
+      "step": 2930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.484430923407302e-05,
+      "loss": 0.312,
+      "step": 2931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4842519685039375e-05,
+      "loss": 0.2548,
+      "step": 2932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4840730136005723e-05,
+      "loss": 0.3772,
+      "step": 2933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4838940586972085e-05,
+      "loss": 0.2902,
+      "step": 2934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.483715103793844e-05,
+      "loss": 0.3586,
+      "step": 2935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4835361488904796e-05,
+      "loss": 0.3111,
+      "step": 2936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.483357193987116e-05,
+      "loss": 0.3796,
+      "step": 2937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.483178239083751e-05,
+      "loss": 0.1867,
+      "step": 2938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482999284180387e-05,
+      "loss": 0.3161,
+      "step": 2939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482820329277022e-05,
+      "loss": 0.4658,
+      "step": 2940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482641374373658e-05,
+      "loss": 0.3451,
+      "step": 2941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4824624194702934e-05,
+      "loss": 0.5548,
+      "step": 2942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4822834645669296e-05,
+      "loss": 0.3347,
+      "step": 2943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482104509663565e-05,
+      "loss": 0.4938,
+      "step": 2944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4819255547602006e-05,
+      "loss": 0.3248,
+      "step": 2945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481746599856836e-05,
+      "loss": 0.2695,
+      "step": 2946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481567644953472e-05,
+      "loss": 0.4267,
+      "step": 2947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481388690050108e-05,
+      "loss": 0.4538,
+      "step": 2948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4812097351467434e-05,
+      "loss": 0.4557,
+      "step": 2949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481030780243379e-05,
+      "loss": 0.2874,
+      "step": 2950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4808518253400144e-05,
+      "loss": 0.2968,
+      "step": 2951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.48067287043665e-05,
+      "loss": 0.487,
+      "step": 2952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.480493915533286e-05,
+      "loss": 0.3549,
+      "step": 2953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4803149606299216e-05,
+      "loss": 0.3803,
+      "step": 2954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.480136005726557e-05,
+      "loss": 0.3447,
+      "step": 2955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.479957050823193e-05,
+      "loss": 0.3428,
+      "step": 2956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.479778095919828e-05,
+      "loss": 0.545,
+      "step": 2957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.479599141016464e-05,
+      "loss": 0.3753,
+      "step": 2958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4794201861131e-05,
+      "loss": 0.4903,
+      "step": 2959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4792412312097354e-05,
+      "loss": 0.4316,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.479062276306371e-05,
+      "loss": 0.425,
+      "step": 2961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478883321403007e-05,
+      "loss": 0.3047,
+      "step": 2962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478704366499643e-05,
+      "loss": 0.3781,
+      "step": 2963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4785254115962775e-05,
+      "loss": 0.3306,
+      "step": 2964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478346456692914e-05,
+      "loss": 0.4391,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.478167501789549e-05,
+      "loss": 0.4485,
+      "step": 2966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.477988546886185e-05,
+      "loss": 0.332,
+      "step": 2967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.47780959198282e-05,
+      "loss": 0.4265,
+      "step": 2968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4776306370794565e-05,
+      "loss": 0.3632,
+      "step": 2969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.477451682176092e-05,
+      "loss": 0.278,
+      "step": 2970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4772727272727275e-05,
+      "loss": 0.2976,
+      "step": 2971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.477093772369363e-05,
+      "loss": 0.2854,
+      "step": 2972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4769148174659985e-05,
+      "loss": 0.4545,
+      "step": 2973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.476735862562634e-05,
+      "loss": 0.284,
+      "step": 2974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.47655690765927e-05,
+      "loss": 0.3917,
+      "step": 2975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.476377952755906e-05,
+      "loss": 0.3573,
+      "step": 2976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.476198997852541e-05,
+      "loss": 0.4453,
+      "step": 2977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4760200429491775e-05,
+      "loss": 0.3551,
+      "step": 2978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.475841088045813e-05,
+      "loss": 0.5063,
+      "step": 2979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.475662133142448e-05,
+      "loss": 0.453,
+      "step": 2980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.475483178239084e-05,
+      "loss": 0.4224,
+      "step": 2981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4753042233357196e-05,
+      "loss": 0.4008,
+      "step": 2982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.475125268432355e-05,
+      "loss": 0.4782,
+      "step": 2983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474946313528991e-05,
+      "loss": 0.2385,
+      "step": 2984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474767358625627e-05,
+      "loss": 0.4187,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474588403722262e-05,
+      "loss": 0.5798,
+      "step": 2986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474409448818898e-05,
+      "loss": 0.3542,
+      "step": 2987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4742304939155334e-05,
+      "loss": 0.344,
+      "step": 2988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.474051539012169e-05,
+      "loss": 0.308,
+      "step": 2989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4738725841088044e-05,
+      "loss": 0.4418,
+      "step": 2990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4736936292054406e-05,
+      "loss": 0.315,
+      "step": 2991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.473514674302076e-05,
+      "loss": 0.3315,
+      "step": 2992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4733357193987116e-05,
+      "loss": 0.4769,
+      "step": 2993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.473156764495348e-05,
+      "loss": 0.3604,
+      "step": 2994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.472977809591983e-05,
+      "loss": 0.3364,
+      "step": 2995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.472798854688618e-05,
+      "loss": 0.4048,
+      "step": 2996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4726198997852544e-05,
+      "loss": 0.3938,
+      "step": 2997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.47244094488189e-05,
+      "loss": 0.3854,
+      "step": 2998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4722619899785254e-05,
+      "loss": 0.3752,
+      "step": 2999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4720830350751616e-05,
+      "loss": 0.3836,
+      "step": 3000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471904080171797e-05,
+      "loss": 0.3265,
+      "step": 3001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4717251252684327e-05,
+      "loss": 0.4343,
+      "step": 3002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471546170365068e-05,
+      "loss": 0.4052,
+      "step": 3003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471367215461704e-05,
+      "loss": 0.5264,
+      "step": 3004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471188260558339e-05,
+      "loss": 0.3903,
+      "step": 3005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4710093056549754e-05,
+      "loss": 0.3277,
+      "step": 3006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.470830350751611e-05,
+      "loss": 0.3595,
+      "step": 3007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4706513958482465e-05,
+      "loss": 0.3907,
+      "step": 3008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.470472440944882e-05,
+      "loss": 0.4361,
+      "step": 3009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.470293486041518e-05,
+      "loss": 0.4442,
+      "step": 3010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.470114531138153e-05,
+      "loss": 0.5147,
+      "step": 3011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4699355762347885e-05,
+      "loss": 0.3376,
+      "step": 3012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.469756621331425e-05,
+      "loss": 0.4449,
+      "step": 3013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.46957766642806e-05,
+      "loss": 0.3424,
+      "step": 3014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.469398711524696e-05,
+      "loss": 0.4193,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.469219756621332e-05,
+      "loss": 0.3411,
+      "step": 3016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4690408017179675e-05,
+      "loss": 0.2264,
+      "step": 3017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.468861846814603e-05,
+      "loss": 0.3718,
+      "step": 3018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4686828919112385e-05,
+      "loss": 0.4069,
+      "step": 3019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.468503937007874e-05,
+      "loss": 0.2506,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4683249821045096e-05,
+      "loss": 0.4376,
+      "step": 3021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.468146027201146e-05,
+      "loss": 0.3283,
+      "step": 3022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.467967072297781e-05,
+      "loss": 0.4309,
+      "step": 3023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.467788117394417e-05,
+      "loss": 0.4408,
+      "step": 3024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.467609162491053e-05,
+      "loss": 0.355,
+      "step": 3025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4674302075876885e-05,
+      "loss": 0.4456,
+      "step": 3026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4672512526843234e-05,
+      "loss": 0.4757,
+      "step": 3027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4670722977809596e-05,
+      "loss": 0.3411,
+      "step": 3028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.466893342877595e-05,
+      "loss": 0.3738,
+      "step": 3029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4667143879742306e-05,
+      "loss": 0.4809,
+      "step": 3030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.466535433070866e-05,
+      "loss": 0.2528,
+      "step": 3031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.466356478167502e-05,
+      "loss": 0.336,
+      "step": 3032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.466177523264138e-05,
+      "loss": 0.2932,
+      "step": 3033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4659985683607733e-05,
+      "loss": 0.3611,
+      "step": 3034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.465819613457409e-05,
+      "loss": 0.4499,
+      "step": 3035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4656406585540444e-05,
+      "loss": 0.2268,
+      "step": 3036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.46546170365068e-05,
+      "loss": 0.3749,
+      "step": 3037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.465282748747316e-05,
+      "loss": 0.4327,
+      "step": 3038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4651037938439516e-05,
+      "loss": 0.3259,
+      "step": 3039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.464924838940587e-05,
+      "loss": 0.2949,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.464745884037223e-05,
+      "loss": 0.4577,
+      "step": 3041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.464566929133858e-05,
+      "loss": 0.4827,
+      "step": 3042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.464387974230494e-05,
+      "loss": 0.388,
+      "step": 3043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.46420901932713e-05,
+      "loss": 0.392,
+      "step": 3044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4640300644237654e-05,
+      "loss": 0.4137,
+      "step": 3045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463851109520401e-05,
+      "loss": 0.3161,
+      "step": 3046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463672154617037e-05,
+      "loss": 0.2923,
+      "step": 3047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4634931997136726e-05,
+      "loss": 0.2552,
+      "step": 3048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463314244810308e-05,
+      "loss": 0.4907,
+      "step": 3049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463135289906944e-05,
+      "loss": 0.3348,
+      "step": 3050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462956335003579e-05,
+      "loss": 0.4199,
+      "step": 3051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462777380100215e-05,
+      "loss": 0.4707,
+      "step": 3052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46259842519685e-05,
+      "loss": 0.4146,
+      "step": 3053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4624194702934864e-05,
+      "loss": 0.366,
+      "step": 3054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462240515390122e-05,
+      "loss": 0.3076,
+      "step": 3055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4620615604867575e-05,
+      "loss": 0.3678,
+      "step": 3056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461882605583394e-05,
+      "loss": 0.3597,
+      "step": 3057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4617036506800285e-05,
+      "loss": 0.3656,
+      "step": 3058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461524695776664e-05,
+      "loss": 0.3593,
+      "step": 3059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4613457408733e-05,
+      "loss": 0.2712,
+      "step": 3060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461166785969936e-05,
+      "loss": 0.3846,
+      "step": 3061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.460987831066571e-05,
+      "loss": 0.3853,
+      "step": 3062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4608088761632075e-05,
+      "loss": 0.3528,
+      "step": 3063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.460629921259843e-05,
+      "loss": 0.4362,
+      "step": 3064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4604509663564785e-05,
+      "loss": 0.4904,
+      "step": 3065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.460272011453114e-05,
+      "loss": 0.3934,
+      "step": 3066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4600930565497495e-05,
+      "loss": 0.4648,
+      "step": 3067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459914101646385e-05,
+      "loss": 0.4807,
+      "step": 3068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459735146743021e-05,
+      "loss": 0.4815,
+      "step": 3069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459556191839657e-05,
+      "loss": 0.2675,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459377236936292e-05,
+      "loss": 0.3378,
+      "step": 3071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459198282032928e-05,
+      "loss": 0.3629,
+      "step": 3072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459019327129564e-05,
+      "loss": 0.4716,
+      "step": 3073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458840372226199e-05,
+      "loss": 0.4199,
+      "step": 3074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4586614173228344e-05,
+      "loss": 0.2319,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4584824624194706e-05,
+      "loss": 0.2792,
+      "step": 3076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458303507516106e-05,
+      "loss": 0.3173,
+      "step": 3077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4581245526127416e-05,
+      "loss": 0.4222,
+      "step": 3078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457945597709378e-05,
+      "loss": 0.2911,
+      "step": 3079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457766642806013e-05,
+      "loss": 0.3259,
+      "step": 3080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457587687902649e-05,
+      "loss": 0.3387,
+      "step": 3081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4574087329992844e-05,
+      "loss": 0.4047,
+      "step": 3082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45722977809592e-05,
+      "loss": 0.3689,
+      "step": 3083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4570508231925554e-05,
+      "loss": 0.5098,
+      "step": 3084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4568718682891916e-05,
+      "loss": 0.3199,
+      "step": 3085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456692913385827e-05,
+      "loss": 0.3418,
+      "step": 3086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4565139584824626e-05,
+      "loss": 0.2943,
+      "step": 3087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456335003579098e-05,
+      "loss": 0.3311,
+      "step": 3088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456156048675734e-05,
+      "loss": 0.4449,
+      "step": 3089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455977093772369e-05,
+      "loss": 0.2908,
+      "step": 3090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4557981388690054e-05,
+      "loss": 0.5063,
+      "step": 3091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455619183965641e-05,
+      "loss": 0.2935,
+      "step": 3092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4554402290622764e-05,
+      "loss": 0.3768,
+      "step": 3093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455261274158912e-05,
+      "loss": 0.4248,
+      "step": 3094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455082319255548e-05,
+      "loss": 0.4887,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454903364352184e-05,
+      "loss": 0.3712,
+      "step": 3096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454724409448819e-05,
+      "loss": 0.3286,
+      "step": 3097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454545454545455e-05,
+      "loss": 0.3478,
+      "step": 3098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.45436649964209e-05,
+      "loss": 0.349,
+      "step": 3099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454187544738726e-05,
+      "loss": 0.3317,
+      "step": 3100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454008589835362e-05,
+      "loss": 0.3257,
+      "step": 3101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4538296349319975e-05,
+      "loss": 0.3966,
+      "step": 3102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453650680028633e-05,
+      "loss": 0.328,
+      "step": 3103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453471725125269e-05,
+      "loss": 0.2694,
+      "step": 3104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453292770221904e-05,
+      "loss": 0.3972,
+      "step": 3105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4531138153185395e-05,
+      "loss": 0.4543,
+      "step": 3106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452934860415176e-05,
+      "loss": 0.2495,
+      "step": 3107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452755905511811e-05,
+      "loss": 0.3986,
+      "step": 3108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452576950608447e-05,
+      "loss": 0.3083,
+      "step": 3109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452397995705083e-05,
+      "loss": 0.2849,
+      "step": 3110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4522190408017185e-05,
+      "loss": 0.3539,
+      "step": 3111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452040085898354e-05,
+      "loss": 0.3324,
+      "step": 3112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4518611309949895e-05,
+      "loss": 0.3177,
+      "step": 3113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451682176091625e-05,
+      "loss": 0.353,
+      "step": 3114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4515032211882606e-05,
+      "loss": 0.3233,
+      "step": 3115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451324266284896e-05,
+      "loss": 0.3513,
+      "step": 3116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451145311381532e-05,
+      "loss": 0.3084,
+      "step": 3117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450966356478168e-05,
+      "loss": 0.458,
+      "step": 3118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450787401574803e-05,
+      "loss": 0.3428,
+      "step": 3119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4506084466714395e-05,
+      "loss": 0.3828,
+      "step": 3120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4504294917680744e-05,
+      "loss": 0.4247,
+      "step": 3121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.45025053686471e-05,
+      "loss": 0.3645,
+      "step": 3122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450071581961346e-05,
+      "loss": 0.3174,
+      "step": 3123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4498926270579816e-05,
+      "loss": 0.4108,
+      "step": 3124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449713672154617e-05,
+      "loss": 0.4268,
+      "step": 3125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449534717251253e-05,
+      "loss": 0.3794,
+      "step": 3126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449355762347889e-05,
+      "loss": 0.4628,
+      "step": 3127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449176807444524e-05,
+      "loss": 0.3103,
+      "step": 3128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44899785254116e-05,
+      "loss": 0.425,
+      "step": 3129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4488188976377954e-05,
+      "loss": 0.3711,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448639942734431e-05,
+      "loss": 0.3704,
+      "step": 3131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448460987831067e-05,
+      "loss": 0.4014,
+      "step": 3132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4482820329277026e-05,
+      "loss": 0.3575,
+      "step": 3133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448103078024338e-05,
+      "loss": 0.3477,
+      "step": 3134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447924123120974e-05,
+      "loss": 0.385,
+      "step": 3135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447745168217609e-05,
+      "loss": 0.2858,
+      "step": 3136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447566213314245e-05,
+      "loss": 0.357,
+      "step": 3137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44738725841088e-05,
+      "loss": 0.2991,
+      "step": 3138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4472083035075164e-05,
+      "loss": 0.3141,
+      "step": 3139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447029348604152e-05,
+      "loss": 0.3255,
+      "step": 3140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4468503937007875e-05,
+      "loss": 0.3259,
+      "step": 3141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4466714387974237e-05,
+      "loss": 0.3499,
+      "step": 3142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446492483894059e-05,
+      "loss": 0.4206,
+      "step": 3143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446313528990694e-05,
+      "loss": 0.374,
+      "step": 3144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44613457408733e-05,
+      "loss": 0.3659,
+      "step": 3145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445955619183966e-05,
+      "loss": 0.4269,
+      "step": 3146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445776664280601e-05,
+      "loss": 0.3667,
+      "step": 3147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4455977093772375e-05,
+      "loss": 0.5174,
+      "step": 3148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445418754473873e-05,
+      "loss": 0.4075,
+      "step": 3149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4452397995705085e-05,
+      "loss": 0.348,
+      "step": 3150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445060844667144e-05,
+      "loss": 0.426,
+      "step": 3151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4448818897637795e-05,
+      "loss": 0.5295,
+      "step": 3152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444702934860415e-05,
+      "loss": 0.3669,
+      "step": 3153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444523979957051e-05,
+      "loss": 0.2722,
+      "step": 3154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444345025053687e-05,
+      "loss": 0.3716,
+      "step": 3155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444166070150322e-05,
+      "loss": 0.4233,
+      "step": 3156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443987115246958e-05,
+      "loss": 0.3159,
+      "step": 3157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443808160343594e-05,
+      "loss": 0.4829,
+      "step": 3158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4436292054402295e-05,
+      "loss": 0.3013,
+      "step": 3159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443450250536865e-05,
+      "loss": 0.2946,
+      "step": 3160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4432712956335006e-05,
+      "loss": 0.5806,
+      "step": 3161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443092340730136e-05,
+      "loss": 0.3726,
+      "step": 3162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4429133858267716e-05,
+      "loss": 0.3767,
+      "step": 3163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442734430923408e-05,
+      "loss": 0.3543,
+      "step": 3164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442555476020043e-05,
+      "loss": 0.4331,
+      "step": 3165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442376521116679e-05,
+      "loss": 0.3832,
+      "step": 3166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442197566213315e-05,
+      "loss": 0.39,
+      "step": 3167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.44201861130995e-05,
+      "loss": 0.2683,
+      "step": 3168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4418396564065854e-05,
+      "loss": 0.3235,
+      "step": 3169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4416607015032216e-05,
+      "loss": 0.4238,
+      "step": 3170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.441481746599857e-05,
+      "loss": 0.388,
+      "step": 3171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4413027916964926e-05,
+      "loss": 0.3974,
+      "step": 3172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.441123836793129e-05,
+      "loss": 0.3884,
+      "step": 3173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4409448818897643e-05,
+      "loss": 0.3057,
+      "step": 3174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.440765926986399e-05,
+      "loss": 0.4766,
+      "step": 3175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4405869720830354e-05,
+      "loss": 0.3741,
+      "step": 3176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.440408017179671e-05,
+      "loss": 0.4535,
+      "step": 3177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4402290622763064e-05,
+      "loss": 0.4632,
+      "step": 3178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.440050107372942e-05,
+      "loss": 0.4207,
+      "step": 3179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.439871152469578e-05,
+      "loss": 0.4056,
+      "step": 3180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4396921975662137e-05,
+      "loss": 0.309,
+      "step": 3181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.439513242662849e-05,
+      "loss": 0.3127,
+      "step": 3182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.439334287759485e-05,
+      "loss": 0.4095,
+      "step": 3183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.43915533285612e-05,
+      "loss": 0.3539,
+      "step": 3184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438976377952756e-05,
+      "loss": 0.3627,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438797423049392e-05,
+      "loss": 0.3558,
+      "step": 3186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4386184681460274e-05,
+      "loss": 0.3856,
+      "step": 3187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438439513242663e-05,
+      "loss": 0.3518,
+      "step": 3188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438260558339299e-05,
+      "loss": 0.3986,
+      "step": 3189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438081603435935e-05,
+      "loss": 0.366,
+      "step": 3190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4379026485325695e-05,
+      "loss": 0.3145,
+      "step": 3191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.437723693629206e-05,
+      "loss": 0.3282,
+      "step": 3192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.437544738725841e-05,
+      "loss": 0.4811,
+      "step": 3193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.437365783822477e-05,
+      "loss": 0.2654,
+      "step": 3194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.437186828919113e-05,
+      "loss": 0.2982,
+      "step": 3195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4370078740157485e-05,
+      "loss": 0.2982,
+      "step": 3196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.436828919112384e-05,
+      "loss": 0.3164,
+      "step": 3197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4366499642090195e-05,
+      "loss": 0.3559,
+      "step": 3198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.436471009305655e-05,
+      "loss": 0.4459,
+      "step": 3199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4362920544022906e-05,
+      "loss": 0.4512,
+      "step": 3200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.436113099498926e-05,
+      "loss": 0.2589,
+      "step": 3201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.435934144595562e-05,
+      "loss": 0.3178,
+      "step": 3202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.435755189692198e-05,
+      "loss": 0.5992,
+      "step": 3203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.435576234788833e-05,
+      "loss": 0.2327,
+      "step": 3204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4353972798854695e-05,
+      "loss": 0.4683,
+      "step": 3205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.435218324982105e-05,
+      "loss": 0.4249,
+      "step": 3206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.43503937007874e-05,
+      "loss": 0.3606,
+      "step": 3207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.434860415175376e-05,
+      "loss": 0.2462,
+      "step": 3208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4346814602720116e-05,
+      "loss": 0.4362,
+      "step": 3209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.434502505368647e-05,
+      "loss": 0.4429,
+      "step": 3210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.434323550465283e-05,
+      "loss": 0.4312,
+      "step": 3211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.434144595561919e-05,
+      "loss": 0.3061,
+      "step": 3212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.433965640658554e-05,
+      "loss": 0.3645,
+      "step": 3213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.43378668575519e-05,
+      "loss": 0.2616,
+      "step": 3214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4336077308518254e-05,
+      "loss": 0.3435,
+      "step": 3215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.433428775948461e-05,
+      "loss": 0.4578,
+      "step": 3216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.433249821045097e-05,
+      "loss": 0.2501,
+      "step": 3217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4330708661417326e-05,
+      "loss": 0.5019,
+      "step": 3218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432891911238368e-05,
+      "loss": 0.3412,
+      "step": 3219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4327129563350036e-05,
+      "loss": 0.3713,
+      "step": 3220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43253400143164e-05,
+      "loss": 0.2751,
+      "step": 3221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432355046528275e-05,
+      "loss": 0.2392,
+      "step": 3222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43217609162491e-05,
+      "loss": 0.2839,
+      "step": 3223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4319971367215464e-05,
+      "loss": 0.2316,
+      "step": 3224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.3091,
+      "step": 3225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4316392269148174e-05,
+      "loss": 0.441,
+      "step": 3226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4314602720114536e-05,
+      "loss": 0.4525,
+      "step": 3227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431281317108089e-05,
+      "loss": 0.3418,
+      "step": 3228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431102362204725e-05,
+      "loss": 0.3773,
+      "step": 3229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43092340730136e-05,
+      "loss": 0.5572,
+      "step": 3230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.430744452397996e-05,
+      "loss": 0.4543,
+      "step": 3231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.430565497494631e-05,
+      "loss": 0.3641,
+      "step": 3232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4303865425912674e-05,
+      "loss": 0.3565,
+      "step": 3233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.430207587687903e-05,
+      "loss": 0.3536,
+      "step": 3234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4300286327845385e-05,
+      "loss": 0.4695,
+      "step": 3235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.429849677881175e-05,
+      "loss": 0.4092,
+      "step": 3236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.42967072297781e-05,
+      "loss": 0.3874,
+      "step": 3237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.429491768074445e-05,
+      "loss": 0.3442,
+      "step": 3238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.429312813171081e-05,
+      "loss": 0.457,
+      "step": 3239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.429133858267717e-05,
+      "loss": 0.2676,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.428954903364352e-05,
+      "loss": 0.3955,
+      "step": 3241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.428775948460988e-05,
+      "loss": 0.3847,
+      "step": 3242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.428596993557624e-05,
+      "loss": 0.2744,
+      "step": 3243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4284180386542595e-05,
+      "loss": 0.4528,
+      "step": 3244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.428239083750895e-05,
+      "loss": 0.425,
+      "step": 3245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4280601288475305e-05,
+      "loss": 0.3493,
+      "step": 3246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.427881173944166e-05,
+      "loss": 0.3525,
+      "step": 3247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4277022190408016e-05,
+      "loss": 0.4213,
+      "step": 3248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.427523264137438e-05,
+      "loss": 0.3598,
+      "step": 3249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.427344309234073e-05,
+      "loss": 0.4474,
+      "step": 3250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.427165354330709e-05,
+      "loss": 0.4313,
+      "step": 3251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.426986399427345e-05,
+      "loss": 0.4082,
+      "step": 3252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4268074445239805e-05,
+      "loss": 0.5209,
+      "step": 3253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4266284896206154e-05,
+      "loss": 0.267,
+      "step": 3254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4264495347172516e-05,
+      "loss": 0.3624,
+      "step": 3255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.426270579813887e-05,
+      "loss": 0.312,
+      "step": 3256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4260916249105226e-05,
+      "loss": 0.3775,
+      "step": 3257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.425912670007159e-05,
+      "loss": 0.2676,
+      "step": 3258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.425733715103794e-05,
+      "loss": 0.3545,
+      "step": 3259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.42555476020043e-05,
+      "loss": 0.3573,
+      "step": 3260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4253758052970654e-05,
+      "loss": 0.3552,
+      "step": 3261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.425196850393701e-05,
+      "loss": 0.3438,
+      "step": 3262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4250178954903364e-05,
+      "loss": 0.4447,
+      "step": 3263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.424838940586972e-05,
+      "loss": 0.3441,
+      "step": 3264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.424659985683608e-05,
+      "loss": 0.4854,
+      "step": 3265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4244810307802436e-05,
+      "loss": 0.332,
+      "step": 3266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.424302075876879e-05,
+      "loss": 0.4883,
+      "step": 3267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4241231209735154e-05,
+      "loss": 0.4167,
+      "step": 3268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.42394416607015e-05,
+      "loss": 0.3114,
+      "step": 3269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.423765211166786e-05,
+      "loss": 0.2826,
+      "step": 3270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.423586256263422e-05,
+      "loss": 0.2395,
+      "step": 3271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4234073013600574e-05,
+      "loss": 0.4246,
+      "step": 3272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.423228346456693e-05,
+      "loss": 0.3099,
+      "step": 3273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.423049391553329e-05,
+      "loss": 0.2645,
+      "step": 3274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422870436649965e-05,
+      "loss": 0.4768,
+      "step": 3275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4226914817466e-05,
+      "loss": 0.4259,
+      "step": 3276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422512526843236e-05,
+      "loss": 0.2577,
+      "step": 3277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422333571939871e-05,
+      "loss": 0.3982,
+      "step": 3278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422154617036507e-05,
+      "loss": 0.3611,
+      "step": 3279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421975662133143e-05,
+      "loss": 0.4059,
+      "step": 3280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4217967072297785e-05,
+      "loss": 0.287,
+      "step": 3281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421617752326414e-05,
+      "loss": 0.5629,
+      "step": 3282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4214387974230495e-05,
+      "loss": 0.3828,
+      "step": 3283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421259842519686e-05,
+      "loss": 0.4227,
+      "step": 3284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4210808876163205e-05,
+      "loss": 0.3308,
+      "step": 3285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.420901932712956e-05,
+      "loss": 0.476,
+      "step": 3286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.420722977809592e-05,
+      "loss": 0.4689,
+      "step": 3287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.420544022906228e-05,
+      "loss": 0.414,
+      "step": 3288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.420365068002863e-05,
+      "loss": 0.4773,
+      "step": 3289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4201861130994995e-05,
+      "loss": 0.3402,
+      "step": 3290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.420007158196135e-05,
+      "loss": 0.328,
+      "step": 3291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4198282032927705e-05,
+      "loss": 0.4259,
+      "step": 3292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.419649248389406e-05,
+      "loss": 0.4183,
+      "step": 3293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4194702934860416e-05,
+      "loss": 0.2684,
+      "step": 3294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.419291338582677e-05,
+      "loss": 0.3191,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.419112383679313e-05,
+      "loss": 0.3302,
+      "step": 3296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.418933428775949e-05,
+      "loss": 0.5,
+      "step": 3297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.418754473872584e-05,
+      "loss": 0.292,
+      "step": 3298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.41857551896922e-05,
+      "loss": 0.2937,
+      "step": 3299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.418396564065856e-05,
+      "loss": 0.3031,
+      "step": 3300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.418217609162491e-05,
+      "loss": 0.3752,
+      "step": 3301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.418038654259127e-05,
+      "loss": 0.4042,
+      "step": 3302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4178596993557626e-05,
+      "loss": 0.4043,
+      "step": 3303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.417680744452398e-05,
+      "loss": 0.2635,
+      "step": 3304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4175017895490336e-05,
+      "loss": 0.2144,
+      "step": 3305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.41732283464567e-05,
+      "loss": 0.4644,
+      "step": 3306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4171438797423053e-05,
+      "loss": 0.5565,
+      "step": 3307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.416964924838941e-05,
+      "loss": 0.2849,
+      "step": 3308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4167859699355764e-05,
+      "loss": 0.5092,
+      "step": 3309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.416607015032212e-05,
+      "loss": 0.3924,
+      "step": 3310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4164280601288474e-05,
+      "loss": 0.3398,
+      "step": 3311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4162491052254836e-05,
+      "loss": 0.341,
+      "step": 3312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.416070150322119e-05,
+      "loss": 0.3275,
+      "step": 3313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4158911954187547e-05,
+      "loss": 0.2901,
+      "step": 3314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.415712240515391e-05,
+      "loss": 0.2616,
+      "step": 3315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.415533285612026e-05,
+      "loss": 0.4808,
+      "step": 3316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.415354330708661e-05,
+      "loss": 0.3121,
+      "step": 3317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4151753758052974e-05,
+      "loss": 0.4784,
+      "step": 3318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.414996420901933e-05,
+      "loss": 0.3473,
+      "step": 3319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4148174659985685e-05,
+      "loss": 0.3723,
+      "step": 3320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4146385110952046e-05,
+      "loss": 0.3684,
+      "step": 3321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.41445955619184e-05,
+      "loss": 0.28,
+      "step": 3322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.414280601288476e-05,
+      "loss": 0.2725,
+      "step": 3323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.414101646385111e-05,
+      "loss": 0.3303,
+      "step": 3324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413922691481747e-05,
+      "loss": 0.4225,
+      "step": 3325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413743736578382e-05,
+      "loss": 0.3732,
+      "step": 3326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413564781675018e-05,
+      "loss": 0.3301,
+      "step": 3327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413385826771654e-05,
+      "loss": 0.3693,
+      "step": 3328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4132068718682895e-05,
+      "loss": 0.4215,
+      "step": 3329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.413027916964925e-05,
+      "loss": 0.2924,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.412848962061561e-05,
+      "loss": 0.4371,
+      "step": 3331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.412670007158196e-05,
+      "loss": 0.4046,
+      "step": 3332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4124910522548316e-05,
+      "loss": 0.3894,
+      "step": 3333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.412312097351468e-05,
+      "loss": 0.3776,
+      "step": 3334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.412133142448103e-05,
+      "loss": 0.4472,
+      "step": 3335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411954187544739e-05,
+      "loss": 0.4298,
+      "step": 3336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411775232641375e-05,
+      "loss": 0.3937,
+      "step": 3337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4115962777380105e-05,
+      "loss": 0.3235,
+      "step": 3338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411417322834646e-05,
+      "loss": 0.4711,
+      "step": 3339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4112383679312815e-05,
+      "loss": 0.3668,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411059413027917e-05,
+      "loss": 0.3314,
+      "step": 3341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4108804581245526e-05,
+      "loss": 0.3464,
+      "step": 3342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.410701503221189e-05,
+      "loss": 0.3539,
+      "step": 3343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.410522548317824e-05,
+      "loss": 0.3042,
+      "step": 3344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.41034359341446e-05,
+      "loss": 0.3668,
+      "step": 3345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4101646385110953e-05,
+      "loss": 0.3819,
+      "step": 3346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.409985683607731e-05,
+      "loss": 0.3834,
+      "step": 3347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4098067287043664e-05,
+      "loss": 0.3008,
+      "step": 3348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.409627773801002e-05,
+      "loss": 0.4025,
+      "step": 3349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.409448818897638e-05,
+      "loss": 0.3857,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4092698639942736e-05,
+      "loss": 0.4552,
+      "step": 3351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.409090909090909e-05,
+      "loss": 0.3607,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.408911954187545e-05,
+      "loss": 0.3001,
+      "step": 3353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.408732999284181e-05,
+      "loss": 0.402,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.408554044380816e-05,
+      "loss": 0.4164,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.408375089477452e-05,
+      "loss": 0.3348,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4081961345740874e-05,
+      "loss": 0.4106,
+      "step": 3357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.408017179670723e-05,
+      "loss": 0.3416,
+      "step": 3358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.407838224767359e-05,
+      "loss": 0.4249,
+      "step": 3359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4076592698639946e-05,
+      "loss": 0.397,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.40748031496063e-05,
+      "loss": 0.3409,
+      "step": 3361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.407301360057266e-05,
+      "loss": 0.3807,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.407122405153901e-05,
+      "loss": 0.3209,
+      "step": 3363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.406943450250537e-05,
+      "loss": 0.3198,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.406764495347173e-05,
+      "loss": 0.46,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4065855404438084e-05,
+      "loss": 0.3581,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.406406585540444e-05,
+      "loss": 0.2337,
+      "step": 3367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4062276306370795e-05,
+      "loss": 0.3347,
+      "step": 3368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.406048675733716e-05,
+      "loss": 0.3562,
+      "step": 3369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.405869720830351e-05,
+      "loss": 0.4214,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.405690765926987e-05,
+      "loss": 0.4481,
+      "step": 3371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.405511811023622e-05,
+      "loss": 0.2928,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.405332856120258e-05,
+      "loss": 0.4791,
+      "step": 3373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.405153901216893e-05,
+      "loss": 0.3303,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4049749463135295e-05,
+      "loss": 0.3594,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.404795991410165e-05,
+      "loss": 0.3369,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4046170365068005e-05,
+      "loss": 0.3238,
+      "step": 3377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.404438081603437e-05,
+      "loss": 0.4856,
+      "step": 3378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4042591267000715e-05,
+      "loss": 0.3913,
+      "step": 3379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.404080171796707e-05,
+      "loss": 0.3449,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.403901216893343e-05,
+      "loss": 0.2847,
+      "step": 3381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.403722261989979e-05,
+      "loss": 0.2777,
+      "step": 3382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.403543307086614e-05,
+      "loss": 0.4309,
+      "step": 3383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4033643521832505e-05,
+      "loss": 0.4142,
+      "step": 3384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.403185397279886e-05,
+      "loss": 0.4877,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4030064423765215e-05,
+      "loss": 0.321,
+      "step": 3386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.402827487473157e-05,
+      "loss": 0.4896,
+      "step": 3387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4026485325697926e-05,
+      "loss": 0.3061,
+      "step": 3388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.402469577666428e-05,
+      "loss": 0.3741,
+      "step": 3389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4022906227630636e-05,
+      "loss": 0.3748,
+      "step": 3390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4021116678597e-05,
+      "loss": 0.2101,
+      "step": 3391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.401932712956335e-05,
+      "loss": 0.3731,
+      "step": 3392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.401753758052971e-05,
+      "loss": 0.3759,
+      "step": 3393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4015748031496064e-05,
+      "loss": 0.3097,
+      "step": 3394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.401395848246242e-05,
+      "loss": 0.2845,
+      "step": 3395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4012168933428774e-05,
+      "loss": 0.2406,
+      "step": 3396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4010379384395136e-05,
+      "loss": 0.368,
+      "step": 3397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.400858983536149e-05,
+      "loss": 0.4426,
+      "step": 3398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4006800286327846e-05,
+      "loss": 0.4543,
+      "step": 3399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.400501073729421e-05,
+      "loss": 0.4881,
+      "step": 3400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4003221188260564e-05,
+      "loss": 0.3372,
+      "step": 3401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.400143163922691e-05,
+      "loss": 0.475,
+      "step": 3402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3999642090193274e-05,
+      "loss": 0.3784,
+      "step": 3403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.399785254115963e-05,
+      "loss": 0.4061,
+      "step": 3404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3996062992125984e-05,
+      "loss": 0.2642,
+      "step": 3405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3994273443092346e-05,
+      "loss": 0.345,
+      "step": 3406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.39924838940587e-05,
+      "loss": 0.2686,
+      "step": 3407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.399069434502506e-05,
+      "loss": 0.2889,
+      "step": 3408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.398890479599141e-05,
+      "loss": 0.4457,
+      "step": 3409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.398711524695777e-05,
+      "loss": 0.3878,
+      "step": 3410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.398532569792412e-05,
+      "loss": 0.5074,
+      "step": 3411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.398353614889048e-05,
+      "loss": 0.3329,
+      "step": 3412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.398174659985684e-05,
+      "loss": 0.4203,
+      "step": 3413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3979957050823195e-05,
+      "loss": 0.416,
+      "step": 3414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.397816750178955e-05,
+      "loss": 0.2899,
+      "step": 3415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.397637795275591e-05,
+      "loss": 0.3782,
+      "step": 3416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.397458840372227e-05,
+      "loss": 0.4299,
+      "step": 3417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3972798854688615e-05,
+      "loss": 0.2652,
+      "step": 3418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.397100930565498e-05,
+      "loss": 0.5363,
+      "step": 3419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.396921975662133e-05,
+      "loss": 0.3502,
+      "step": 3420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.396743020758769e-05,
+      "loss": 0.5802,
+      "step": 3421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.396564065855405e-05,
+      "loss": 0.2968,
+      "step": 3422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3963851109520405e-05,
+      "loss": 0.4929,
+      "step": 3423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.396206156048676e-05,
+      "loss": 0.3806,
+      "step": 3424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3960272011453115e-05,
+      "loss": 0.377,
+      "step": 3425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.395848246241947e-05,
+      "loss": 0.3119,
+      "step": 3426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3956692913385826e-05,
+      "loss": 0.2666,
+      "step": 3427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.395490336435219e-05,
+      "loss": 0.3421,
+      "step": 3428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.395311381531854e-05,
+      "loss": 0.3239,
+      "step": 3429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.39513242662849e-05,
+      "loss": 0.4529,
+      "step": 3430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.394953471725125e-05,
+      "loss": 0.3318,
+      "step": 3431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3947745168217615e-05,
+      "loss": 0.2411,
+      "step": 3432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3945955619183964e-05,
+      "loss": 0.4119,
+      "step": 3433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.394416607015032e-05,
+      "loss": 0.3973,
+      "step": 3434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.394237652111668e-05,
+      "loss": 0.3015,
+      "step": 3435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3940586972083036e-05,
+      "loss": 0.4754,
+      "step": 3436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.393879742304939e-05,
+      "loss": 0.3647,
+      "step": 3437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.393700787401575e-05,
+      "loss": 0.4692,
+      "step": 3438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.393521832498211e-05,
+      "loss": 0.4631,
+      "step": 3439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3933428775948464e-05,
+      "loss": 0.3632,
+      "step": 3440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.393163922691482e-05,
+      "loss": 0.3699,
+      "step": 3441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3929849677881174e-05,
+      "loss": 0.2836,
+      "step": 3442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.392806012884753e-05,
+      "loss": 0.2863,
+      "step": 3443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.392627057981389e-05,
+      "loss": 0.3703,
+      "step": 3444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3924481030780246e-05,
+      "loss": 0.3238,
+      "step": 3445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.39226914817466e-05,
+      "loss": 0.2446,
+      "step": 3446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3920901932712963e-05,
+      "loss": 0.3663,
+      "step": 3447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.391911238367932e-05,
+      "loss": 0.4331,
+      "step": 3448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.391732283464567e-05,
+      "loss": 0.2957,
+      "step": 3449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.391553328561203e-05,
+      "loss": 0.4963,
+      "step": 3450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3913743736578384e-05,
+      "loss": 0.3257,
+      "step": 3451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.391195418754474e-05,
+      "loss": 0.3606,
+      "step": 3452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3910164638511095e-05,
+      "loss": 0.2748,
+      "step": 3453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3908375089477457e-05,
+      "loss": 0.3939,
+      "step": 3454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.390658554044381e-05,
+      "loss": 0.3473,
+      "step": 3455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.390479599141017e-05,
+      "loss": 0.422,
+      "step": 3456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.390300644237652e-05,
+      "loss": 0.459,
+      "step": 3457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.390121689334288e-05,
+      "loss": 0.1929,
+      "step": 3458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.389942734430923e-05,
+      "loss": 0.2798,
+      "step": 3459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3897637795275594e-05,
+      "loss": 0.2543,
+      "step": 3460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.389584824624195e-05,
+      "loss": 0.3907,
+      "step": 3461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3894058697208305e-05,
+      "loss": 0.4113,
+      "step": 3462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.389226914817467e-05,
+      "loss": 0.3418,
+      "step": 3463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.389047959914102e-05,
+      "loss": 0.192,
+      "step": 3464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.388869005010737e-05,
+      "loss": 0.5695,
+      "step": 3465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.388690050107373e-05,
+      "loss": 0.3689,
+      "step": 3466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.388511095204009e-05,
+      "loss": 0.3127,
+      "step": 3467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.388332140300644e-05,
+      "loss": 0.2684,
+      "step": 3468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3881531853972805e-05,
+      "loss": 0.3806,
+      "step": 3469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.387974230493916e-05,
+      "loss": 0.4116,
+      "step": 3470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3877952755905515e-05,
+      "loss": 0.3608,
+      "step": 3471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.387616320687187e-05,
+      "loss": 0.3701,
+      "step": 3472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3874373657838226e-05,
+      "loss": 0.3661,
+      "step": 3473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.387258410880458e-05,
+      "loss": 0.332,
+      "step": 3474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3870794559770936e-05,
+      "loss": 0.3827,
+      "step": 3475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.38690050107373e-05,
+      "loss": 0.4209,
+      "step": 3476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.386721546170365e-05,
+      "loss": 0.5157,
+      "step": 3477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.386542591267001e-05,
+      "loss": 0.2998,
+      "step": 3478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.386363636363637e-05,
+      "loss": 0.2562,
+      "step": 3479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.386184681460272e-05,
+      "loss": 0.385,
+      "step": 3480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3860057265569074e-05,
+      "loss": 0.4799,
+      "step": 3481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3858267716535436e-05,
+      "loss": 0.3934,
+      "step": 3482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.385647816750179e-05,
+      "loss": 0.306,
+      "step": 3483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3854688618468146e-05,
+      "loss": 0.3939,
+      "step": 3484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.385289906943451e-05,
+      "loss": 0.3761,
+      "step": 3485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3851109520400863e-05,
+      "loss": 0.3775,
+      "step": 3486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.384931997136722e-05,
+      "loss": 0.3314,
+      "step": 3487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3847530422333574e-05,
+      "loss": 0.3135,
+      "step": 3488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.384574087329993e-05,
+      "loss": 0.3162,
+      "step": 3489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3843951324266284e-05,
+      "loss": 0.3889,
+      "step": 3490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3842161775232646e-05,
+      "loss": 0.4129,
+      "step": 3491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3840372226199e-05,
+      "loss": 0.3047,
+      "step": 3492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3838582677165357e-05,
+      "loss": 0.3118,
+      "step": 3493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.383679312813171e-05,
+      "loss": 0.3276,
+      "step": 3494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3835003579098074e-05,
+      "loss": 0.4627,
+      "step": 3495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.383321403006442e-05,
+      "loss": 0.2379,
+      "step": 3496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.383142448103078e-05,
+      "loss": 0.4746,
+      "step": 3497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.382963493199714e-05,
+      "loss": 0.367,
+      "step": 3498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3827845382963494e-05,
+      "loss": 0.4866,
+      "step": 3499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.382605583392985e-05,
+      "loss": 0.2926,
+      "step": 3500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.382426628489621e-05,
+      "loss": 0.3005,
+      "step": 3501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.382247673586257e-05,
+      "loss": 0.3975,
+      "step": 3502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.382068718682892e-05,
+      "loss": 0.3057,
+      "step": 3503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381889763779528e-05,
+      "loss": 0.2659,
+      "step": 3504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381710808876163e-05,
+      "loss": 0.3436,
+      "step": 3505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381531853972799e-05,
+      "loss": 0.323,
+      "step": 3506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381352899069435e-05,
+      "loss": 0.3033,
+      "step": 3507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3811739441660705e-05,
+      "loss": 0.3161,
+      "step": 3508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380994989262706e-05,
+      "loss": 0.3454,
+      "step": 3509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3808160343593415e-05,
+      "loss": 0.4213,
+      "step": 3510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380637079455978e-05,
+      "loss": 0.4104,
+      "step": 3511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3804581245526126e-05,
+      "loss": 0.4082,
+      "step": 3512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380279169649249e-05,
+      "loss": 0.2427,
+      "step": 3513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380100214745884e-05,
+      "loss": 0.4736,
+      "step": 3514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.37992125984252e-05,
+      "loss": 0.2297,
+      "step": 3515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.379742304939155e-05,
+      "loss": 0.3483,
+      "step": 3516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3795633500357915e-05,
+      "loss": 0.3538,
+      "step": 3517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.379384395132427e-05,
+      "loss": 0.3219,
+      "step": 3518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3792054402290625e-05,
+      "loss": 0.5032,
+      "step": 3519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.379026485325698e-05,
+      "loss": 0.3454,
+      "step": 3520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3788475304223336e-05,
+      "loss": 0.4457,
+      "step": 3521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.378668575518969e-05,
+      "loss": 0.2442,
+      "step": 3522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.378489620615605e-05,
+      "loss": 0.4372,
+      "step": 3523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.378310665712241e-05,
+      "loss": 0.4359,
+      "step": 3524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.378131710808876e-05,
+      "loss": 0.3444,
+      "step": 3525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3779527559055125e-05,
+      "loss": 0.4926,
+      "step": 3526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3777738010021474e-05,
+      "loss": 0.4067,
+      "step": 3527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.377594846098783e-05,
+      "loss": 0.3862,
+      "step": 3528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.377415891195419e-05,
+      "loss": 0.404,
+      "step": 3529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3772369362920546e-05,
+      "loss": 0.4032,
+      "step": 3530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.37705798138869e-05,
+      "loss": 0.3344,
+      "step": 3531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.376879026485326e-05,
+      "loss": 0.4369,
+      "step": 3532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.376700071581962e-05,
+      "loss": 0.3304,
+      "step": 3533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3765211166785974e-05,
+      "loss": 0.3304,
+      "step": 3534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.376342161775233e-05,
+      "loss": 0.3441,
+      "step": 3535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3761632068718684e-05,
+      "loss": 0.3207,
+      "step": 3536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.375984251968504e-05,
+      "loss": 0.467,
+      "step": 3537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3758052970651394e-05,
+      "loss": 0.4068,
+      "step": 3538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3756263421617756e-05,
+      "loss": 0.3764,
+      "step": 3539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.375447387258411e-05,
+      "loss": 0.329,
+      "step": 3540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.375268432355047e-05,
+      "loss": 0.2586,
+      "step": 3541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.375089477451683e-05,
+      "loss": 0.4112,
+      "step": 3542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.374910522548318e-05,
+      "loss": 0.4682,
+      "step": 3543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.374731567644953e-05,
+      "loss": 0.3761,
+      "step": 3544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3745526127415894e-05,
+      "loss": 0.3669,
+      "step": 3545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.374373657838225e-05,
+      "loss": 0.3707,
+      "step": 3546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3741947029348605e-05,
+      "loss": 0.4676,
+      "step": 3547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.374015748031497e-05,
+      "loss": 0.3428,
+      "step": 3548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373836793128132e-05,
+      "loss": 0.5015,
+      "step": 3549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373657838224768e-05,
+      "loss": 0.2792,
+      "step": 3550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373478883321403e-05,
+      "loss": 0.38,
+      "step": 3551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373299928418039e-05,
+      "loss": 0.3461,
+      "step": 3552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.373120973514674e-05,
+      "loss": 0.3632,
+      "step": 3553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3729420186113105e-05,
+      "loss": 0.4488,
+      "step": 3554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372763063707946e-05,
+      "loss": 0.3027,
+      "step": 3555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3725841088045815e-05,
+      "loss": 0.2857,
+      "step": 3556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372405153901217e-05,
+      "loss": 0.3419,
+      "step": 3557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372226198997853e-05,
+      "loss": 0.316,
+      "step": 3558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372047244094488e-05,
+      "loss": 0.377,
+      "step": 3559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3718682891911236e-05,
+      "loss": 0.3485,
+      "step": 3560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.37168933428776e-05,
+      "loss": 0.2846,
+      "step": 3561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.371510379384395e-05,
+      "loss": 0.3672,
+      "step": 3562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.371331424481031e-05,
+      "loss": 0.2869,
+      "step": 3563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.371152469577667e-05,
+      "loss": 0.428,
+      "step": 3564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3709735146743025e-05,
+      "loss": 0.4994,
+      "step": 3565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3707945597709374e-05,
+      "loss": 0.3266,
+      "step": 3566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3706156048675736e-05,
+      "loss": 0.2416,
+      "step": 3567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.370436649964209e-05,
+      "loss": 0.3305,
+      "step": 3568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3702576950608446e-05,
+      "loss": 0.4448,
+      "step": 3569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.370078740157481e-05,
+      "loss": 0.2897,
+      "step": 3570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.369899785254116e-05,
+      "loss": 0.3659,
+      "step": 3571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.369720830350752e-05,
+      "loss": 0.2814,
+      "step": 3572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3695418754473874e-05,
+      "loss": 0.4439,
+      "step": 3573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.369362920544023e-05,
+      "loss": 0.3276,
+      "step": 3574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3691839656406584e-05,
+      "loss": 0.4191,
+      "step": 3575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3690050107372946e-05,
+      "loss": 0.4273,
+      "step": 3576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.36882605583393e-05,
+      "loss": 0.3582,
+      "step": 3577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3686471009305656e-05,
+      "loss": 0.4284,
+      "step": 3578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.368468146027201e-05,
+      "loss": 0.4393,
+      "step": 3579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3682891911238374e-05,
+      "loss": 0.2903,
+      "step": 3580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.368110236220473e-05,
+      "loss": 0.2674,
+      "step": 3581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367931281317108e-05,
+      "loss": 0.4741,
+      "step": 3582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367752326413744e-05,
+      "loss": 0.3493,
+      "step": 3583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3675733715103794e-05,
+      "loss": 0.3699,
+      "step": 3584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367394416607015e-05,
+      "loss": 0.2813,
+      "step": 3585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367215461703651e-05,
+      "loss": 0.3315,
+      "step": 3586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367036506800287e-05,
+      "loss": 0.4439,
+      "step": 3587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.366857551896922e-05,
+      "loss": 0.3361,
+      "step": 3588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3666785969935584e-05,
+      "loss": 0.2174,
+      "step": 3589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.366499642090193e-05,
+      "loss": 0.2601,
+      "step": 3590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.366320687186829e-05,
+      "loss": 0.2617,
+      "step": 3591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.366141732283465e-05,
+      "loss": 0.3382,
+      "step": 3592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3659627773801005e-05,
+      "loss": 0.3928,
+      "step": 3593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.365783822476736e-05,
+      "loss": 0.281,
+      "step": 3594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.365604867573372e-05,
+      "loss": 0.4636,
+      "step": 3595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.365425912670008e-05,
+      "loss": 0.2692,
+      "step": 3596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.365246957766643e-05,
+      "loss": 0.2568,
+      "step": 3597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.365068002863279e-05,
+      "loss": 0.4804,
+      "step": 3598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.364889047959914e-05,
+      "loss": 0.2482,
+      "step": 3599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.36471009305655e-05,
+      "loss": 0.212,
+      "step": 3600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.364531138153185e-05,
+      "loss": 0.2555,
+      "step": 3601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3643521832498215e-05,
+      "loss": 0.3787,
+      "step": 3602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.364173228346457e-05,
+      "loss": 0.2864,
+      "step": 3603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3639942734430925e-05,
+      "loss": 0.3867,
+      "step": 3604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.363815318539729e-05,
+      "loss": 0.444,
+      "step": 3605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 0.4981,
+      "step": 3606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.363457408732999e-05,
+      "loss": 0.4748,
+      "step": 3607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.363278453829635e-05,
+      "loss": 0.473,
+      "step": 3608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.363099498926271e-05,
+      "loss": 0.2339,
+      "step": 3609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.362920544022906e-05,
+      "loss": 0.2642,
+      "step": 3610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3627415891195425e-05,
+      "loss": 0.4308,
+      "step": 3611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.362562634216178e-05,
+      "loss": 0.3165,
+      "step": 3612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.362383679312813e-05,
+      "loss": 0.2772,
+      "step": 3613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.362204724409449e-05,
+      "loss": 0.4333,
+      "step": 3614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3620257695060846e-05,
+      "loss": 0.3413,
+      "step": 3615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.36184681460272e-05,
+      "loss": 0.3738,
+      "step": 3616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361667859699356e-05,
+      "loss": 0.2517,
+      "step": 3617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361488904795992e-05,
+      "loss": 0.3139,
+      "step": 3618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3613099498926273e-05,
+      "loss": 0.2772,
+      "step": 3619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361130994989263e-05,
+      "loss": 0.3857,
+      "step": 3620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3609520400858984e-05,
+      "loss": 0.4156,
+      "step": 3621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.360773085182534e-05,
+      "loss": 0.2628,
+      "step": 3622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3605941302791694e-05,
+      "loss": 0.3134,
+      "step": 3623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3604151753758056e-05,
+      "loss": 0.5779,
+      "step": 3624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.360236220472441e-05,
+      "loss": 0.4362,
+      "step": 3625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3600572655690767e-05,
+      "loss": 0.2834,
+      "step": 3626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.359878310665713e-05,
+      "loss": 0.436,
+      "step": 3627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3596993557623484e-05,
+      "loss": 0.4509,
+      "step": 3628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.359520400858983e-05,
+      "loss": 0.336,
+      "step": 3629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3593414459556194e-05,
+      "loss": 0.4204,
+      "step": 3630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.359162491052255e-05,
+      "loss": 0.3537,
+      "step": 3631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3589835361488905e-05,
+      "loss": 0.3029,
+      "step": 3632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3588045812455266e-05,
+      "loss": 0.3625,
+      "step": 3633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.358625626342162e-05,
+      "loss": 0.3258,
+      "step": 3634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.358446671438798e-05,
+      "loss": 0.3678,
+      "step": 3635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.358267716535433e-05,
+      "loss": 0.3745,
+      "step": 3636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.358088761632069e-05,
+      "loss": 0.4724,
+      "step": 3637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.357909806728704e-05,
+      "loss": 0.4163,
+      "step": 3638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3577308518253404e-05,
+      "loss": 0.3331,
+      "step": 3639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.357551896921976e-05,
+      "loss": 0.3637,
+      "step": 3640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3573729420186115e-05,
+      "loss": 0.2531,
+      "step": 3641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.357193987115247e-05,
+      "loss": 0.295,
+      "step": 3642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.357015032211883e-05,
+      "loss": 0.4675,
+      "step": 3643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.356836077308519e-05,
+      "loss": 0.3004,
+      "step": 3644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3566571224051536e-05,
+      "loss": 0.3678,
+      "step": 3645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.35647816750179e-05,
+      "loss": 0.3407,
+      "step": 3646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.356299212598425e-05,
+      "loss": 0.4026,
+      "step": 3647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.356120257695061e-05,
+      "loss": 0.3464,
+      "step": 3648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.355941302791697e-05,
+      "loss": 0.3691,
+      "step": 3649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3557623478883325e-05,
+      "loss": 0.3116,
+      "step": 3650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.355583392984968e-05,
+      "loss": 0.3317,
+      "step": 3651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.355404438081604e-05,
+      "loss": 0.4969,
+      "step": 3652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.355225483178239e-05,
+      "loss": 0.3657,
+      "step": 3653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3550465282748746e-05,
+      "loss": 0.4241,
+      "step": 3654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.354867573371511e-05,
+      "loss": 0.3786,
+      "step": 3655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.354688618468146e-05,
+      "loss": 0.3609,
+      "step": 3656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.354509663564782e-05,
+      "loss": 0.4279,
+      "step": 3657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3543307086614173e-05,
+      "loss": 0.3869,
+      "step": 3658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3541517537580535e-05,
+      "loss": 0.3485,
+      "step": 3659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3539727988546884e-05,
+      "loss": 0.359,
+      "step": 3660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3537938439513246e-05,
+      "loss": 0.4101,
+      "step": 3661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.35361488904796e-05,
+      "loss": 0.4734,
+      "step": 3662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3534359341445956e-05,
+      "loss": 0.3456,
+      "step": 3663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.353256979241231e-05,
+      "loss": 0.3354,
+      "step": 3664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.353078024337867e-05,
+      "loss": 0.4877,
+      "step": 3665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.352899069434503e-05,
+      "loss": 0.3299,
+      "step": 3666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3527201145311384e-05,
+      "loss": 0.3319,
+      "step": 3667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.352541159627774e-05,
+      "loss": 0.3997,
+      "step": 3668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3523622047244094e-05,
+      "loss": 0.3775,
+      "step": 3669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.352183249821045e-05,
+      "loss": 0.3576,
+      "step": 3670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.352004294917681e-05,
+      "loss": 0.3317,
+      "step": 3671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3518253400143166e-05,
+      "loss": 0.4407,
+      "step": 3672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.351646385110952e-05,
+      "loss": 0.3944,
+      "step": 3673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3514674302075884e-05,
+      "loss": 0.356,
+      "step": 3674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.351288475304224e-05,
+      "loss": 0.2758,
+      "step": 3675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.351109520400859e-05,
+      "loss": 0.3188,
+      "step": 3676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350930565497495e-05,
+      "loss": 0.552,
+      "step": 3677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3507516105941304e-05,
+      "loss": 0.4037,
+      "step": 3678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350572655690766e-05,
+      "loss": 0.3285,
+      "step": 3679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350393700787402e-05,
+      "loss": 0.3106,
+      "step": 3680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350214745884038e-05,
+      "loss": 0.4369,
+      "step": 3681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350035790980673e-05,
+      "loss": 0.2426,
+      "step": 3682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.349856836077309e-05,
+      "loss": 0.3509,
+      "step": 3683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.349677881173944e-05,
+      "loss": 0.3467,
+      "step": 3684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.34949892627058e-05,
+      "loss": 0.3844,
+      "step": 3685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.349319971367215e-05,
+      "loss": 0.2824,
+      "step": 3686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3491410164638515e-05,
+      "loss": 0.2818,
+      "step": 3687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.348962061560487e-05,
+      "loss": 0.4231,
+      "step": 3688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3487831066571225e-05,
+      "loss": 0.3451,
+      "step": 3689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.348604151753759e-05,
+      "loss": 0.4126,
+      "step": 3690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.348425196850394e-05,
+      "loss": 0.2436,
+      "step": 3691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.348246241947029e-05,
+      "loss": 0.3116,
+      "step": 3692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.348067287043665e-05,
+      "loss": 0.4009,
+      "step": 3693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.347888332140301e-05,
+      "loss": 0.1885,
+      "step": 3694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.347709377236936e-05,
+      "loss": 0.5115,
+      "step": 3695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3475304223335725e-05,
+      "loss": 0.4593,
+      "step": 3696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.347351467430208e-05,
+      "loss": 0.4119,
+      "step": 3697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3471725125268435e-05,
+      "loss": 0.3171,
+      "step": 3698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.346993557623479e-05,
+      "loss": 0.4509,
+      "step": 3699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3468146027201146e-05,
+      "loss": 0.32,
+      "step": 3700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.34663564781675e-05,
+      "loss": 0.4864,
+      "step": 3701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.346456692913386e-05,
+      "loss": 0.4233,
+      "step": 3702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.346277738010022e-05,
+      "loss": 0.339,
+      "step": 3703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.346098783106657e-05,
+      "loss": 0.4466,
+      "step": 3704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.345919828203293e-05,
+      "loss": 0.4218,
+      "step": 3705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.345740873299929e-05,
+      "loss": 0.3237,
+      "step": 3706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.345561918396564e-05,
+      "loss": 0.4454,
+      "step": 3707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3453829634931994e-05,
+      "loss": 0.2856,
+      "step": 3708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3452040085898356e-05,
+      "loss": 0.3645,
+      "step": 3709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.345025053686471e-05,
+      "loss": 0.4938,
+      "step": 3710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3448460987831066e-05,
+      "loss": 0.365,
+      "step": 3711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.344667143879743e-05,
+      "loss": 0.3547,
+      "step": 3712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3444881889763784e-05,
+      "loss": 0.2553,
+      "step": 3713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.344309234073014e-05,
+      "loss": 0.3437,
+      "step": 3714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3441302791696494e-05,
+      "loss": 0.2862,
+      "step": 3715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.343951324266285e-05,
+      "loss": 0.3345,
+      "step": 3716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3437723693629204e-05,
+      "loss": 0.3661,
+      "step": 3717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3435934144595566e-05,
+      "loss": 0.3771,
+      "step": 3718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.343414459556192e-05,
+      "loss": 0.3373,
+      "step": 3719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.343235504652828e-05,
+      "loss": 0.3995,
+      "step": 3720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.343056549749463e-05,
+      "loss": 0.3,
+      "step": 3721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3428775948460994e-05,
+      "loss": 0.3857,
+      "step": 3722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.342698639942734e-05,
+      "loss": 0.3716,
+      "step": 3723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3425196850393704e-05,
+      "loss": 0.331,
+      "step": 3724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.342340730136006e-05,
+      "loss": 0.3854,
+      "step": 3725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3421617752326415e-05,
+      "loss": 0.3083,
+      "step": 3726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341982820329277e-05,
+      "loss": 0.3654,
+      "step": 3727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341803865425913e-05,
+      "loss": 0.245,
+      "step": 3728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341624910522549e-05,
+      "loss": 0.4008,
+      "step": 3729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341445955619184e-05,
+      "loss": 0.3402,
+      "step": 3730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.34126700071582e-05,
+      "loss": 0.3839,
+      "step": 3731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341088045812455e-05,
+      "loss": 0.4699,
+      "step": 3732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340909090909091e-05,
+      "loss": 0.355,
+      "step": 3733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340730136005727e-05,
+      "loss": 0.3212,
+      "step": 3734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3405511811023625e-05,
+      "loss": 0.3778,
+      "step": 3735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340372226198998e-05,
+      "loss": 0.362,
+      "step": 3736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340193271295634e-05,
+      "loss": 0.4723,
+      "step": 3737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.34001431639227e-05,
+      "loss": 0.3412,
+      "step": 3738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3398353614889046e-05,
+      "loss": 0.2936,
+      "step": 3739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.339656406585541e-05,
+      "loss": 0.3011,
+      "step": 3740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.339477451682176e-05,
+      "loss": 0.2607,
+      "step": 3741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.339298496778812e-05,
+      "loss": 0.3707,
+      "step": 3742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.339119541875448e-05,
+      "loss": 0.2923,
+      "step": 3743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3389405869720835e-05,
+      "loss": 0.3724,
+      "step": 3744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.338761632068719e-05,
+      "loss": 0.3297,
+      "step": 3745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3385826771653546e-05,
+      "loss": 0.3238,
+      "step": 3746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.33840372226199e-05,
+      "loss": 0.3341,
+      "step": 3747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3382247673586256e-05,
+      "loss": 0.2562,
+      "step": 3748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.338045812455261e-05,
+      "loss": 0.336,
+      "step": 3749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.337866857551897e-05,
+      "loss": 0.2405,
+      "step": 3750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.337687902648533e-05,
+      "loss": 0.2684,
+      "step": 3751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3375089477451684e-05,
+      "loss": 0.5003,
+      "step": 3752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3373299928418045e-05,
+      "loss": 0.3038,
+      "step": 3753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3371510379384394e-05,
+      "loss": 0.4137,
+      "step": 3754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.336972083035075e-05,
+      "loss": 0.3423,
+      "step": 3755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.336793128131711e-05,
+      "loss": 0.3619,
+      "step": 3756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3366141732283466e-05,
+      "loss": 0.3185,
+      "step": 3757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.336435218324982e-05,
+      "loss": 0.4266,
+      "step": 3758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3362562634216183e-05,
+      "loss": 0.2961,
+      "step": 3759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.336077308518254e-05,
+      "loss": 0.2407,
+      "step": 3760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3358983536148894e-05,
+      "loss": 0.397,
+      "step": 3761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.335719398711525e-05,
+      "loss": 0.2822,
+      "step": 3762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3355404438081604e-05,
+      "loss": 0.3529,
+      "step": 3763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.335361488904796e-05,
+      "loss": 0.3058,
+      "step": 3764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.335182534001432e-05,
+      "loss": 0.2192,
+      "step": 3765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3350035790980677e-05,
+      "loss": 0.4413,
+      "step": 3766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.334824624194703e-05,
+      "loss": 0.2954,
+      "step": 3767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.334645669291339e-05,
+      "loss": 0.547,
+      "step": 3768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.334466714387975e-05,
+      "loss": 0.5653,
+      "step": 3769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.33428775948461e-05,
+      "loss": 0.3552,
+      "step": 3770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.334108804581245e-05,
+      "loss": 0.2637,
+      "step": 3771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3339298496778814e-05,
+      "loss": 0.3083,
+      "step": 3772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.333750894774517e-05,
+      "loss": 0.3762,
+      "step": 3773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3335719398711525e-05,
+      "loss": 0.2958,
+      "step": 3774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.333392984967789e-05,
+      "loss": 0.3209,
+      "step": 3775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.333214030064424e-05,
+      "loss": 0.2193,
+      "step": 3776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.33303507516106e-05,
+      "loss": 0.3804,
+      "step": 3777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.332856120257695e-05,
+      "loss": 0.4106,
+      "step": 3778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332677165354331e-05,
+      "loss": 0.4037,
+      "step": 3779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332498210450966e-05,
+      "loss": 0.3601,
+      "step": 3780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3323192555476025e-05,
+      "loss": 0.3607,
+      "step": 3781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332140300644238e-05,
+      "loss": 0.2305,
+      "step": 3782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3319613457408735e-05,
+      "loss": 0.3978,
+      "step": 3783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331782390837509e-05,
+      "loss": 0.2803,
+      "step": 3784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3316034359341446e-05,
+      "loss": 0.2832,
+      "step": 3785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.33142448103078e-05,
+      "loss": 0.2431,
+      "step": 3786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331245526127416e-05,
+      "loss": 0.267,
+      "step": 3787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331066571224052e-05,
+      "loss": 0.2964,
+      "step": 3788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.330887616320687e-05,
+      "loss": 0.4098,
+      "step": 3789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.330708661417323e-05,
+      "loss": 0.35,
+      "step": 3790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.330529706513959e-05,
+      "loss": 0.2501,
+      "step": 3791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3303507516105945e-05,
+      "loss": 0.2491,
+      "step": 3792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3301717967072294e-05,
+      "loss": 0.2941,
+      "step": 3793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3299928418038656e-05,
+      "loss": 0.3858,
+      "step": 3794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.329813886900501e-05,
+      "loss": 0.4254,
+      "step": 3795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3296349319971366e-05,
+      "loss": 0.3483,
+      "step": 3796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.329455977093773e-05,
+      "loss": 0.4291,
+      "step": 3797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.329277022190408e-05,
+      "loss": 0.264,
+      "step": 3798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.329098067287044e-05,
+      "loss": 0.2987,
+      "step": 3799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.32891911238368e-05,
+      "loss": 0.3457,
+      "step": 3800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.328740157480315e-05,
+      "loss": 0.3658,
+      "step": 3801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3285612025769504e-05,
+      "loss": 0.3657,
+      "step": 3802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3283822476735866e-05,
+      "loss": 0.226,
+      "step": 3803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.328203292770222e-05,
+      "loss": 0.1575,
+      "step": 3804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3280243378668577e-05,
+      "loss": 0.4497,
+      "step": 3805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.327845382963494e-05,
+      "loss": 0.3929,
+      "step": 3806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3276664280601294e-05,
+      "loss": 0.5594,
+      "step": 3807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.327487473156765e-05,
+      "loss": 0.4157,
+      "step": 3808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3273085182534004e-05,
+      "loss": 0.3094,
+      "step": 3809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.327129563350036e-05,
+      "loss": 0.382,
+      "step": 3810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3269506084466714e-05,
+      "loss": 0.388,
+      "step": 3811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.326771653543307e-05,
+      "loss": 0.2972,
+      "step": 3812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.326592698639943e-05,
+      "loss": 0.3499,
+      "step": 3813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.326413743736579e-05,
+      "loss": 0.4102,
+      "step": 3814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.326234788833214e-05,
+      "loss": 0.3443,
+      "step": 3815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3260558339298504e-05,
+      "loss": 0.4117,
+      "step": 3816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.325876879026485e-05,
+      "loss": 0.271,
+      "step": 3817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.325697924123121e-05,
+      "loss": 0.2913,
+      "step": 3818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.325518969219757e-05,
+      "loss": 0.3926,
+      "step": 3819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3253400143163925e-05,
+      "loss": 0.4627,
+      "step": 3820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.325161059413028e-05,
+      "loss": 0.45,
+      "step": 3821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324982104509664e-05,
+      "loss": 0.2876,
+      "step": 3822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3248031496063e-05,
+      "loss": 0.3008,
+      "step": 3823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324624194702935e-05,
+      "loss": 0.3543,
+      "step": 3824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324445239799571e-05,
+      "loss": 0.288,
+      "step": 3825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324266284896206e-05,
+      "loss": 0.3664,
+      "step": 3826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.324087329992842e-05,
+      "loss": 0.3517,
+      "step": 3827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.323908375089478e-05,
+      "loss": 0.3362,
+      "step": 3828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3237294201861135e-05,
+      "loss": 0.3767,
+      "step": 3829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.323550465282749e-05,
+      "loss": 0.2858,
+      "step": 3830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3233715103793845e-05,
+      "loss": 0.3959,
+      "step": 3831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.32319255547602e-05,
+      "loss": 0.2846,
+      "step": 3832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3230136005726556e-05,
+      "loss": 0.3722,
+      "step": 3833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.322834645669291e-05,
+      "loss": 0.3493,
+      "step": 3834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322655690765927e-05,
+      "loss": 0.2558,
+      "step": 3835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322476735862563e-05,
+      "loss": 0.4273,
+      "step": 3836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322297780959198e-05,
+      "loss": 0.2723,
+      "step": 3837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3221188260558345e-05,
+      "loss": 0.4377,
+      "step": 3838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.32193987115247e-05,
+      "loss": 0.4076,
+      "step": 3839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321760916249105e-05,
+      "loss": 0.2891,
+      "step": 3840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321581961345741e-05,
+      "loss": 0.4082,
+      "step": 3841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3214030064423766e-05,
+      "loss": 0.3552,
+      "step": 3842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321224051539012e-05,
+      "loss": 0.3556,
+      "step": 3843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321045096635648e-05,
+      "loss": 0.348,
+      "step": 3844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320866141732284e-05,
+      "loss": 0.4513,
+      "step": 3845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3206871868289194e-05,
+      "loss": 0.3797,
+      "step": 3846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320508231925555e-05,
+      "loss": 0.373,
+      "step": 3847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3203292770221904e-05,
+      "loss": 0.3218,
+      "step": 3848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320150322118826e-05,
+      "loss": 0.3408,
+      "step": 3849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.319971367215462e-05,
+      "loss": 0.2701,
+      "step": 3850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3197924123120976e-05,
+      "loss": 0.3114,
+      "step": 3851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.319613457408733e-05,
+      "loss": 0.3784,
+      "step": 3852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.319434502505369e-05,
+      "loss": 0.2713,
+      "step": 3853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.319255547602005e-05,
+      "loss": 0.3136,
+      "step": 3854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3190765926986404e-05,
+      "loss": 0.3878,
+      "step": 3855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318897637795275e-05,
+      "loss": 0.3361,
+      "step": 3856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3187186828919114e-05,
+      "loss": 0.2766,
+      "step": 3857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318539727988547e-05,
+      "loss": 0.2685,
+      "step": 3858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3183607730851825e-05,
+      "loss": 0.4537,
+      "step": 3859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.3684,
+      "step": 3860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318002863278454e-05,
+      "loss": 0.2897,
+      "step": 3861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31782390837509e-05,
+      "loss": 0.3753,
+      "step": 3862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.317644953471726e-05,
+      "loss": 0.3374,
+      "step": 3863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.317465998568361e-05,
+      "loss": 0.4301,
+      "step": 3864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.317287043664996e-05,
+      "loss": 0.3193,
+      "step": 3865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3171080887616325e-05,
+      "loss": 0.3561,
+      "step": 3866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316929133858268e-05,
+      "loss": 0.2678,
+      "step": 3867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3167501789549035e-05,
+      "loss": 0.2823,
+      "step": 3868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316571224051539e-05,
+      "loss": 0.313,
+      "step": 3869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316392269148175e-05,
+      "loss": 0.2817,
+      "step": 3870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316213314244811e-05,
+      "loss": 0.3391,
+      "step": 3871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316034359341446e-05,
+      "loss": 0.298,
+      "step": 3872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.315855404438082e-05,
+      "loss": 0.3182,
+      "step": 3873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.315676449534717e-05,
+      "loss": 0.5021,
+      "step": 3874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.315497494631353e-05,
+      "loss": 0.2845,
+      "step": 3875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.315318539727989e-05,
+      "loss": 0.3177,
+      "step": 3876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3151395848246245e-05,
+      "loss": 0.2842,
+      "step": 3877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31496062992126e-05,
+      "loss": 0.4161,
+      "step": 3878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3147816750178956e-05,
+      "loss": 0.2717,
+      "step": 3879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314602720114531e-05,
+      "loss": 0.4023,
+      "step": 3880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3144237652111666e-05,
+      "loss": 0.2536,
+      "step": 3881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314244810307803e-05,
+      "loss": 0.338,
+      "step": 3882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314065855404438e-05,
+      "loss": 0.4079,
+      "step": 3883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.313886900501074e-05,
+      "loss": 0.3616,
+      "step": 3884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31370794559771e-05,
+      "loss": 0.3663,
+      "step": 3885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3135289906943456e-05,
+      "loss": 0.369,
+      "step": 3886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3133500357909804e-05,
+      "loss": 0.3732,
+      "step": 3887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3131710808876166e-05,
+      "loss": 0.4234,
+      "step": 3888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.312992125984252e-05,
+      "loss": 0.3582,
+      "step": 3889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3128131710808876e-05,
+      "loss": 0.3775,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312634216177524e-05,
+      "loss": 0.3005,
+      "step": 3891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3124552612741593e-05,
+      "loss": 0.2781,
+      "step": 3892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312276306370795e-05,
+      "loss": 0.3544,
+      "step": 3893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3120973514674304e-05,
+      "loss": 0.2879,
+      "step": 3894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311918396564066e-05,
+      "loss": 0.2197,
+      "step": 3895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3117394416607014e-05,
+      "loss": 0.3486,
+      "step": 3896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311560486757337e-05,
+      "loss": 0.2172,
+      "step": 3897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311381531853973e-05,
+      "loss": 0.3682,
+      "step": 3898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311202576950609e-05,
+      "loss": 0.3812,
+      "step": 3899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311023622047244e-05,
+      "loss": 0.3347,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3108446671438804e-05,
+      "loss": 0.2207,
+      "step": 3901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310665712240516e-05,
+      "loss": 0.2646,
+      "step": 3902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310486757337151e-05,
+      "loss": 0.3049,
+      "step": 3903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310307802433787e-05,
+      "loss": 0.4117,
+      "step": 3904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3101288475304225e-05,
+      "loss": 0.3503,
+      "step": 3905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309949892627058e-05,
+      "loss": 0.4237,
+      "step": 3906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309770937723694e-05,
+      "loss": 0.2458,
+      "step": 3907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30959198282033e-05,
+      "loss": 0.3183,
+      "step": 3908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309413027916965e-05,
+      "loss": 0.2936,
+      "step": 3909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309234073013601e-05,
+      "loss": 0.3898,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309055118110236e-05,
+      "loss": 0.3544,
+      "step": 3911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308876163206872e-05,
+      "loss": 0.3204,
+      "step": 3912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308697208303508e-05,
+      "loss": 0.4207,
+      "step": 3913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3085182534001435e-05,
+      "loss": 0.452,
+      "step": 3914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308339298496779e-05,
+      "loss": 0.3199,
+      "step": 3915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3081603435934145e-05,
+      "loss": 0.2419,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307981388690051e-05,
+      "loss": 0.3245,
+      "step": 3917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3078024337866856e-05,
+      "loss": 0.2871,
+      "step": 3918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307623478883321e-05,
+      "loss": 0.3295,
+      "step": 3919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307444523979957e-05,
+      "loss": 0.3304,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307265569076593e-05,
+      "loss": 0.3572,
+      "step": 3921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307086614173228e-05,
+      "loss": 0.4698,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3069076592698645e-05,
+      "loss": 0.2857,
+      "step": 3923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3067287043665e-05,
+      "loss": 0.3985,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3065497494631356e-05,
+      "loss": 0.2994,
+      "step": 3925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306370794559771e-05,
+      "loss": 0.3041,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3061918396564066e-05,
+      "loss": 0.3113,
+      "step": 3927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306012884753042e-05,
+      "loss": 0.3178,
+      "step": 3928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305833929849678e-05,
+      "loss": 0.3848,
+      "step": 3929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305654974946314e-05,
+      "loss": 0.2315,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3054760200429493e-05,
+      "loss": 0.3829,
+      "step": 3931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305297065139585e-05,
+      "loss": 0.4734,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305118110236221e-05,
+      "loss": 0.2763,
+      "step": 3933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304939155332856e-05,
+      "loss": 0.3905,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304760200429492e-05,
+      "loss": 0.2853,
+      "step": 3935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3045812455261276e-05,
+      "loss": 0.2415,
+      "step": 3936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304402290622763e-05,
+      "loss": 0.2505,
+      "step": 3937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3042233357193987e-05,
+      "loss": 0.3041,
+      "step": 3938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304044380816035e-05,
+      "loss": 0.2568,
+      "step": 3939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3038654259126704e-05,
+      "loss": 0.4342,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303686471009306e-05,
+      "loss": 0.4155,
+      "step": 3941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3035075161059414e-05,
+      "loss": 0.4214,
+      "step": 3942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303328561202577e-05,
+      "loss": 0.3255,
+      "step": 3943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3031496062992125e-05,
+      "loss": 0.3974,
+      "step": 3944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3029706513958486e-05,
+      "loss": 0.3919,
+      "step": 3945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302791696492484e-05,
+      "loss": 0.3658,
+      "step": 3946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.30261274158912e-05,
+      "loss": 0.3261,
+      "step": 3947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.302433786685756e-05,
+      "loss": 0.2647,
+      "step": 3948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3022548317823914e-05,
+      "loss": 0.2501,
+      "step": 3949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.302075876879026e-05,
+      "loss": 0.2675,
+      "step": 3950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3018969219756624e-05,
+      "loss": 0.158,
+      "step": 3951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301717967072298e-05,
+      "loss": 0.3091,
+      "step": 3952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3015390121689335e-05,
+      "loss": 0.3204,
+      "step": 3953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.30136005726557e-05,
+      "loss": 0.464,
+      "step": 3954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301181102362205e-05,
+      "loss": 0.3456,
+      "step": 3955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301002147458841e-05,
+      "loss": 0.4487,
+      "step": 3956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300823192555476e-05,
+      "loss": 0.4262,
+      "step": 3957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300644237652112e-05,
+      "loss": 0.1999,
+      "step": 3958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300465282748747e-05,
+      "loss": 0.3532,
+      "step": 3959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300286327845383e-05,
+      "loss": 0.3613,
+      "step": 3960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300107372942019e-05,
+      "loss": 0.31,
+      "step": 3961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2999284180386545e-05,
+      "loss": 0.4697,
+      "step": 3962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.29974946313529e-05,
+      "loss": 0.3305,
+      "step": 3963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.299570508231926e-05,
+      "loss": 0.3251,
+      "step": 3964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.299391553328561e-05,
+      "loss": 0.3117,
+      "step": 3965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2992125984251966e-05,
+      "loss": 0.4118,
+      "step": 3966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.299033643521833e-05,
+      "loss": 0.3272,
+      "step": 3967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.298854688618468e-05,
+      "loss": 0.351,
+      "step": 3968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.298675733715104e-05,
+      "loss": 0.4955,
+      "step": 3969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.29849677881174e-05,
+      "loss": 0.3248,
+      "step": 3970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2983178239083755e-05,
+      "loss": 0.3116,
+      "step": 3971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.298138869005011e-05,
+      "loss": 0.4173,
+      "step": 3972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2979599141016466e-05,
+      "loss": 0.3447,
+      "step": 3973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297780959198282e-05,
+      "loss": 0.3137,
+      "step": 3974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2976020042949176e-05,
+      "loss": 0.2557,
+      "step": 3975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297423049391554e-05,
+      "loss": 0.312,
+      "step": 3976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297244094488189e-05,
+      "loss": 0.2601,
+      "step": 3977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297065139584825e-05,
+      "loss": 0.4416,
+      "step": 3978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2968861846814604e-05,
+      "loss": 0.2794,
+      "step": 3979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2967072297780966e-05,
+      "loss": 0.324,
+      "step": 3980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2965282748747314e-05,
+      "loss": 0.3079,
+      "step": 3981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.296349319971367e-05,
+      "loss": 0.3757,
+      "step": 3982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.296170365068003e-05,
+      "loss": 0.3631,
+      "step": 3983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2959914101646386e-05,
+      "loss": 0.39,
+      "step": 3984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.295812455261274e-05,
+      "loss": 0.3307,
+      "step": 3985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2956335003579104e-05,
+      "loss": 0.2781,
+      "step": 3986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.295454545454546e-05,
+      "loss": 0.351,
+      "step": 3987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2952755905511814e-05,
+      "loss": 0.3065,
+      "step": 3988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.295096635647817e-05,
+      "loss": 0.2445,
+      "step": 3989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2949176807444524e-05,
+      "loss": 0.2729,
+      "step": 3990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.294738725841088e-05,
+      "loss": 0.287,
+      "step": 3991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.294559770937724e-05,
+      "loss": 0.3515,
+      "step": 3992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.29438081603436e-05,
+      "loss": 0.3443,
+      "step": 3993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.294201861130995e-05,
+      "loss": 0.5172,
+      "step": 3994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.294022906227631e-05,
+      "loss": 0.2907,
+      "step": 3995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.293843951324267e-05,
+      "loss": 0.2487,
+      "step": 3996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.293664996420902e-05,
+      "loss": 0.442,
+      "step": 3997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.293486041517538e-05,
+      "loss": 0.3038,
+      "step": 3998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2933070866141735e-05,
+      "loss": 0.4015,
+      "step": 3999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.293128131710809e-05,
+      "loss": 0.2404,
+      "step": 4000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2929491768074445e-05,
+      "loss": 0.3664,
+      "step": 4001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.292770221904081e-05,
+      "loss": 0.2889,
+      "step": 4002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292591267000716e-05,
+      "loss": 0.3254,
+      "step": 4003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292412312097351e-05,
+      "loss": 0.3626,
+      "step": 4004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292233357193987e-05,
+      "loss": 0.4126,
+      "step": 4005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292054402290623e-05,
+      "loss": 0.3817,
+      "step": 4006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.291875447387258e-05,
+      "loss": 0.3442,
+      "step": 4007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2916964924838945e-05,
+      "loss": 0.3861,
+      "step": 4008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29151753758053e-05,
+      "loss": 0.3243,
+      "step": 4009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2913385826771655e-05,
+      "loss": 0.345,
+      "step": 4010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.291159627773802e-05,
+      "loss": 0.2543,
+      "step": 4011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2909806728704366e-05,
+      "loss": 0.4541,
+      "step": 4012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290801717967072e-05,
+      "loss": 0.2993,
+      "step": 4013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290622763063708e-05,
+      "loss": 0.4193,
+      "step": 4014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290443808160344e-05,
+      "loss": 0.3707,
+      "step": 4015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290264853256979e-05,
+      "loss": 0.2704,
+      "step": 4016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2900858983536155e-05,
+      "loss": 0.2551,
+      "step": 4017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.289906943450251e-05,
+      "loss": 0.3711,
+      "step": 4018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2897279885468866e-05,
+      "loss": 0.3437,
+      "step": 4019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.289549033643522e-05,
+      "loss": 0.3403,
+      "step": 4020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2893700787401576e-05,
+      "loss": 0.2797,
+      "step": 4021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.289191123836793e-05,
+      "loss": 0.291,
+      "step": 4022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2890121689334286e-05,
+      "loss": 0.2496,
+      "step": 4023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288833214030065e-05,
+      "loss": 0.4601,
+      "step": 4024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2886542591267004e-05,
+      "loss": 0.2918,
+      "step": 4025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288475304223336e-05,
+      "loss": 0.3745,
+      "step": 4026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288296349319972e-05,
+      "loss": 0.2904,
+      "step": 4027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.288117394416607e-05,
+      "loss": 0.3345,
+      "step": 4028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2879384395132424e-05,
+      "loss": 0.2893,
+      "step": 4029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2877594846098786e-05,
+      "loss": 0.3081,
+      "step": 4030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.287580529706514e-05,
+      "loss": 0.2839,
+      "step": 4031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.28740157480315e-05,
+      "loss": 0.2449,
+      "step": 4032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.287222619899786e-05,
+      "loss": 0.3832,
+      "step": 4033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2870436649964214e-05,
+      "loss": 0.3827,
+      "step": 4034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.286864710093057e-05,
+      "loss": 0.2366,
+      "step": 4035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2866857551896924e-05,
+      "loss": 0.5089,
+      "step": 4036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.286506800286328e-05,
+      "loss": 0.3516,
+      "step": 4037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2863278453829635e-05,
+      "loss": 0.3087,
+      "step": 4038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2861488904795997e-05,
+      "loss": 0.2254,
+      "step": 4039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285969935576235e-05,
+      "loss": 0.3069,
+      "step": 4040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285790980672871e-05,
+      "loss": 0.4149,
+      "step": 4041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285612025769506e-05,
+      "loss": 0.2912,
+      "step": 4042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2854330708661424e-05,
+      "loss": 0.4772,
+      "step": 4043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285254115962777e-05,
+      "loss": 0.2765,
+      "step": 4044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.285075161059413e-05,
+      "loss": 0.3971,
+      "step": 4045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.284896206156049e-05,
+      "loss": 0.3879,
+      "step": 4046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2847172512526845e-05,
+      "loss": 0.2657,
+      "step": 4047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.28453829634932e-05,
+      "loss": 0.1536,
+      "step": 4048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.284359341445956e-05,
+      "loss": 0.3347,
+      "step": 4049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.284180386542592e-05,
+      "loss": 0.359,
+      "step": 4050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2840014316392266e-05,
+      "loss": 0.2861,
+      "step": 4051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283822476735863e-05,
+      "loss": 0.269,
+      "step": 4052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283643521832498e-05,
+      "loss": 0.4234,
+      "step": 4053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.283464566929134e-05,
+      "loss": 0.4026,
+      "step": 4054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.28328561202577e-05,
+      "loss": 0.4267,
+      "step": 4055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2831066571224055e-05,
+      "loss": 0.2821,
+      "step": 4056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.282927702219041e-05,
+      "loss": 0.2589,
+      "step": 4057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2827487473156766e-05,
+      "loss": 0.2784,
+      "step": 4058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.282569792412312e-05,
+      "loss": 0.3079,
+      "step": 4059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2823908375089476e-05,
+      "loss": 0.32,
+      "step": 4060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.282211882605584e-05,
+      "loss": 0.3957,
+      "step": 4061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.282032927702219e-05,
+      "loss": 0.3238,
+      "step": 4062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281853972798855e-05,
+      "loss": 0.2688,
+      "step": 4063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2816750178954904e-05,
+      "loss": 0.3828,
+      "step": 4064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2814960629921265e-05,
+      "loss": 0.5077,
+      "step": 4065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281317108088762e-05,
+      "loss": 0.2602,
+      "step": 4066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281138153185397e-05,
+      "loss": 0.3039,
+      "step": 4067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.280959198282033e-05,
+      "loss": 0.237,
+      "step": 4068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2807802433786686e-05,
+      "loss": 0.3966,
+      "step": 4069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.280601288475304e-05,
+      "loss": 0.2659,
+      "step": 4070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2804223335719403e-05,
+      "loss": 0.3885,
+      "step": 4071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.280243378668576e-05,
+      "loss": 0.2818,
+      "step": 4072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2800644237652114e-05,
+      "loss": 0.3919,
+      "step": 4073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2798854688618476e-05,
+      "loss": 0.369,
+      "step": 4074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2797065139584824e-05,
+      "loss": 0.3399,
+      "step": 4075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.279527559055118e-05,
+      "loss": 0.2901,
+      "step": 4076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.279348604151754e-05,
+      "loss": 0.3504,
+      "step": 4077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2791696492483897e-05,
+      "loss": 0.369,
+      "step": 4078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.278990694345025e-05,
+      "loss": 0.5817,
+      "step": 4079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.278811739441661e-05,
+      "loss": 0.3195,
+      "step": 4080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.278632784538297e-05,
+      "loss": 0.3081,
+      "step": 4081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2784538296349324e-05,
+      "loss": 0.2181,
+      "step": 4082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.278274874731568e-05,
+      "loss": 0.3025,
+      "step": 4083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2780959198282034e-05,
+      "loss": 0.5926,
+      "step": 4084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.277916964924839e-05,
+      "loss": 0.3646,
+      "step": 4085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2777380100214745e-05,
+      "loss": 0.3362,
+      "step": 4086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.277559055118111e-05,
+      "loss": 0.3917,
+      "step": 4087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.277380100214746e-05,
+      "loss": 0.2596,
+      "step": 4088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.277201145311382e-05,
+      "loss": 0.3483,
+      "step": 4089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.277022190408018e-05,
+      "loss": 0.286,
+      "step": 4090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.276843235504653e-05,
+      "loss": 0.2689,
+      "step": 4091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.276664280601288e-05,
+      "loss": 0.2006,
+      "step": 4092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2764853256979245e-05,
+      "loss": 0.3233,
+      "step": 4093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.27630637079456e-05,
+      "loss": 0.398,
+      "step": 4094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2761274158911955e-05,
+      "loss": 0.1964,
+      "step": 4095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275948460987832e-05,
+      "loss": 0.3878,
+      "step": 4096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275769506084467e-05,
+      "loss": 0.3642,
+      "step": 4097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275590551181102e-05,
+      "loss": 0.2392,
+      "step": 4098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275411596277738e-05,
+      "loss": 0.2419,
+      "step": 4099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275232641374374e-05,
+      "loss": 0.4394,
+      "step": 4100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.275053686471009e-05,
+      "loss": 0.3008,
+      "step": 4101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2748747315676455e-05,
+      "loss": 0.3435,
+      "step": 4102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.274695776664281e-05,
+      "loss": 0.2519,
+      "step": 4103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2745168217609165e-05,
+      "loss": 0.2926,
+      "step": 4104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.274337866857552e-05,
+      "loss": 0.3821,
+      "step": 4105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2741589119541876e-05,
+      "loss": 0.382,
+      "step": 4106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.273979957050823e-05,
+      "loss": 0.3452,
+      "step": 4107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2738010021474586e-05,
+      "loss": 0.1886,
+      "step": 4108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.273622047244095e-05,
+      "loss": 0.3143,
+      "step": 4109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.27344309234073e-05,
+      "loss": 0.268,
+      "step": 4110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.273264137437366e-05,
+      "loss": 0.2565,
+      "step": 4111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.273085182534002e-05,
+      "loss": 0.3156,
+      "step": 4112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2729062276306376e-05,
+      "loss": 0.2893,
+      "step": 4113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2727272727272724e-05,
+      "loss": 0.3176,
+      "step": 4114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2725483178239086e-05,
+      "loss": 0.3066,
+      "step": 4115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.272369362920544e-05,
+      "loss": 0.262,
+      "step": 4116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2721904080171797e-05,
+      "loss": 0.3239,
+      "step": 4117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.272011453113816e-05,
+      "loss": 0.3061,
+      "step": 4118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2718324982104514e-05,
+      "loss": 0.3126,
+      "step": 4119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.271653543307087e-05,
+      "loss": 0.3491,
+      "step": 4120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2714745884037224e-05,
+      "loss": 0.247,
+      "step": 4121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.271295633500358e-05,
+      "loss": 0.3528,
+      "step": 4122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2711166785969934e-05,
+      "loss": 0.3418,
+      "step": 4123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2709377236936296e-05,
+      "loss": 0.3405,
+      "step": 4124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.270758768790265e-05,
+      "loss": 0.262,
+      "step": 4125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.270579813886901e-05,
+      "loss": 0.2201,
+      "step": 4126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.270400858983536e-05,
+      "loss": 0.2729,
+      "step": 4127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2702219040801724e-05,
+      "loss": 0.2838,
+      "step": 4128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.270042949176808e-05,
+      "loss": 0.3298,
+      "step": 4129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.269863994273443e-05,
+      "loss": 0.3135,
+      "step": 4130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.269685039370079e-05,
+      "loss": 0.2896,
+      "step": 4131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2695060844667145e-05,
+      "loss": 0.6434,
+      "step": 4132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.26932712956335e-05,
+      "loss": 0.4306,
+      "step": 4133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.269148174659986e-05,
+      "loss": 0.3575,
+      "step": 4134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.268969219756622e-05,
+      "loss": 0.4071,
+      "step": 4135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.268790264853257e-05,
+      "loss": 0.4413,
+      "step": 4136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.268611309949893e-05,
+      "loss": 0.4213,
+      "step": 4137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.268432355046528e-05,
+      "loss": 0.3888,
+      "step": 4138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.268253400143164e-05,
+      "loss": 0.3502,
+      "step": 4139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2680744452398e-05,
+      "loss": 0.3355,
+      "step": 4140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2678954903364355e-05,
+      "loss": 0.2961,
+      "step": 4141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.267716535433071e-05,
+      "loss": 0.3758,
+      "step": 4142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2675375805297065e-05,
+      "loss": 0.2953,
+      "step": 4143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.267358625626343e-05,
+      "loss": 0.4698,
+      "step": 4144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2671796707229776e-05,
+      "loss": 0.3826,
+      "step": 4145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.267000715819614e-05,
+      "loss": 0.2595,
+      "step": 4146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.266821760916249e-05,
+      "loss": 0.3228,
+      "step": 4147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.266642806012885e-05,
+      "loss": 0.2112,
+      "step": 4148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.26646385110952e-05,
+      "loss": 0.4391,
+      "step": 4149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2662848962061565e-05,
+      "loss": 0.3936,
+      "step": 4150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.266105941302792e-05,
+      "loss": 0.3652,
+      "step": 4151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2659269863994276e-05,
+      "loss": 0.403,
+      "step": 4152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.265748031496063e-05,
+      "loss": 0.3212,
+      "step": 4153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2655690765926986e-05,
+      "loss": 0.422,
+      "step": 4154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.265390121689334e-05,
+      "loss": 0.4687,
+      "step": 4155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.26521116678597e-05,
+      "loss": 0.3113,
+      "step": 4156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.265032211882606e-05,
+      "loss": 0.3474,
+      "step": 4157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2648532569792414e-05,
+      "loss": 0.5268,
+      "step": 4158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2646743020758776e-05,
+      "loss": 0.2427,
+      "step": 4159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.264495347172513e-05,
+      "loss": 0.1947,
+      "step": 4160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.264316392269148e-05,
+      "loss": 0.3278,
+      "step": 4161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.264137437365784e-05,
+      "loss": 0.2471,
+      "step": 4162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2639584824624196e-05,
+      "loss": 0.3718,
+      "step": 4163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.263779527559055e-05,
+      "loss": 0.2751,
+      "step": 4164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2636005726556914e-05,
+      "loss": 0.3286,
+      "step": 4165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.263421617752327e-05,
+      "loss": 0.4112,
+      "step": 4166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2632426628489624e-05,
+      "loss": 0.2673,
+      "step": 4167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.263063707945598e-05,
+      "loss": 0.3576,
+      "step": 4168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2628847530422334e-05,
+      "loss": 0.298,
+      "step": 4169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.262705798138869e-05,
+      "loss": 0.381,
+      "step": 4170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2625268432355045e-05,
+      "loss": 0.4353,
+      "step": 4171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262347888332141e-05,
+      "loss": 0.2503,
+      "step": 4172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262168933428776e-05,
+      "loss": 0.376,
+      "step": 4173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261989978525412e-05,
+      "loss": 0.402,
+      "step": 4174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261811023622048e-05,
+      "loss": 0.258,
+      "step": 4175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2616320687186834e-05,
+      "loss": 0.3884,
+      "step": 4176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261453113815318e-05,
+      "loss": 0.3268,
+      "step": 4177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2612741589119545e-05,
+      "loss": 0.3263,
+      "step": 4178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26109520400859e-05,
+      "loss": 0.2139,
+      "step": 4179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2609162491052255e-05,
+      "loss": 0.4438,
+      "step": 4180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260737294201862e-05,
+      "loss": 0.3511,
+      "step": 4181
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260558339298497e-05,
+      "loss": 0.2637,
+      "step": 4182
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260379384395133e-05,
+      "loss": 0.2704,
+      "step": 4183
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260200429491768e-05,
+      "loss": 0.3189,
+      "step": 4184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260021474588404e-05,
+      "loss": 0.4339,
+      "step": 4185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.259842519685039e-05,
+      "loss": 0.2884,
+      "step": 4186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2596635647816755e-05,
+      "loss": 0.2924,
+      "step": 4187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.259484609878311e-05,
+      "loss": 0.4594,
+      "step": 4188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2593056549749465e-05,
+      "loss": 0.4367,
+      "step": 4189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.259126700071582e-05,
+      "loss": 0.304,
+      "step": 4190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.258947745168218e-05,
+      "loss": 0.3938,
+      "step": 4191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.258768790264853e-05,
+      "loss": 0.4456,
+      "step": 4192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2585898353614886e-05,
+      "loss": 0.3097,
+      "step": 4193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.258410880458125e-05,
+      "loss": 0.4087,
+      "step": 4194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.25823192555476e-05,
+      "loss": 0.324,
+      "step": 4195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.258052970651396e-05,
+      "loss": 0.3432,
+      "step": 4196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.257874015748032e-05,
+      "loss": 0.2301,
+      "step": 4197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2576950608446676e-05,
+      "loss": 0.2399,
+      "step": 4198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.257516105941303e-05,
+      "loss": 0.4723,
+      "step": 4199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2573371510379386e-05,
+      "loss": 0.3344,
+      "step": 4200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.257158196134574e-05,
+      "loss": 0.2963,
+      "step": 4201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2569792412312096e-05,
+      "loss": 0.3393,
+      "step": 4202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.256800286327846e-05,
+      "loss": 0.4812,
+      "step": 4203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2566213314244813e-05,
+      "loss": 0.4711,
+      "step": 4204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.256442376521117e-05,
+      "loss": 0.2919,
+      "step": 4205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2562634216177524e-05,
+      "loss": 0.2845,
+      "step": 4206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2560844667143886e-05,
+      "loss": 0.4729,
+      "step": 4207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2559055118110234e-05,
+      "loss": 0.3278,
+      "step": 4208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2557265569076596e-05,
+      "loss": 0.4984,
+      "step": 4209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.255547602004295e-05,
+      "loss": 0.3256,
+      "step": 4210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2553686471009307e-05,
+      "loss": 0.3305,
+      "step": 4211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.255189692197566e-05,
+      "loss": 0.2336,
+      "step": 4212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2550107372942024e-05,
+      "loss": 0.267,
+      "step": 4213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.254831782390838e-05,
+      "loss": 0.152,
+      "step": 4214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2546528274874734e-05,
+      "loss": 0.256,
+      "step": 4215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.254473872584109e-05,
+      "loss": 0.3598,
+      "step": 4216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2542949176807445e-05,
+      "loss": 0.4233,
+      "step": 4217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.25411596277738e-05,
+      "loss": 0.5255,
+      "step": 4218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.253937007874016e-05,
+      "loss": 0.4637,
+      "step": 4219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.253758052970652e-05,
+      "loss": 0.2867,
+      "step": 4220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.253579098067287e-05,
+      "loss": 0.2599,
+      "step": 4221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2534001431639234e-05,
+      "loss": 0.2309,
+      "step": 4222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.253221188260559e-05,
+      "loss": 0.3222,
+      "step": 4223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.253042233357194e-05,
+      "loss": 0.4558,
+      "step": 4224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.25286327845383e-05,
+      "loss": 0.2078,
+      "step": 4225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2526843235504655e-05,
+      "loss": 0.4888,
+      "step": 4226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252505368647101e-05,
+      "loss": 0.2888,
+      "step": 4227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2523264137437365e-05,
+      "loss": 0.3504,
+      "step": 4228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252147458840373e-05,
+      "loss": 0.3707,
+      "step": 4229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251968503937008e-05,
+      "loss": 0.4236,
+      "step": 4230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251789549033644e-05,
+      "loss": 0.315,
+      "step": 4231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251610594130279e-05,
+      "loss": 0.2723,
+      "step": 4232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.251431639226915e-05,
+      "loss": 0.2948,
+      "step": 4233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25125268432355e-05,
+      "loss": 0.6033,
+      "step": 4234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2510737294201865e-05,
+      "loss": 0.268,
+      "step": 4235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250894774516822e-05,
+      "loss": 0.2856,
+      "step": 4236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2507158196134576e-05,
+      "loss": 0.2739,
+      "step": 4237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250536864710094e-05,
+      "loss": 0.3163,
+      "step": 4238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2503579098067286e-05,
+      "loss": 0.354,
+      "step": 4239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250178954903364e-05,
+      "loss": 0.3399,
+      "step": 4240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25e-05,
+      "loss": 0.2681,
+      "step": 4241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.249821045096636e-05,
+      "loss": 0.2586,
+      "step": 4242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2496420901932713e-05,
+      "loss": 0.4718,
+      "step": 4243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2494631352899075e-05,
+      "loss": 0.3324,
+      "step": 4244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.249284180386543e-05,
+      "loss": 0.3325,
+      "step": 4245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2491052254831786e-05,
+      "loss": 0.372,
+      "step": 4246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.248926270579814e-05,
+      "loss": 0.3905,
+      "step": 4247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2487473156764496e-05,
+      "loss": 0.2827,
+      "step": 4248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.248568360773085e-05,
+      "loss": 0.3,
+      "step": 4249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.248389405869721e-05,
+      "loss": 0.4124,
+      "step": 4250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.248210450966357e-05,
+      "loss": 0.4172,
+      "step": 4251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2480314960629924e-05,
+      "loss": 0.2851,
+      "step": 4252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.247852541159628e-05,
+      "loss": 0.3736,
+      "step": 4253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.247673586256264e-05,
+      "loss": 0.3191,
+      "step": 4254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.247494631352899e-05,
+      "loss": 0.2317,
+      "step": 4255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2473156764495344e-05,
+      "loss": 0.2729,
+      "step": 4256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2471367215461706e-05,
+      "loss": 0.3028,
+      "step": 4257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246957766642806e-05,
+      "loss": 0.2427,
+      "step": 4258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246778811739442e-05,
+      "loss": 0.2862,
+      "step": 4259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246599856836078e-05,
+      "loss": 0.2668,
+      "step": 4260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2464209019327134e-05,
+      "loss": 0.3453,
+      "step": 4261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246241947029349e-05,
+      "loss": 0.2078,
+      "step": 4262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2460629921259844e-05,
+      "loss": 0.2608,
+      "step": 4263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.24588403722262e-05,
+      "loss": 0.321,
+      "step": 4264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2457050823192555e-05,
+      "loss": 0.3427,
+      "step": 4265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245526127415892e-05,
+      "loss": 0.2579,
+      "step": 4266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245347172512527e-05,
+      "loss": 0.4057,
+      "step": 4267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.245168217609163e-05,
+      "loss": 0.5235,
+      "step": 4268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.244989262705798e-05,
+      "loss": 0.3196,
+      "step": 4269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.244810307802434e-05,
+      "loss": 0.3296,
+      "step": 4270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.244631352899069e-05,
+      "loss": 0.4507,
+      "step": 4271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2444523979957055e-05,
+      "loss": 0.3279,
+      "step": 4272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.244273443092341e-05,
+      "loss": 0.3282,
+      "step": 4273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2440944881889765e-05,
+      "loss": 0.5068,
+      "step": 4274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243915533285612e-05,
+      "loss": 0.3936,
+      "step": 4275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243736578382248e-05,
+      "loss": 0.3854,
+      "step": 4276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243557623478884e-05,
+      "loss": 0.4417,
+      "step": 4277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2433786685755186e-05,
+      "loss": 0.2667,
+      "step": 4278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243199713672155e-05,
+      "loss": 0.4939,
+      "step": 4279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.24302075876879e-05,
+      "loss": 0.2764,
+      "step": 4280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242841803865426e-05,
+      "loss": 0.3796,
+      "step": 4281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.242662848962062e-05,
+      "loss": 0.3771,
+      "step": 4282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2424838940586975e-05,
+      "loss": 0.198,
+      "step": 4283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.242304939155333e-05,
+      "loss": 0.2456,
+      "step": 4284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.242125984251969e-05,
+      "loss": 0.4493,
+      "step": 4285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.241947029348604e-05,
+      "loss": 0.4919,
+      "step": 4286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2417680744452396e-05,
+      "loss": 0.342,
+      "step": 4287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.241589119541876e-05,
+      "loss": 0.4024,
+      "step": 4288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.241410164638511e-05,
+      "loss": 0.2498,
+      "step": 4289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.241231209735147e-05,
+      "loss": 0.3729,
+      "step": 4290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2410522548317824e-05,
+      "loss": 0.3843,
+      "step": 4291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2408732999284186e-05,
+      "loss": 0.3923,
+      "step": 4292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240694345025054e-05,
+      "loss": 0.3162,
+      "step": 4293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2405153901216896e-05,
+      "loss": 0.3236,
+      "step": 4294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240336435218325e-05,
+      "loss": 0.2811,
+      "step": 4295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2401574803149606e-05,
+      "loss": 0.2987,
+      "step": 4296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.239978525411596e-05,
+      "loss": 0.3086,
+      "step": 4297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2397995705082324e-05,
+      "loss": 0.4732,
+      "step": 4298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.239620615604868e-05,
+      "loss": 0.2236,
+      "step": 4299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2394416607015034e-05,
+      "loss": 0.228,
+      "step": 4300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2392627057981396e-05,
+      "loss": 0.3079,
+      "step": 4301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2390837508947744e-05,
+      "loss": 0.2364,
+      "step": 4302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.23890479599141e-05,
+      "loss": 0.3228,
+      "step": 4303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238725841088046e-05,
+      "loss": 0.4201,
+      "step": 4304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238546886184682e-05,
+      "loss": 0.3585,
+      "step": 4305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238367931281317e-05,
+      "loss": 0.3816,
+      "step": 4306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2381889763779534e-05,
+      "loss": 0.3596,
+      "step": 4307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.238010021474589e-05,
+      "loss": 0.4597,
+      "step": 4308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2378310665712244e-05,
+      "loss": 0.3445,
+      "step": 4309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.23765211166786e-05,
+      "loss": 0.3874,
+      "step": 4310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2374731567644955e-05,
+      "loss": 0.3672,
+      "step": 4311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.237294201861131e-05,
+      "loss": 0.2284,
+      "step": 4312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.237115246957767e-05,
+      "loss": 0.3218,
+      "step": 4313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236936292054403e-05,
+      "loss": 0.3484,
+      "step": 4314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236757337151038e-05,
+      "loss": 0.5091,
+      "step": 4315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236578382247674e-05,
+      "loss": 0.2652,
+      "step": 4316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236399427344309e-05,
+      "loss": 0.2647,
+      "step": 4317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.236220472440945e-05,
+      "loss": 0.237,
+      "step": 4318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.23604151753758e-05,
+      "loss": 0.291,
+      "step": 4319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2358625626342165e-05,
+      "loss": 0.3528,
+      "step": 4320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.235683607730852e-05,
+      "loss": 0.317,
+      "step": 4321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2355046528274875e-05,
+      "loss": 0.3484,
+      "step": 4322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.235325697924124e-05,
+      "loss": 0.4037,
+      "step": 4323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.235146743020759e-05,
+      "loss": 0.3452,
+      "step": 4324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234967788117394e-05,
+      "loss": 0.3237,
+      "step": 4325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.23478883321403e-05,
+      "loss": 0.3605,
+      "step": 4326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234609878310666e-05,
+      "loss": 0.4505,
+      "step": 4327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234430923407301e-05,
+      "loss": 0.2197,
+      "step": 4328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2342519685039375e-05,
+      "loss": 0.21,
+      "step": 4329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.234073013600573e-05,
+      "loss": 0.2583,
+      "step": 4330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2338940586972086e-05,
+      "loss": 0.3399,
+      "step": 4331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233715103793844e-05,
+      "loss": 0.3662,
+      "step": 4332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2335361488904796e-05,
+      "loss": 0.4279,
+      "step": 4333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233357193987115e-05,
+      "loss": 0.1856,
+      "step": 4334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.233178239083751e-05,
+      "loss": 0.3923,
+      "step": 4335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232999284180387e-05,
+      "loss": 0.2673,
+      "step": 4336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2328203292770224e-05,
+      "loss": 0.2891,
+      "step": 4337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.232641374373658e-05,
+      "loss": 0.2821,
+      "step": 4338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.232462419470294e-05,
+      "loss": 0.3853,
+      "step": 4339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2322834645669296e-05,
+      "loss": 0.3351,
+      "step": 4340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2321045096635644e-05,
+      "loss": 0.3586,
+      "step": 4341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2319255547602006e-05,
+      "loss": 0.3248,
+      "step": 4342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231746599856836e-05,
+      "loss": 0.2596,
+      "step": 4343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231567644953472e-05,
+      "loss": 0.3419,
+      "step": 4344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231388690050108e-05,
+      "loss": 0.2817,
+      "step": 4345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2312097351467434e-05,
+      "loss": 0.3392,
+      "step": 4346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231030780243379e-05,
+      "loss": 0.4162,
+      "step": 4347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.230851825340015e-05,
+      "loss": 0.353,
+      "step": 4348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.23067287043665e-05,
+      "loss": 0.4661,
+      "step": 4349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2304939155332855e-05,
+      "loss": 0.3627,
+      "step": 4350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2303149606299217e-05,
+      "loss": 0.2413,
+      "step": 4351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.230136005726557e-05,
+      "loss": 0.2,
+      "step": 4352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.229957050823193e-05,
+      "loss": 0.2618,
+      "step": 4353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.229778095919828e-05,
+      "loss": 0.3287,
+      "step": 4354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2295991410164644e-05,
+      "loss": 0.2779,
+      "step": 4355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.229420186113099e-05,
+      "loss": 0.2509,
+      "step": 4356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2292412312097355e-05,
+      "loss": 0.3644,
+      "step": 4357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.229062276306371e-05,
+      "loss": 0.2852,
+      "step": 4358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2288833214030065e-05,
+      "loss": 0.3841,
+      "step": 4359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.228704366499642e-05,
+      "loss": 0.2419,
+      "step": 4360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.228525411596278e-05,
+      "loss": 0.3739,
+      "step": 4361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.228346456692914e-05,
+      "loss": 0.4167,
+      "step": 4362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.228167501789549e-05,
+      "loss": 0.3652,
+      "step": 4363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.227988546886185e-05,
+      "loss": 0.3724,
+      "step": 4364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.22780959198282e-05,
+      "loss": 0.2389,
+      "step": 4365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.227630637079456e-05,
+      "loss": 0.3546,
+      "step": 4366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.227451682176092e-05,
+      "loss": 0.3689,
+      "step": 4367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2272727272727275e-05,
+      "loss": 0.3411,
+      "step": 4368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.227093772369363e-05,
+      "loss": 0.4922,
+      "step": 4369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.226914817465999e-05,
+      "loss": 0.2357,
+      "step": 4370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.226735862562635e-05,
+      "loss": 0.5392,
+      "step": 4371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2265569076592696e-05,
+      "loss": 0.3685,
+      "step": 4372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.226377952755906e-05,
+      "loss": 0.3842,
+      "step": 4373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.226198997852541e-05,
+      "loss": 0.3478,
+      "step": 4374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.226020042949177e-05,
+      "loss": 0.4532,
+      "step": 4375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.225841088045813e-05,
+      "loss": 0.4004,
+      "step": 4376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2256621331424485e-05,
+      "loss": 0.2507,
+      "step": 4377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.225483178239084e-05,
+      "loss": 0.3558,
+      "step": 4378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2253042233357196e-05,
+      "loss": 0.3478,
+      "step": 4379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.225125268432355e-05,
+      "loss": 0.2489,
+      "step": 4380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2249463135289906e-05,
+      "loss": 0.3211,
+      "step": 4381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.224767358625626e-05,
+      "loss": 0.3928,
+      "step": 4382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2245884037222623e-05,
+      "loss": 0.3838,
+      "step": 4383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.224409448818898e-05,
+      "loss": 0.3417,
+      "step": 4384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2242304939155334e-05,
+      "loss": 0.2547,
+      "step": 4385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2240515390121696e-05,
+      "loss": 0.3913,
+      "step": 4386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.223872584108805e-05,
+      "loss": 0.2531,
+      "step": 4387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.22369362920544e-05,
+      "loss": 0.4702,
+      "step": 4388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.223514674302076e-05,
+      "loss": 0.3251,
+      "step": 4389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2233357193987117e-05,
+      "loss": 0.2241,
+      "step": 4390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.223156764495347e-05,
+      "loss": 0.2019,
+      "step": 4391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2229778095919834e-05,
+      "loss": 0.3581,
+      "step": 4392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.222798854688619e-05,
+      "loss": 0.3887,
+      "step": 4393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2226198997852544e-05,
+      "loss": 0.3946,
+      "step": 4394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.22244094488189e-05,
+      "loss": 0.3244,
+      "step": 4395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2222619899785254e-05,
+      "loss": 0.4872,
+      "step": 4396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.222083035075161e-05,
+      "loss": 0.3887,
+      "step": 4397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221904080171797e-05,
+      "loss": 0.2172,
+      "step": 4398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221725125268433e-05,
+      "loss": 0.2907,
+      "step": 4399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221546170365068e-05,
+      "loss": 0.3315,
+      "step": 4400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221367215461704e-05,
+      "loss": 0.2733,
+      "step": 4401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.22118826055834e-05,
+      "loss": 0.1804,
+      "step": 4402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221009305654975e-05,
+      "loss": 0.3241,
+      "step": 4403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.22083035075161e-05,
+      "loss": 0.2408,
+      "step": 4404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2206513958482465e-05,
+      "loss": 0.3321,
+      "step": 4405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220472440944882e-05,
+      "loss": 0.328,
+      "step": 4406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2202934860415175e-05,
+      "loss": 0.3348,
+      "step": 4407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220114531138154e-05,
+      "loss": 0.3137,
+      "step": 4408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219935576234789e-05,
+      "loss": 0.3038,
+      "step": 4409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219756621331425e-05,
+      "loss": 0.3816,
+      "step": 4410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21957766642806e-05,
+      "loss": 0.4691,
+      "step": 4411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219398711524696e-05,
+      "loss": 0.2892,
+      "step": 4412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219219756621331e-05,
+      "loss": 0.304,
+      "step": 4413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2190408017179675e-05,
+      "loss": 0.4054,
+      "step": 4414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218861846814603e-05,
+      "loss": 0.2781,
+      "step": 4415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2186828919112385e-05,
+      "loss": 0.2731,
+      "step": 4416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218503937007874e-05,
+      "loss": 0.3056,
+      "step": 4417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21832498210451e-05,
+      "loss": 0.3084,
+      "step": 4418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218146027201145e-05,
+      "loss": 0.3517,
+      "step": 4419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217967072297781e-05,
+      "loss": 0.2873,
+      "step": 4420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217788117394417e-05,
+      "loss": 0.3505,
+      "step": 4421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217609162491052e-05,
+      "loss": 0.3443,
+      "step": 4422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217430207587688e-05,
+      "loss": 0.2821,
+      "step": 4423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217251252684324e-05,
+      "loss": 0.4261,
+      "step": 4424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2170722977809596e-05,
+      "loss": 0.2478,
+      "step": 4425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.216893342877595e-05,
+      "loss": 0.3008,
+      "step": 4426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2167143879742306e-05,
+      "loss": 0.2567,
+      "step": 4427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.216535433070866e-05,
+      "loss": 0.4346,
+      "step": 4428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2163564781675016e-05,
+      "loss": 0.3298,
+      "step": 4429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.216177523264138e-05,
+      "loss": 0.2795,
+      "step": 4430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2159985683607734e-05,
+      "loss": 0.25,
+      "step": 4431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.215819613457409e-05,
+      "loss": 0.326,
+      "step": 4432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.215640658554045e-05,
+      "loss": 0.2967,
+      "step": 4433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2154617036506806e-05,
+      "loss": 0.2742,
+      "step": 4434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2152827487473154e-05,
+      "loss": 0.4003,
+      "step": 4435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2151037938439516e-05,
+      "loss": 0.3039,
+      "step": 4436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214924838940587e-05,
+      "loss": 0.203,
+      "step": 4437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214745884037223e-05,
+      "loss": 0.3007,
+      "step": 4438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214566929133858e-05,
+      "loss": 0.3739,
+      "step": 4439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2143879742304944e-05,
+      "loss": 0.4272,
+      "step": 4440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21420901932713e-05,
+      "loss": 0.3905,
+      "step": 4441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2140300644237654e-05,
+      "loss": 0.2134,
+      "step": 4442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213851109520401e-05,
+      "loss": 0.1713,
+      "step": 4443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2136721546170365e-05,
+      "loss": 0.2738,
+      "step": 4444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213493199713672e-05,
+      "loss": 0.2976,
+      "step": 4445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213314244810308e-05,
+      "loss": 0.3906,
+      "step": 4446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213135289906944e-05,
+      "loss": 0.2821,
+      "step": 4447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.212956335003579e-05,
+      "loss": 0.2262,
+      "step": 4448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2127773801002154e-05,
+      "loss": 0.3531,
+      "step": 4449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21259842519685e-05,
+      "loss": 0.2577,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212419470293486e-05,
+      "loss": 0.3525,
+      "step": 4451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212240515390122e-05,
+      "loss": 0.2401,
+      "step": 4452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2120615604867575e-05,
+      "loss": 0.2861,
+      "step": 4453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211882605583393e-05,
+      "loss": 0.3182,
+      "step": 4454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211703650680029e-05,
+      "loss": 0.3294,
+      "step": 4455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211524695776665e-05,
+      "loss": 0.4573,
+      "step": 4456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2113457408733e-05,
+      "loss": 0.325,
+      "step": 4457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211166785969936e-05,
+      "loss": 0.3881,
+      "step": 4458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210987831066571e-05,
+      "loss": 0.3658,
+      "step": 4459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210808876163207e-05,
+      "loss": 0.2869,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210629921259843e-05,
+      "loss": 0.4521,
+      "step": 4461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2104509663564785e-05,
+      "loss": 0.2556,
+      "step": 4462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210272011453114e-05,
+      "loss": 0.325,
+      "step": 4463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2100930565497496e-05,
+      "loss": 0.2253,
+      "step": 4464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.209914101646386e-05,
+      "loss": 0.4487,
+      "step": 4465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2097351467430206e-05,
+      "loss": 0.3259,
+      "step": 4466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.209556191839656e-05,
+      "loss": 0.264,
+      "step": 4467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.209377236936292e-05,
+      "loss": 0.1913,
+      "step": 4468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.209198282032928e-05,
+      "loss": 0.386,
+      "step": 4469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2090193271295634e-05,
+      "loss": 0.2728,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2088403722261996e-05,
+      "loss": 0.4697,
+      "step": 4471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.208661417322835e-05,
+      "loss": 0.3327,
+      "step": 4472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2084824624194706e-05,
+      "loss": 0.3733,
+      "step": 4473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.208303507516106e-05,
+      "loss": 0.3125,
+      "step": 4474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2081245526127416e-05,
+      "loss": 0.273,
+      "step": 4475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.207945597709377e-05,
+      "loss": 0.3441,
+      "step": 4476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2077666428060134e-05,
+      "loss": 0.4027,
+      "step": 4477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.207587687902649e-05,
+      "loss": 0.3488,
+      "step": 4478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2074087329992844e-05,
+      "loss": 0.2537,
+      "step": 4479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20722977809592e-05,
+      "loss": 0.2837,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.207050823192556e-05,
+      "loss": 0.3593,
+      "step": 4481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206871868289191e-05,
+      "loss": 0.328,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206692913385827e-05,
+      "loss": 0.2764,
+      "step": 4483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206513958482463e-05,
+      "loss": 0.3194,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206335003579098e-05,
+      "loss": 0.3126,
+      "step": 4485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.206156048675734e-05,
+      "loss": 0.3144,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20597709377237e-05,
+      "loss": 0.4465,
+      "step": 4487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2057981388690054e-05,
+      "loss": 0.3941,
+      "step": 4488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20561918396564e-05,
+      "loss": 0.3107,
+      "step": 4489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2054402290622765e-05,
+      "loss": 0.3368,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.205261274158912e-05,
+      "loss": 0.3934,
+      "step": 4491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2050823192555475e-05,
+      "loss": 0.3363,
+      "step": 4492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204903364352184e-05,
+      "loss": 0.2274,
+      "step": 4493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204724409448819e-05,
+      "loss": 0.2584,
+      "step": 4494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 0.2452,
+      "step": 4495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204366499642091e-05,
+      "loss": 0.2836,
+      "step": 4496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204187544738726e-05,
+      "loss": 0.3781,
+      "step": 4497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.204008589835361e-05,
+      "loss": 0.3161,
+      "step": 4498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2038296349319975e-05,
+      "loss": 0.4618,
+      "step": 4499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.203650680028633e-05,
+      "loss": 0.2263,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2034717251252685e-05,
+      "loss": 0.4496,
+      "step": 4501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.203292770221904e-05,
+      "loss": 0.1732,
+      "step": 4502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20311381531854e-05,
+      "loss": 0.1899,
+      "step": 4503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.202934860415176e-05,
+      "loss": 0.4436,
+      "step": 4504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.202755905511811e-05,
+      "loss": 0.243,
+      "step": 4505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.202576950608447e-05,
+      "loss": 0.2078,
+      "step": 4506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.202397995705082e-05,
+      "loss": 0.362,
+      "step": 4507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.202219040801718e-05,
+      "loss": 0.3138,
+      "step": 4508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.202040085898354e-05,
+      "loss": 0.4371,
+      "step": 4509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2018611309949896e-05,
+      "loss": 0.2548,
+      "step": 4510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.201682176091625e-05,
+      "loss": 0.2929,
+      "step": 4511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.201503221188261e-05,
+      "loss": 0.2669,
+      "step": 4512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.201324266284896e-05,
+      "loss": 0.332,
+      "step": 4513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2011453113815316e-05,
+      "loss": 0.2743,
+      "step": 4514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200966356478168e-05,
+      "loss": 0.2817,
+      "step": 4515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2007874015748033e-05,
+      "loss": 0.3361,
+      "step": 4516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200608446671439e-05,
+      "loss": 0.2393,
+      "step": 4517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200429491768075e-05,
+      "loss": 0.3218,
+      "step": 4518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2002505368647106e-05,
+      "loss": 0.2606,
+      "step": 4519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200071581961346e-05,
+      "loss": 0.2034,
+      "step": 4520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1998926270579816e-05,
+      "loss": 0.3466,
+      "step": 4521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.199713672154617e-05,
+      "loss": 0.3229,
+      "step": 4522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1995347172512527e-05,
+      "loss": 0.3872,
+      "step": 4523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.199355762347889e-05,
+      "loss": 0.3225,
+      "step": 4524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1991768074445244e-05,
+      "loss": 0.3198,
+      "step": 4525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.19899785254116e-05,
+      "loss": 0.2888,
+      "step": 4526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1988188976377954e-05,
+      "loss": 0.3395,
+      "step": 4527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1986399427344316e-05,
+      "loss": 0.2179,
+      "step": 4528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1984609878310665e-05,
+      "loss": 0.2871,
+      "step": 4529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.198282032927702e-05,
+      "loss": 0.2177,
+      "step": 4530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.198103078024338e-05,
+      "loss": 0.4114,
+      "step": 4531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.197924123120974e-05,
+      "loss": 0.2176,
+      "step": 4532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.197745168217609e-05,
+      "loss": 0.2855,
+      "step": 4533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1975662133142454e-05,
+      "loss": 0.2936,
+      "step": 4534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.197387258410881e-05,
+      "loss": 0.3581,
+      "step": 4535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.197208303507516e-05,
+      "loss": 0.1808,
+      "step": 4536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.197029348604152e-05,
+      "loss": 0.4985,
+      "step": 4537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1968503937007875e-05,
+      "loss": 0.3549,
+      "step": 4538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.196671438797423e-05,
+      "loss": 0.1787,
+      "step": 4539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.196492483894059e-05,
+      "loss": 0.2813,
+      "step": 4540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.196313528990695e-05,
+      "loss": 0.4224,
+      "step": 4541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.19613457408733e-05,
+      "loss": 0.2345,
+      "step": 4542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195955619183966e-05,
+      "loss": 0.2084,
+      "step": 4543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195776664280601e-05,
+      "loss": 0.3659,
+      "step": 4544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195597709377237e-05,
+      "loss": 0.2292,
+      "step": 4545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195418754473873e-05,
+      "loss": 0.2167,
+      "step": 4546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1952397995705085e-05,
+      "loss": 0.3149,
+      "step": 4547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195060844667144e-05,
+      "loss": 0.2143,
+      "step": 4548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1948818897637795e-05,
+      "loss": 0.3462,
+      "step": 4549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194702934860416e-05,
+      "loss": 0.3315,
+      "step": 4550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194523979957051e-05,
+      "loss": 0.2193,
+      "step": 4551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194345025053686e-05,
+      "loss": 0.3934,
+      "step": 4552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.194166070150322e-05,
+      "loss": 0.3338,
+      "step": 4553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.193987115246958e-05,
+      "loss": 0.235,
+      "step": 4554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1938081603435933e-05,
+      "loss": 0.3717,
+      "step": 4555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1936292054402295e-05,
+      "loss": 0.3042,
+      "step": 4556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.193450250536865e-05,
+      "loss": 0.2296,
+      "step": 4557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1932712956335006e-05,
+      "loss": 0.3221,
+      "step": 4558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.193092340730137e-05,
+      "loss": 0.4369,
+      "step": 4559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1929133858267716e-05,
+      "loss": 0.2682,
+      "step": 4560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.192734430923407e-05,
+      "loss": 0.168,
+      "step": 4561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.192555476020043e-05,
+      "loss": 0.2802,
+      "step": 4562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.192376521116679e-05,
+      "loss": 0.4873,
+      "step": 4563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1921975662133144e-05,
+      "loss": 0.3097,
+      "step": 4564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.19201861130995e-05,
+      "loss": 0.3422,
+      "step": 4565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.191839656406586e-05,
+      "loss": 0.3959,
+      "step": 4566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1916607015032216e-05,
+      "loss": 0.3753,
+      "step": 4567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.191481746599857e-05,
+      "loss": 0.2656,
+      "step": 4568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1913027916964926e-05,
+      "loss": 0.3508,
+      "step": 4569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.191123836793128e-05,
+      "loss": 0.3392,
+      "step": 4570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190944881889764e-05,
+      "loss": 0.3571,
+      "step": 4571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1907659269864e-05,
+      "loss": 0.3215,
+      "step": 4572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1905869720830354e-05,
+      "loss": 0.407,
+      "step": 4573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190408017179671e-05,
+      "loss": 0.2633,
+      "step": 4574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1902290622763064e-05,
+      "loss": 0.2825,
+      "step": 4575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190050107372942e-05,
+      "loss": 0.4296,
+      "step": 4576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1898711524695775e-05,
+      "loss": 0.2679,
+      "step": 4577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.189692197566214e-05,
+      "loss": 0.3416,
+      "step": 4578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.189513242662849e-05,
+      "loss": 0.2908,
+      "step": 4579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.189334287759485e-05,
+      "loss": 0.3808,
+      "step": 4580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.189155332856121e-05,
+      "loss": 0.2382,
+      "step": 4581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1889763779527564e-05,
+      "loss": 0.3513,
+      "step": 4582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.188797423049391e-05,
+      "loss": 0.1679,
+      "step": 4583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1886184681460275e-05,
+      "loss": 0.3937,
+      "step": 4584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.188439513242663e-05,
+      "loss": 0.3589,
+      "step": 4585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1882605583392985e-05,
+      "loss": 0.3487,
+      "step": 4586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.188081603435934e-05,
+      "loss": 0.273,
+      "step": 4587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.18790264853257e-05,
+      "loss": 0.3012,
+      "step": 4588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187723693629206e-05,
+      "loss": 0.4022,
+      "step": 4589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187544738725841e-05,
+      "loss": 0.2514,
+      "step": 4590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187365783822477e-05,
+      "loss": 0.3058,
+      "step": 4591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187186828919112e-05,
+      "loss": 0.2061,
+      "step": 4592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.187007874015748e-05,
+      "loss": 0.421,
+      "step": 4593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.186828919112384e-05,
+      "loss": 0.2214,
+      "step": 4594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1866499642090195e-05,
+      "loss": 0.3859,
+      "step": 4595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.186471009305655e-05,
+      "loss": 0.2268,
+      "step": 4596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.186292054402291e-05,
+      "loss": 0.3071,
+      "step": 4597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.186113099498927e-05,
+      "loss": 0.3118,
+      "step": 4598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1859341445955616e-05,
+      "loss": 0.4119,
+      "step": 4599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.185755189692198e-05,
+      "loss": 0.3126,
+      "step": 4600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.185576234788833e-05,
+      "loss": 0.2961,
+      "step": 4601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.185397279885469e-05,
+      "loss": 0.3417,
+      "step": 4602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.185218324982105e-05,
+      "loss": 0.4044,
+      "step": 4603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1850393700787406e-05,
+      "loss": 0.2335,
+      "step": 4604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.184860415175376e-05,
+      "loss": 0.4206,
+      "step": 4605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1846814602720116e-05,
+      "loss": 0.3083,
+      "step": 4606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.184502505368647e-05,
+      "loss": 0.3236,
+      "step": 4607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1843235504652826e-05,
+      "loss": 0.3879,
+      "step": 4608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.184144595561919e-05,
+      "loss": 0.5328,
+      "step": 4609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1839656406585544e-05,
+      "loss": 0.3252,
+      "step": 4610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.18378668575519e-05,
+      "loss": 0.189,
+      "step": 4611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1836077308518254e-05,
+      "loss": 0.2729,
+      "step": 4612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1834287759484616e-05,
+      "loss": 0.2734,
+      "step": 4613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.183249821045097e-05,
+      "loss": 0.2318,
+      "step": 4614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.183070866141732e-05,
+      "loss": 0.3184,
+      "step": 4615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.182891911238368e-05,
+      "loss": 0.2946,
+      "step": 4616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.182712956335004e-05,
+      "loss": 0.3197,
+      "step": 4617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.182534001431639e-05,
+      "loss": 0.2651,
+      "step": 4618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1823550465282754e-05,
+      "loss": 0.3071,
+      "step": 4619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.182176091624911e-05,
+      "loss": 0.5697,
+      "step": 4620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1819971367215464e-05,
+      "loss": 0.2237,
+      "step": 4621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.4279,
+      "step": 4622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1816392269148175e-05,
+      "loss": 0.2159,
+      "step": 4623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181460272011453e-05,
+      "loss": 0.1672,
+      "step": 4624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181281317108089e-05,
+      "loss": 0.2032,
+      "step": 4625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181102362204725e-05,
+      "loss": 0.3045,
+      "step": 4626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.18092340730136e-05,
+      "loss": 0.344,
+      "step": 4627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180744452397996e-05,
+      "loss": 0.1915,
+      "step": 4628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180565497494632e-05,
+      "loss": 0.3216,
+      "step": 4629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180386542591267e-05,
+      "loss": 0.3092,
+      "step": 4630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180207587687903e-05,
+      "loss": 0.3065,
+      "step": 4631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1800286327845385e-05,
+      "loss": 0.2923,
+      "step": 4632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.179849677881174e-05,
+      "loss": 0.2937,
+      "step": 4633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1796707229778095e-05,
+      "loss": 0.3462,
+      "step": 4634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.179491768074446e-05,
+      "loss": 0.2549,
+      "step": 4635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.179312813171081e-05,
+      "loss": 0.237,
+      "step": 4636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.179133858267717e-05,
+      "loss": 0.2522,
+      "step": 4637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.178954903364352e-05,
+      "loss": 0.2842,
+      "step": 4638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.178775948460988e-05,
+      "loss": 0.3451,
+      "step": 4639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.178596993557623e-05,
+      "loss": 0.2464,
+      "step": 4640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1784180386542595e-05,
+      "loss": 0.242,
+      "step": 4641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.178239083750895e-05,
+      "loss": 0.3832,
+      "step": 4642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1780601288475306e-05,
+      "loss": 0.1532,
+      "step": 4643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177881173944167e-05,
+      "loss": 0.3094,
+      "step": 4644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177702219040802e-05,
+      "loss": 0.4123,
+      "step": 4645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177523264137437e-05,
+      "loss": 0.3911,
+      "step": 4646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177344309234073e-05,
+      "loss": 0.3286,
+      "step": 4647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177165354330709e-05,
+      "loss": 0.3062,
+      "step": 4648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1769863994273444e-05,
+      "loss": 0.2319,
+      "step": 4649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.17680744452398e-05,
+      "loss": 0.2492,
+      "step": 4650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.176628489620616e-05,
+      "loss": 0.3572,
+      "step": 4651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1764495347172516e-05,
+      "loss": 0.2422,
+      "step": 4652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.176270579813887e-05,
+      "loss": 0.4067,
+      "step": 4653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1760916249105226e-05,
+      "loss": 0.3472,
+      "step": 4654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.175912670007158e-05,
+      "loss": 0.3655,
+      "step": 4655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.175733715103794e-05,
+      "loss": 0.3425,
+      "step": 4656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.17555476020043e-05,
+      "loss": 0.4021,
+      "step": 4657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1753758052970654e-05,
+      "loss": 0.3435,
+      "step": 4658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.175196850393701e-05,
+      "loss": 0.261,
+      "step": 4659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.175017895490337e-05,
+      "loss": 0.3012,
+      "step": 4660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1748389405869726e-05,
+      "loss": 0.2604,
+      "step": 4661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1746599856836075e-05,
+      "loss": 0.3248,
+      "step": 4662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1744810307802437e-05,
+      "loss": 0.3494,
+      "step": 4663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.174302075876879e-05,
+      "loss": 0.4146,
+      "step": 4664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.174123120973515e-05,
+      "loss": 0.2732,
+      "step": 4665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.173944166070151e-05,
+      "loss": 0.3838,
+      "step": 4666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1737652111667864e-05,
+      "loss": 0.2705,
+      "step": 4667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.173586256263422e-05,
+      "loss": 0.2921,
+      "step": 4668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1734073013600575e-05,
+      "loss": 0.2745,
+      "step": 4669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.173228346456693e-05,
+      "loss": 0.3799,
+      "step": 4670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1730493915533285e-05,
+      "loss": 0.3423,
+      "step": 4671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.172870436649965e-05,
+      "loss": 0.3612,
+      "step": 4672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1726914817466e-05,
+      "loss": 0.3142,
+      "step": 4673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.172512526843236e-05,
+      "loss": 0.2294,
+      "step": 4674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.172333571939871e-05,
+      "loss": 0.2374,
+      "step": 4675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1721546170365074e-05,
+      "loss": 0.3838,
+      "step": 4676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171975662133142e-05,
+      "loss": 0.3221,
+      "step": 4677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171796707229778e-05,
+      "loss": 0.3055,
+      "step": 4678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171617752326414e-05,
+      "loss": 0.3775,
+      "step": 4679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1714387974230495e-05,
+      "loss": 0.2878,
+      "step": 4680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171259842519685e-05,
+      "loss": 0.4088,
+      "step": 4681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171080887616321e-05,
+      "loss": 0.1858,
+      "step": 4682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170901932712957e-05,
+      "loss": 0.3441,
+      "step": 4683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170722977809592e-05,
+      "loss": 0.3722,
+      "step": 4684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170544022906228e-05,
+      "loss": 0.3981,
+      "step": 4685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170365068002863e-05,
+      "loss": 0.4493,
+      "step": 4686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170186113099499e-05,
+      "loss": 0.3978,
+      "step": 4687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170007158196135e-05,
+      "loss": 0.2221,
+      "step": 4688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1698282032927705e-05,
+      "loss": 0.3947,
+      "step": 4689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.169649248389406e-05,
+      "loss": 0.3465,
+      "step": 4690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1694702934860416e-05,
+      "loss": 0.2877,
+      "step": 4691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.169291338582678e-05,
+      "loss": 0.2848,
+      "step": 4692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1691123836793126e-05,
+      "loss": 0.3117,
+      "step": 4693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.168933428775949e-05,
+      "loss": 0.3142,
+      "step": 4694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1687544738725843e-05,
+      "loss": 0.3097,
+      "step": 4695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.16857551896922e-05,
+      "loss": 0.2338,
+      "step": 4696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1683965640658554e-05,
+      "loss": 0.2616,
+      "step": 4697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1682176091624916e-05,
+      "loss": 0.3831,
+      "step": 4698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.168038654259127e-05,
+      "loss": 0.3236,
+      "step": 4699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1678596993557626e-05,
+      "loss": 0.2934,
+      "step": 4700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.167680744452398e-05,
+      "loss": 0.3963,
+      "step": 4701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1675017895490337e-05,
+      "loss": 0.2905,
+      "step": 4702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.167322834645669e-05,
+      "loss": 0.2291,
+      "step": 4703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1671438797423054e-05,
+      "loss": 0.4259,
+      "step": 4704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.166964924838941e-05,
+      "loss": 0.3485,
+      "step": 4705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1667859699355764e-05,
+      "loss": 0.3764,
+      "step": 4706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1666070150322126e-05,
+      "loss": 0.3903,
+      "step": 4707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1664280601288474e-05,
+      "loss": 0.2515,
+      "step": 4708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.166249105225483e-05,
+      "loss": 0.2583,
+      "step": 4709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.166070150322119e-05,
+      "loss": 0.3145,
+      "step": 4710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.165891195418755e-05,
+      "loss": 0.285,
+      "step": 4711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.16571224051539e-05,
+      "loss": 0.3673,
+      "step": 4712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.165533285612026e-05,
+      "loss": 0.4474,
+      "step": 4713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.165354330708662e-05,
+      "loss": 0.4091,
+      "step": 4714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1651753758052974e-05,
+      "loss": 0.4272,
+      "step": 4715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.164996420901933e-05,
+      "loss": 0.2865,
+      "step": 4716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1648174659985685e-05,
+      "loss": 0.4715,
+      "step": 4717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.164638511095204e-05,
+      "loss": 0.4914,
+      "step": 4718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1644595561918395e-05,
+      "loss": 0.2195,
+      "step": 4719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.164280601288476e-05,
+      "loss": 0.3659,
+      "step": 4720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.164101646385111e-05,
+      "loss": 0.4853,
+      "step": 4721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.163922691481747e-05,
+      "loss": 0.2852,
+      "step": 4722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.163743736578383e-05,
+      "loss": 0.4936,
+      "step": 4723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.163564781675018e-05,
+      "loss": 0.3976,
+      "step": 4724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.163385826771653e-05,
+      "loss": 0.2182,
+      "step": 4725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1632068718682895e-05,
+      "loss": 0.4623,
+      "step": 4726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.163027916964925e-05,
+      "loss": 0.3053,
+      "step": 4727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1628489620615605e-05,
+      "loss": 0.3362,
+      "step": 4728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.162670007158197e-05,
+      "loss": 0.3107,
+      "step": 4729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.162491052254832e-05,
+      "loss": 0.6064,
+      "step": 4730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.162312097351468e-05,
+      "loss": 0.324,
+      "step": 4731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.162133142448103e-05,
+      "loss": 0.2573,
+      "step": 4732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161954187544739e-05,
+      "loss": 0.4213,
+      "step": 4733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161775232641374e-05,
+      "loss": 0.2831,
+      "step": 4734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1615962777380105e-05,
+      "loss": 0.3952,
+      "step": 4735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161417322834646e-05,
+      "loss": 0.4008,
+      "step": 4736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1612383679312816e-05,
+      "loss": 0.4222,
+      "step": 4737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161059413027917e-05,
+      "loss": 0.2642,
+      "step": 4738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.160880458124553e-05,
+      "loss": 0.3455,
+      "step": 4739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.160701503221188e-05,
+      "loss": 0.4173,
+      "step": 4740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1605225483178236e-05,
+      "loss": 0.4065,
+      "step": 4741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.16034359341446e-05,
+      "loss": 0.2769,
+      "step": 4742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1601646385110954e-05,
+      "loss": 0.3454,
+      "step": 4743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.159985683607731e-05,
+      "loss": 0.2195,
+      "step": 4744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.159806728704367e-05,
+      "loss": 0.2406,
+      "step": 4745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1596277738010026e-05,
+      "loss": 0.2888,
+      "step": 4746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.159448818897638e-05,
+      "loss": 0.2604,
+      "step": 4747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1592698639942736e-05,
+      "loss": 0.3617,
+      "step": 4748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.159090909090909e-05,
+      "loss": 0.3774,
+      "step": 4749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.158911954187545e-05,
+      "loss": 0.296,
+      "step": 4750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.158732999284181e-05,
+      "loss": 0.3853,
+      "step": 4751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1585540443808164e-05,
+      "loss": 0.3031,
+      "step": 4752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.158375089477452e-05,
+      "loss": 0.388,
+      "step": 4753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1581961345740874e-05,
+      "loss": 0.2833,
+      "step": 4754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.158017179670723e-05,
+      "loss": 0.325,
+      "step": 4755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1578382247673585e-05,
+      "loss": 0.2165,
+      "step": 4756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.157659269863995e-05,
+      "loss": 0.2275,
+      "step": 4757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.15748031496063e-05,
+      "loss": 0.3827,
+      "step": 4758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.157301360057266e-05,
+      "loss": 0.2648,
+      "step": 4759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.157122405153901e-05,
+      "loss": 0.376,
+      "step": 4760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1569434502505374e-05,
+      "loss": 0.3745,
+      "step": 4761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.156764495347173e-05,
+      "loss": 0.2025,
+      "step": 4762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.156585540443808e-05,
+      "loss": 0.4313,
+      "step": 4763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.156406585540444e-05,
+      "loss": 0.3366,
+      "step": 4764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1562276306370795e-05,
+      "loss": 0.3091,
+      "step": 4765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.156048675733715e-05,
+      "loss": 0.3271,
+      "step": 4766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.155869720830351e-05,
+      "loss": 0.2571,
+      "step": 4767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.155690765926987e-05,
+      "loss": 0.2442,
+      "step": 4768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.155511811023622e-05,
+      "loss": 0.3786,
+      "step": 4769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1553328561202585e-05,
+      "loss": 0.3536,
+      "step": 4770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.155153901216893e-05,
+      "loss": 0.2491,
+      "step": 4771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.154974946313529e-05,
+      "loss": 0.2593,
+      "step": 4772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.154795991410165e-05,
+      "loss": 0.2705,
+      "step": 4773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1546170365068005e-05,
+      "loss": 0.2292,
+      "step": 4774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.154438081603436e-05,
+      "loss": 0.3561,
+      "step": 4775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1542591267000716e-05,
+      "loss": 0.2889,
+      "step": 4776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.154080171796708e-05,
+      "loss": 0.2789,
+      "step": 4777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.153901216893343e-05,
+      "loss": 0.3169,
+      "step": 4778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.153722261989979e-05,
+      "loss": 0.2364,
+      "step": 4779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.153543307086614e-05,
+      "loss": 0.3732,
+      "step": 4780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.15336435218325e-05,
+      "loss": 0.2511,
+      "step": 4781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1531853972798854e-05,
+      "loss": 0.3352,
+      "step": 4782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1530064423765216e-05,
+      "loss": 0.1805,
+      "step": 4783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.152827487473157e-05,
+      "loss": 0.2518,
+      "step": 4784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1526485325697926e-05,
+      "loss": 0.2093,
+      "step": 4785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.152469577666429e-05,
+      "loss": 0.2022,
+      "step": 4786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1522906227630636e-05,
+      "loss": 0.4723,
+      "step": 4787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.152111667859699e-05,
+      "loss": 0.1853,
+      "step": 4788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1519327129563354e-05,
+      "loss": 0.3174,
+      "step": 4789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.151753758052971e-05,
+      "loss": 0.318,
+      "step": 4790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1515748031496064e-05,
+      "loss": 0.3956,
+      "step": 4791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1513958482462426e-05,
+      "loss": 0.2681,
+      "step": 4792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.151216893342878e-05,
+      "loss": 0.3457,
+      "step": 4793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1510379384395136e-05,
+      "loss": 0.2009,
+      "step": 4794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150858983536149e-05,
+      "loss": 0.3118,
+      "step": 4795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150680028632785e-05,
+      "loss": 0.2636,
+      "step": 4796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.15050107372942e-05,
+      "loss": 0.3413,
+      "step": 4797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150322118826056e-05,
+      "loss": 0.3485,
+      "step": 4798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150143163922692e-05,
+      "loss": 0.17,
+      "step": 4799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1499642090193274e-05,
+      "loss": 0.372,
+      "step": 4800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.149785254115963e-05,
+      "loss": 0.3029,
+      "step": 4801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1496062992125985e-05,
+      "loss": 0.3656,
+      "step": 4802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.149427344309234e-05,
+      "loss": 0.2828,
+      "step": 4803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1492483894058695e-05,
+      "loss": 0.418,
+      "step": 4804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.149069434502506e-05,
+      "loss": 0.3582,
+      "step": 4805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.148890479599141e-05,
+      "loss": 0.335,
+      "step": 4806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.148711524695777e-05,
+      "loss": 0.3786,
+      "step": 4807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.148532569792413e-05,
+      "loss": 0.2727,
+      "step": 4808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1483536148890484e-05,
+      "loss": 0.37,
+      "step": 4809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.148174659985683e-05,
+      "loss": 0.4939,
+      "step": 4810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1479957050823195e-05,
+      "loss": 0.5665,
+      "step": 4811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.147816750178955e-05,
+      "loss": 0.2781,
+      "step": 4812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1476377952755905e-05,
+      "loss": 0.402,
+      "step": 4813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.147458840372227e-05,
+      "loss": 0.3441,
+      "step": 4814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.147279885468862e-05,
+      "loss": 0.154,
+      "step": 4815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.147100930565498e-05,
+      "loss": 0.2955,
+      "step": 4816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.146921975662133e-05,
+      "loss": 0.3624,
+      "step": 4817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.146743020758769e-05,
+      "loss": 0.245,
+      "step": 4818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.146564065855404e-05,
+      "loss": 0.2722,
+      "step": 4819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1463851109520405e-05,
+      "loss": 0.2967,
+      "step": 4820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.146206156048676e-05,
+      "loss": 0.4907,
+      "step": 4821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1460272011453116e-05,
+      "loss": 0.2728,
+      "step": 4822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.145848246241947e-05,
+      "loss": 0.2776,
+      "step": 4823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.145669291338583e-05,
+      "loss": 0.4098,
+      "step": 4824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.145490336435219e-05,
+      "loss": 0.2602,
+      "step": 4825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1453113815318536e-05,
+      "loss": 0.3085,
+      "step": 4826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.14513242662849e-05,
+      "loss": 0.3451,
+      "step": 4827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1449534717251253e-05,
+      "loss": 0.3339,
+      "step": 4828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.144774516821761e-05,
+      "loss": 0.4655,
+      "step": 4829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.144595561918397e-05,
+      "loss": 0.3116,
+      "step": 4830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1444166070150326e-05,
+      "loss": 0.3121,
+      "step": 4831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.144237652111668e-05,
+      "loss": 0.3268,
+      "step": 4832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.144058697208304e-05,
+      "loss": 0.2913,
+      "step": 4833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.143879742304939e-05,
+      "loss": 0.3064,
+      "step": 4834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1437007874015747e-05,
+      "loss": 0.3941,
+      "step": 4835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.143521832498211e-05,
+      "loss": 0.2765,
+      "step": 4836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1433428775948464e-05,
+      "loss": 0.2335,
+      "step": 4837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.143163922691482e-05,
+      "loss": 0.3588,
+      "step": 4838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1429849677881174e-05,
+      "loss": 0.4175,
+      "step": 4839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1428060128847536e-05,
+      "loss": 0.5924,
+      "step": 4840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1426270579813885e-05,
+      "loss": 0.2728,
+      "step": 4841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1424481030780246e-05,
+      "loss": 0.4979,
+      "step": 4842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14226914817466e-05,
+      "loss": 0.3459,
+      "step": 4843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.142090193271296e-05,
+      "loss": 0.4149,
+      "step": 4844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.141911238367931e-05,
+      "loss": 0.3425,
+      "step": 4845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1417322834645674e-05,
+      "loss": 0.2064,
+      "step": 4846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.141553328561203e-05,
+      "loss": 0.3552,
+      "step": 4847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1413743736578384e-05,
+      "loss": 0.3055,
+      "step": 4848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.141195418754474e-05,
+      "loss": 0.4002,
+      "step": 4849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1410164638511095e-05,
+      "loss": 0.3637,
+      "step": 4850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140837508947745e-05,
+      "loss": 0.3301,
+      "step": 4851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140658554044381e-05,
+      "loss": 0.3332,
+      "step": 4852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140479599141017e-05,
+      "loss": 0.2642,
+      "step": 4853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140300644237652e-05,
+      "loss": 0.3272,
+      "step": 4854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1401216893342884e-05,
+      "loss": 0.2192,
+      "step": 4855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139942734430924e-05,
+      "loss": 0.2751,
+      "step": 4856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139763779527559e-05,
+      "loss": 0.2619,
+      "step": 4857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139584824624195e-05,
+      "loss": 0.263,
+      "step": 4858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1394058697208305e-05,
+      "loss": 0.2416,
+      "step": 4859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139226914817466e-05,
+      "loss": 0.3414,
+      "step": 4860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1390479599141015e-05,
+      "loss": 0.391,
+      "step": 4861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.138869005010738e-05,
+      "loss": 0.2567,
+      "step": 4862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.138690050107373e-05,
+      "loss": 0.2506,
+      "step": 4863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.138511095204009e-05,
+      "loss": 0.3555,
+      "step": 4864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.138332140300644e-05,
+      "loss": 0.3039,
+      "step": 4865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.13815318539728e-05,
+      "loss": 0.3268,
+      "step": 4866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1379742304939153e-05,
+      "loss": 0.2624,
+      "step": 4867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1377952755905515e-05,
+      "loss": 0.264,
+      "step": 4868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.137616320687187e-05,
+      "loss": 0.4346,
+      "step": 4869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1374373657838226e-05,
+      "loss": 0.254,
+      "step": 4870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.137258410880459e-05,
+      "loss": 0.3467,
+      "step": 4871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.137079455977094e-05,
+      "loss": 0.3651,
+      "step": 4872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136900501073729e-05,
+      "loss": 0.2455,
+      "step": 4873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136721546170365e-05,
+      "loss": 0.2878,
+      "step": 4874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136542591267001e-05,
+      "loss": 0.2803,
+      "step": 4875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1363636363636364e-05,
+      "loss": 0.3244,
+      "step": 4876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1361846814602726e-05,
+      "loss": 0.2579,
+      "step": 4877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136005726556908e-05,
+      "loss": 0.4437,
+      "step": 4878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1358267716535436e-05,
+      "loss": 0.3047,
+      "step": 4879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.135647816750179e-05,
+      "loss": 0.3104,
+      "step": 4880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1354688618468146e-05,
+      "loss": 0.2113,
+      "step": 4881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.13528990694345e-05,
+      "loss": 0.3788,
+      "step": 4882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1351109520400864e-05,
+      "loss": 0.2554,
+      "step": 4883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134931997136722e-05,
+      "loss": 0.1747,
+      "step": 4884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1347530422333574e-05,
+      "loss": 0.3607,
+      "step": 4885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134574087329993e-05,
+      "loss": 0.2066,
+      "step": 4886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134395132426629e-05,
+      "loss": 0.3164,
+      "step": 4887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134216177523264e-05,
+      "loss": 0.2875,
+      "step": 4888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1340372226198995e-05,
+      "loss": 0.1711,
+      "step": 4889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.133858267716536e-05,
+      "loss": 0.3092,
+      "step": 4890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.133679312813171e-05,
+      "loss": 0.3463,
+      "step": 4891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.133500357909807e-05,
+      "loss": 0.2575,
+      "step": 4892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.133321403006443e-05,
+      "loss": 0.2731,
+      "step": 4893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1331424481030784e-05,
+      "loss": 0.3031,
+      "step": 4894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.132963493199714e-05,
+      "loss": 0.2591,
+      "step": 4895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1327845382963495e-05,
+      "loss": 0.3744,
+      "step": 4896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.132605583392985e-05,
+      "loss": 0.2331,
+      "step": 4897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1324266284896205e-05,
+      "loss": 0.4449,
+      "step": 4898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.132247673586257e-05,
+      "loss": 0.262,
+      "step": 4899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.132068718682892e-05,
+      "loss": 0.3081,
+      "step": 4900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.131889763779528e-05,
+      "loss": 0.3247,
+      "step": 4901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.131710808876163e-05,
+      "loss": 0.328,
+      "step": 4902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1315318539727995e-05,
+      "loss": 0.3569,
+      "step": 4903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.131352899069434e-05,
+      "loss": 0.3287,
+      "step": 4904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1311739441660705e-05,
+      "loss": 0.249,
+      "step": 4905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130994989262706e-05,
+      "loss": 0.2118,
+      "step": 4906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1308160343593415e-05,
+      "loss": 0.411,
+      "step": 4907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130637079455977e-05,
+      "loss": 0.235,
+      "step": 4908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130458124552613e-05,
+      "loss": 0.3085,
+      "step": 4909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130279169649249e-05,
+      "loss": 0.3532,
+      "step": 4910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130100214745884e-05,
+      "loss": 0.3156,
+      "step": 4911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.12992125984252e-05,
+      "loss": 0.4322,
+      "step": 4912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129742304939155e-05,
+      "loss": 0.3302,
+      "step": 4913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129563350035791e-05,
+      "loss": 0.354,
+      "step": 4914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129384395132427e-05,
+      "loss": 0.2776,
+      "step": 4915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1292054402290626e-05,
+      "loss": 0.3105,
+      "step": 4916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129026485325698e-05,
+      "loss": 0.3456,
+      "step": 4917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.128847530422334e-05,
+      "loss": 0.3529,
+      "step": 4918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.12866857551897e-05,
+      "loss": 0.3099,
+      "step": 4919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1284896206156046e-05,
+      "loss": 0.2708,
+      "step": 4920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.128310665712241e-05,
+      "loss": 0.318,
+      "step": 4921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1281317108088764e-05,
+      "loss": 0.3016,
+      "step": 4922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.127952755905512e-05,
+      "loss": 0.2356,
+      "step": 4923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1277738010021474e-05,
+      "loss": 0.3475,
+      "step": 4924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1275948460987836e-05,
+      "loss": 0.3164,
+      "step": 4925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.127415891195419e-05,
+      "loss": 0.2913,
+      "step": 4926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1272369362920546e-05,
+      "loss": 0.3402,
+      "step": 4927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.12705798138869e-05,
+      "loss": 0.3042,
+      "step": 4928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.126879026485326e-05,
+      "loss": 0.3735,
+      "step": 4929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.126700071581961e-05,
+      "loss": 0.3532,
+      "step": 4930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1265211166785974e-05,
+      "loss": 0.289,
+      "step": 4931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.126342161775233e-05,
+      "loss": 0.3369,
+      "step": 4932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1261632068718684e-05,
+      "loss": 0.5121,
+      "step": 4933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1259842519685046e-05,
+      "loss": 0.4321,
+      "step": 4934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1258052970651395e-05,
+      "loss": 0.3098,
+      "step": 4935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.125626342161775e-05,
+      "loss": 0.3019,
+      "step": 4936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.125447387258411e-05,
+      "loss": 0.3105,
+      "step": 4937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.125268432355047e-05,
+      "loss": 0.4136,
+      "step": 4938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.125089477451682e-05,
+      "loss": 0.3662,
+      "step": 4939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1249105225483184e-05,
+      "loss": 0.2481,
+      "step": 4940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.124731567644954e-05,
+      "loss": 0.3042,
+      "step": 4941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1245526127415895e-05,
+      "loss": 0.1787,
+      "step": 4942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.124373657838225e-05,
+      "loss": 0.2385,
+      "step": 4943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1241947029348605e-05,
+      "loss": 0.4101,
+      "step": 4944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.124015748031496e-05,
+      "loss": 0.3832,
+      "step": 4945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123836793128132e-05,
+      "loss": 0.278,
+      "step": 4946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123657838224768e-05,
+      "loss": 0.3656,
+      "step": 4947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123478883321403e-05,
+      "loss": 0.3081,
+      "step": 4948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123299928418039e-05,
+      "loss": 0.3324,
+      "step": 4949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.123120973514675e-05,
+      "loss": 0.2539,
+      "step": 4950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.12294201861131e-05,
+      "loss": 0.3459,
+      "step": 4951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.122763063707945e-05,
+      "loss": 0.2419,
+      "step": 4952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1225841088045815e-05,
+      "loss": 0.3814,
+      "step": 4953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.122405153901217e-05,
+      "loss": 0.342,
+      "step": 4954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1222261989978526e-05,
+      "loss": 0.2782,
+      "step": 4955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.122047244094489e-05,
+      "loss": 0.3653,
+      "step": 4956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121868289191124e-05,
+      "loss": 0.3261,
+      "step": 4957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.12168933428776e-05,
+      "loss": 0.3939,
+      "step": 4958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121510379384395e-05,
+      "loss": 0.3045,
+      "step": 4959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121331424481031e-05,
+      "loss": 0.3013,
+      "step": 4960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1211524695776664e-05,
+      "loss": 0.3129,
+      "step": 4961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1209735146743025e-05,
+      "loss": 0.3954,
+      "step": 4962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.120794559770938e-05,
+      "loss": 0.3077,
+      "step": 4963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1206156048675736e-05,
+      "loss": 0.3039,
+      "step": 4964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.120436649964209e-05,
+      "loss": 0.2296,
+      "step": 4965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.120257695060845e-05,
+      "loss": 0.3168,
+      "step": 4966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.12007874015748e-05,
+      "loss": 0.4133,
+      "step": 4967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1198997852541163e-05,
+      "loss": 0.3838,
+      "step": 4968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.119720830350752e-05,
+      "loss": 0.434,
+      "step": 4969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1195418754473874e-05,
+      "loss": 0.2235,
+      "step": 4970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.119362920544023e-05,
+      "loss": 0.3331,
+      "step": 4971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.119183965640659e-05,
+      "loss": 0.4356,
+      "step": 4972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1190050107372946e-05,
+      "loss": 0.286,
+      "step": 4973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1188260558339295e-05,
+      "loss": 0.2916,
+      "step": 4974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1186471009305657e-05,
+      "loss": 0.2741,
+      "step": 4975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.118468146027201e-05,
+      "loss": 0.2748,
+      "step": 4976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.118289191123837e-05,
+      "loss": 0.2252,
+      "step": 4977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.118110236220473e-05,
+      "loss": 0.3069,
+      "step": 4978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1179312813171084e-05,
+      "loss": 0.3075,
+      "step": 4979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.117752326413744e-05,
+      "loss": 0.2848,
+      "step": 4980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.11757337151038e-05,
+      "loss": 0.4122,
+      "step": 4981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.117394416607015e-05,
+      "loss": 0.2926,
+      "step": 4982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1172154617036505e-05,
+      "loss": 0.2557,
+      "step": 4983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.117036506800287e-05,
+      "loss": 0.3845,
+      "step": 4984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.116857551896922e-05,
+      "loss": 0.2538,
+      "step": 4985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.116678596993558e-05,
+      "loss": 0.2802,
+      "step": 4986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.116499642090193e-05,
+      "loss": 0.2736,
+      "step": 4987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1163206871868294e-05,
+      "loss": 0.2935,
+      "step": 4988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.116141732283465e-05,
+      "loss": 0.3856,
+      "step": 4989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1159627773801005e-05,
+      "loss": 0.2368,
+      "step": 4990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.115783822476736e-05,
+      "loss": 0.336,
+      "step": 4991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1156048675733715e-05,
+      "loss": 0.2407,
+      "step": 4992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.115425912670007e-05,
+      "loss": 0.4064,
+      "step": 4993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.115246957766643e-05,
+      "loss": 0.406,
+      "step": 4994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.115068002863279e-05,
+      "loss": 0.3237,
+      "step": 4995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.114889047959914e-05,
+      "loss": 0.3616,
+      "step": 4996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1147100930565505e-05,
+      "loss": 0.319,
+      "step": 4997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.114531138153185e-05,
+      "loss": 0.2403,
+      "step": 4998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.114352183249821e-05,
+      "loss": 0.3344,
+      "step": 4999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.114173228346457e-05,
+      "loss": 0.2871,
+      "step": 5000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1139942734430925e-05,
+      "loss": 0.2941,
+      "step": 5001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113815318539728e-05,
+      "loss": 0.3163,
+      "step": 5002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113636363636364e-05,
+      "loss": 0.3208,
+      "step": 5003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113457408733e-05,
+      "loss": 0.2854,
+      "step": 5004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113278453829635e-05,
+      "loss": 0.2479,
+      "step": 5005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.113099498926271e-05,
+      "loss": 0.3373,
+      "step": 5006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.112920544022906e-05,
+      "loss": 0.2951,
+      "step": 5007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.112741589119542e-05,
+      "loss": 0.2165,
+      "step": 5008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1125626342161774e-05,
+      "loss": 0.2192,
+      "step": 5009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1123836793128136e-05,
+      "loss": 0.2838,
+      "step": 5010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.112204724409449e-05,
+      "loss": 0.3156,
+      "step": 5011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1120257695060846e-05,
+      "loss": 0.2819,
+      "step": 5012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.111846814602721e-05,
+      "loss": 0.2286,
+      "step": 5013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1116678596993557e-05,
+      "loss": 0.2345,
+      "step": 5014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.111488904795991e-05,
+      "loss": 0.3091,
+      "step": 5015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1113099498926274e-05,
+      "loss": 0.2389,
+      "step": 5016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.111130994989263e-05,
+      "loss": 0.341,
+      "step": 5017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1109520400858984e-05,
+      "loss": 0.2388,
+      "step": 5018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1107730851825346e-05,
+      "loss": 0.3945,
+      "step": 5019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.11059413027917e-05,
+      "loss": 0.2465,
+      "step": 5020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.110415175375805e-05,
+      "loss": 0.2977,
+      "step": 5021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.110236220472441e-05,
+      "loss": 0.3309,
+      "step": 5022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.110057265569077e-05,
+      "loss": 0.2995,
+      "step": 5023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.109878310665712e-05,
+      "loss": 0.3728,
+      "step": 5024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1096993557623484e-05,
+      "loss": 0.5126,
+      "step": 5025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.109520400858984e-05,
+      "loss": 0.1984,
+      "step": 5026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1093414459556194e-05,
+      "loss": 0.3948,
+      "step": 5027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.109162491052255e-05,
+      "loss": 0.2681,
+      "step": 5028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1089835361488905e-05,
+      "loss": 0.2542,
+      "step": 5029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.108804581245526e-05,
+      "loss": 0.3161,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.108625626342162e-05,
+      "loss": 0.2612,
+      "step": 5031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.108446671438798e-05,
+      "loss": 0.167,
+      "step": 5032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.108267716535433e-05,
+      "loss": 0.2972,
+      "step": 5033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.108088761632069e-05,
+      "loss": 0.1915,
+      "step": 5034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.107909806728705e-05,
+      "loss": 0.3304,
+      "step": 5035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1077308518253405e-05,
+      "loss": 0.4795,
+      "step": 5036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.107551896921975e-05,
+      "loss": 0.3726,
+      "step": 5037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1073729420186115e-05,
+      "loss": 0.2641,
+      "step": 5038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.107193987115247e-05,
+      "loss": 0.3413,
+      "step": 5039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1070150322118825e-05,
+      "loss": 0.2905,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.106836077308519e-05,
+      "loss": 0.2073,
+      "step": 5041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.106657122405154e-05,
+      "loss": 0.2697,
+      "step": 5042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.10647816750179e-05,
+      "loss": 0.2576,
+      "step": 5043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.106299212598426e-05,
+      "loss": 0.2833,
+      "step": 5044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.106120257695061e-05,
+      "loss": 0.3233,
+      "step": 5045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.105941302791696e-05,
+      "loss": 0.3082,
+      "step": 5046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1057623478883325e-05,
+      "loss": 0.3427,
+      "step": 5047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.105583392984968e-05,
+      "loss": 0.1381,
+      "step": 5048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1054044380816036e-05,
+      "loss": 0.2688,
+      "step": 5049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.105225483178239e-05,
+      "loss": 0.2688,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.105046528274875e-05,
+      "loss": 0.2812,
+      "step": 5051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.104867573371511e-05,
+      "loss": 0.2526,
+      "step": 5052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.104688618468146e-05,
+      "loss": 0.3102,
+      "step": 5053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.104509663564782e-05,
+      "loss": 0.2977,
+      "step": 5054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1043307086614174e-05,
+      "loss": 0.3697,
+      "step": 5055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.104151753758053e-05,
+      "loss": 0.2806,
+      "step": 5056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.103972798854689e-05,
+      "loss": 0.2524,
+      "step": 5057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1037938439513246e-05,
+      "loss": 0.3813,
+      "step": 5058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.10361488904796e-05,
+      "loss": 0.1952,
+      "step": 5059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1034359341445956e-05,
+      "loss": 0.2501,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.103256979241231e-05,
+      "loss": 0.4268,
+      "step": 5061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.103078024337867e-05,
+      "loss": 0.3077,
+      "step": 5062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.102899069434503e-05,
+      "loss": 0.1794,
+      "step": 5063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1027201145311384e-05,
+      "loss": 0.2478,
+      "step": 5064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.102541159627774e-05,
+      "loss": 0.2819,
+      "step": 5065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.10236220472441e-05,
+      "loss": 0.3037,
+      "step": 5066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1021832498210456e-05,
+      "loss": 0.3579,
+      "step": 5067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1020042949176805e-05,
+      "loss": 0.177,
+      "step": 5068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101825340014317e-05,
+      "loss": 0.2245,
+      "step": 5069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101646385110952e-05,
+      "loss": 0.1859,
+      "step": 5070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101467430207588e-05,
+      "loss": 0.4628,
+      "step": 5071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101288475304223e-05,
+      "loss": 0.3656,
+      "step": 5072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1011095204008594e-05,
+      "loss": 0.2767,
+      "step": 5073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100930565497495e-05,
+      "loss": 0.2791,
+      "step": 5074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1007516105941305e-05,
+      "loss": 0.1996,
+      "step": 5075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100572655690766e-05,
+      "loss": 0.3014,
+      "step": 5076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1003937007874015e-05,
+      "loss": 0.2036,
+      "step": 5077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100214745884037e-05,
+      "loss": 0.2523,
+      "step": 5078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100035790980673e-05,
+      "loss": 0.2107,
+      "step": 5079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099856836077309e-05,
+      "loss": 0.3546,
+      "step": 5080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099677881173944e-05,
+      "loss": 0.2402,
+      "step": 5081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0994989262705805e-05,
+      "loss": 0.2267,
+      "step": 5082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099319971367216e-05,
+      "loss": 0.2995,
+      "step": 5083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.099141016463851e-05,
+      "loss": 0.2874,
+      "step": 5084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098962061560487e-05,
+      "loss": 0.4516,
+      "step": 5085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0987831066571225e-05,
+      "loss": 0.3661,
+      "step": 5086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098604151753758e-05,
+      "loss": 0.3605,
+      "step": 5087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098425196850394e-05,
+      "loss": 0.2447,
+      "step": 5088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.09824624194703e-05,
+      "loss": 0.3246,
+      "step": 5089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098067287043665e-05,
+      "loss": 0.2913,
+      "step": 5090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097888332140301e-05,
+      "loss": 0.2053,
+      "step": 5091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097709377236936e-05,
+      "loss": 0.2764,
+      "step": 5092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097530422333572e-05,
+      "loss": 0.2657,
+      "step": 5093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097351467430208e-05,
+      "loss": 0.2529,
+      "step": 5094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0971725125268436e-05,
+      "loss": 0.4446,
+      "step": 5095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096993557623479e-05,
+      "loss": 0.2547,
+      "step": 5096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0968146027201146e-05,
+      "loss": 0.2569,
+      "step": 5097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096635647816751e-05,
+      "loss": 0.2938,
+      "step": 5098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096456692913386e-05,
+      "loss": 0.1696,
+      "step": 5099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096277738010021e-05,
+      "loss": 0.4251,
+      "step": 5100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0960987831066573e-05,
+      "loss": 0.215,
+      "step": 5101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.095919828203293e-05,
+      "loss": 0.3146,
+      "step": 5102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0957408732999284e-05,
+      "loss": 0.3397,
+      "step": 5103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0955619183965646e-05,
+      "loss": 0.2377,
+      "step": 5104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0953829634932e-05,
+      "loss": 0.2617,
+      "step": 5105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0952040085898356e-05,
+      "loss": 0.4142,
+      "step": 5106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.095025053686471e-05,
+      "loss": 0.4389,
+      "step": 5107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.094846098783107e-05,
+      "loss": 0.3727,
+      "step": 5108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.094667143879742e-05,
+      "loss": 0.4793,
+      "step": 5109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0944881889763784e-05,
+      "loss": 0.4957,
+      "step": 5110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.094309234073014e-05,
+      "loss": 0.2078,
+      "step": 5111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0941302791696494e-05,
+      "loss": 0.4899,
+      "step": 5112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093951324266285e-05,
+      "loss": 0.2784,
+      "step": 5113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093772369362921e-05,
+      "loss": 0.2272,
+      "step": 5114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093593414459556e-05,
+      "loss": 0.261,
+      "step": 5115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093414459556192e-05,
+      "loss": 0.3111,
+      "step": 5116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093235504652828e-05,
+      "loss": 0.2588,
+      "step": 5117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093056549749463e-05,
+      "loss": 0.4312,
+      "step": 5118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092877594846099e-05,
+      "loss": 0.2419,
+      "step": 5119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092698639942735e-05,
+      "loss": 0.2605,
+      "step": 5120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0925196850393704e-05,
+      "loss": 0.3238,
+      "step": 5121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092340730136006e-05,
+      "loss": 0.2407,
+      "step": 5122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0921617752326415e-05,
+      "loss": 0.2635,
+      "step": 5123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.091982820329277e-05,
+      "loss": 0.2987,
+      "step": 5124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0918038654259125e-05,
+      "loss": 0.325,
+      "step": 5125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.091624910522549e-05,
+      "loss": 0.3953,
+      "step": 5126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.091445955619184e-05,
+      "loss": 0.2252,
+      "step": 5127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.09126700071582e-05,
+      "loss": 0.3878,
+      "step": 5128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.091088045812456e-05,
+      "loss": 0.3309,
+      "step": 5129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.3292,
+      "step": 5130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.090730136005726e-05,
+      "loss": 0.3144,
+      "step": 5131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0905511811023625e-05,
+      "loss": 0.2515,
+      "step": 5132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.090372226198998e-05,
+      "loss": 0.3832,
+      "step": 5133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0901932712956336e-05,
+      "loss": 0.3996,
+      "step": 5134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.090014316392269e-05,
+      "loss": 0.2955,
+      "step": 5135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.089835361488905e-05,
+      "loss": 0.417,
+      "step": 5136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.089656406585541e-05,
+      "loss": 0.2603,
+      "step": 5137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.089477451682176e-05,
+      "loss": 0.3147,
+      "step": 5138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.089298496778812e-05,
+      "loss": 0.2166,
+      "step": 5139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0891195418754473e-05,
+      "loss": 0.298,
+      "step": 5140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.088940586972083e-05,
+      "loss": 0.329,
+      "step": 5141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.088761632068719e-05,
+      "loss": 0.319,
+      "step": 5142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0885826771653546e-05,
+      "loss": 0.2689,
+      "step": 5143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.08840372226199e-05,
+      "loss": 0.3709,
+      "step": 5144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.088224767358626e-05,
+      "loss": 0.4274,
+      "step": 5145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.088045812455261e-05,
+      "loss": 0.3443,
+      "step": 5146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0878668575518967e-05,
+      "loss": 0.3301,
+      "step": 5147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.087687902648533e-05,
+      "loss": 0.2477,
+      "step": 5148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0875089477451684e-05,
+      "loss": 0.2151,
+      "step": 5149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.087329992841804e-05,
+      "loss": 0.276,
+      "step": 5150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.08715103793844e-05,
+      "loss": 0.3634,
+      "step": 5151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0869720830350756e-05,
+      "loss": 0.3547,
+      "step": 5152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086793128131711e-05,
+      "loss": 0.2543,
+      "step": 5153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0866141732283466e-05,
+      "loss": 0.2801,
+      "step": 5154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086435218324982e-05,
+      "loss": 0.2002,
+      "step": 5155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086256263421618e-05,
+      "loss": 0.3517,
+      "step": 5156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086077308518253e-05,
+      "loss": 0.2509,
+      "step": 5157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0858983536148894e-05,
+      "loss": 0.2306,
+      "step": 5158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.085719398711525e-05,
+      "loss": 0.2487,
+      "step": 5159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0855404438081604e-05,
+      "loss": 0.35,
+      "step": 5160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0853614889047966e-05,
+      "loss": 0.3901,
+      "step": 5161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0851825340014315e-05,
+      "loss": 0.3261,
+      "step": 5162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.085003579098067e-05,
+      "loss": 0.3385,
+      "step": 5163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.084824624194703e-05,
+      "loss": 0.3939,
+      "step": 5164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.084645669291339e-05,
+      "loss": 0.2161,
+      "step": 5165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.084466714387974e-05,
+      "loss": 0.3253,
+      "step": 5166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0842877594846104e-05,
+      "loss": 0.2931,
+      "step": 5167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.084108804581246e-05,
+      "loss": 0.4312,
+      "step": 5168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0839298496778815e-05,
+      "loss": 0.3319,
+      "step": 5169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.083750894774517e-05,
+      "loss": 0.2998,
+      "step": 5170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0835719398711525e-05,
+      "loss": 0.2652,
+      "step": 5171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.083392984967788e-05,
+      "loss": 0.2432,
+      "step": 5172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.083214030064424e-05,
+      "loss": 0.2939,
+      "step": 5173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.08303507516106e-05,
+      "loss": 0.3005,
+      "step": 5174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.082856120257695e-05,
+      "loss": 0.3099,
+      "step": 5175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.082677165354331e-05,
+      "loss": 0.2667,
+      "step": 5176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.082498210450967e-05,
+      "loss": 0.3365,
+      "step": 5177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.082319255547602e-05,
+      "loss": 0.4617,
+      "step": 5178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.082140300644238e-05,
+      "loss": 0.3241,
+      "step": 5179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0819613457408735e-05,
+      "loss": 0.3016,
+      "step": 5180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081782390837509e-05,
+      "loss": 0.3225,
+      "step": 5181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0816034359341446e-05,
+      "loss": 0.2199,
+      "step": 5182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081424481030781e-05,
+      "loss": 0.1799,
+      "step": 5183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081245526127416e-05,
+      "loss": 0.161,
+      "step": 5184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081066571224052e-05,
+      "loss": 0.2973,
+      "step": 5185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.080887616320687e-05,
+      "loss": 0.286,
+      "step": 5186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.080708661417323e-05,
+      "loss": 0.2417,
+      "step": 5187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0805297065139584e-05,
+      "loss": 0.1805,
+      "step": 5188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0803507516105946e-05,
+      "loss": 0.2834,
+      "step": 5189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.08017179670723e-05,
+      "loss": 0.2781,
+      "step": 5190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0799928418038656e-05,
+      "loss": 0.2467,
+      "step": 5191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.079813886900502e-05,
+      "loss": 0.4538,
+      "step": 5192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0796349319971366e-05,
+      "loss": 0.1416,
+      "step": 5193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.079455977093772e-05,
+      "loss": 0.277,
+      "step": 5194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0792770221904084e-05,
+      "loss": 0.2142,
+      "step": 5195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.079098067287044e-05,
+      "loss": 0.3344,
+      "step": 5196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0789191123836794e-05,
+      "loss": 0.1992,
+      "step": 5197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.078740157480315e-05,
+      "loss": 0.2954,
+      "step": 5198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.078561202576951e-05,
+      "loss": 0.2469,
+      "step": 5199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0783822476735866e-05,
+      "loss": 0.3905,
+      "step": 5200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.078203292770222e-05,
+      "loss": 0.3733,
+      "step": 5201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.078024337866858e-05,
+      "loss": 0.256,
+      "step": 5202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.077845382963493e-05,
+      "loss": 0.2579,
+      "step": 5203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.077666428060129e-05,
+      "loss": 0.1874,
+      "step": 5204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.077487473156765e-05,
+      "loss": 0.3548,
+      "step": 5205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0773085182534004e-05,
+      "loss": 0.1838,
+      "step": 5206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.077129563350036e-05,
+      "loss": 0.3914,
+      "step": 5207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.076950608446672e-05,
+      "loss": 0.3088,
+      "step": 5208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.076771653543307e-05,
+      "loss": 0.2244,
+      "step": 5209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0765926986399425e-05,
+      "loss": 0.274,
+      "step": 5210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.076413743736579e-05,
+      "loss": 0.261,
+      "step": 5211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.076234788833214e-05,
+      "loss": 0.3126,
+      "step": 5212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.07605583392985e-05,
+      "loss": 0.3316,
+      "step": 5213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.075876879026486e-05,
+      "loss": 0.2006,
+      "step": 5214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0756979241231215e-05,
+      "loss": 0.3722,
+      "step": 5215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.075518969219757e-05,
+      "loss": 0.4019,
+      "step": 5216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0753400143163925e-05,
+      "loss": 0.3618,
+      "step": 5217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.075161059413028e-05,
+      "loss": 0.2629,
+      "step": 5218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0749821045096635e-05,
+      "loss": 0.3729,
+      "step": 5219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.074803149606299e-05,
+      "loss": 0.2676,
+      "step": 5220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.074624194702935e-05,
+      "loss": 0.3818,
+      "step": 5221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.074445239799571e-05,
+      "loss": 0.3007,
+      "step": 5222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.074266284896206e-05,
+      "loss": 0.3269,
+      "step": 5223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0740873299928425e-05,
+      "loss": 0.2084,
+      "step": 5224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.073908375089477e-05,
+      "loss": 0.4046,
+      "step": 5225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.073729420186113e-05,
+      "loss": 0.3092,
+      "step": 5226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.073550465282749e-05,
+      "loss": 0.2069,
+      "step": 5227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0733715103793846e-05,
+      "loss": 0.2005,
+      "step": 5228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.07319255547602e-05,
+      "loss": 0.4062,
+      "step": 5229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.073013600572656e-05,
+      "loss": 0.3099,
+      "step": 5230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.072834645669292e-05,
+      "loss": 0.2269,
+      "step": 5231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.072655690765927e-05,
+      "loss": 0.2977,
+      "step": 5232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.072476735862563e-05,
+      "loss": 0.3091,
+      "step": 5233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0722977809591984e-05,
+      "loss": 0.2967,
+      "step": 5234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.072118826055834e-05,
+      "loss": 0.3325,
+      "step": 5235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.07193987115247e-05,
+      "loss": 0.3932,
+      "step": 5236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0717609162491056e-05,
+      "loss": 0.3853,
+      "step": 5237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071581961345741e-05,
+      "loss": 0.3287,
+      "step": 5238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0714030064423766e-05,
+      "loss": 0.3608,
+      "step": 5239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071224051539012e-05,
+      "loss": 0.2334,
+      "step": 5240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071045096635648e-05,
+      "loss": 0.3005,
+      "step": 5241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.070866141732284e-05,
+      "loss": 0.4121,
+      "step": 5242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0706871868289194e-05,
+      "loss": 0.2283,
+      "step": 5243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.070508231925555e-05,
+      "loss": 0.2553,
+      "step": 5244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0703292770221904e-05,
+      "loss": 0.3457,
+      "step": 5245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0701503221188266e-05,
+      "loss": 0.23,
+      "step": 5246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069971367215462e-05,
+      "loss": 0.3082,
+      "step": 5247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069792412312097e-05,
+      "loss": 0.299,
+      "step": 5248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069613457408733e-05,
+      "loss": 0.3056,
+      "step": 5249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069434502505369e-05,
+      "loss": 0.3076,
+      "step": 5250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069255547602004e-05,
+      "loss": 0.1714,
+      "step": 5251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0690765926986404e-05,
+      "loss": 0.1818,
+      "step": 5252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.068897637795276e-05,
+      "loss": 0.2673,
+      "step": 5253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0687186828919115e-05,
+      "loss": 0.3349,
+      "step": 5254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0685397279885476e-05,
+      "loss": 0.351,
+      "step": 5255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0683607730851825e-05,
+      "loss": 0.415,
+      "step": 5256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.068181818181818e-05,
+      "loss": 0.2712,
+      "step": 5257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.068002863278454e-05,
+      "loss": 0.2708,
+      "step": 5258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.06782390837509e-05,
+      "loss": 0.3301,
+      "step": 5259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.067644953471725e-05,
+      "loss": 0.305,
+      "step": 5260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.067465998568361e-05,
+      "loss": 0.265,
+      "step": 5261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.067287043664997e-05,
+      "loss": 0.1835,
+      "step": 5262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0671080887616325e-05,
+      "loss": 0.3,
+      "step": 5263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.066929133858268e-05,
+      "loss": 0.228,
+      "step": 5264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0667501789549035e-05,
+      "loss": 0.277,
+      "step": 5265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.066571224051539e-05,
+      "loss": 0.1877,
+      "step": 5266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0663922691481746e-05,
+      "loss": 0.2535,
+      "step": 5267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.066213314244811e-05,
+      "loss": 0.28,
+      "step": 5268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.066034359341446e-05,
+      "loss": 0.1935,
+      "step": 5269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.065855404438082e-05,
+      "loss": 0.3056,
+      "step": 5270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.065676449534718e-05,
+      "loss": 0.2006,
+      "step": 5271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.065497494631353e-05,
+      "loss": 0.289,
+      "step": 5272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0653185397279884e-05,
+      "loss": 0.2569,
+      "step": 5273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0651395848246245e-05,
+      "loss": 0.3915,
+      "step": 5274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.06496062992126e-05,
+      "loss": 0.2646,
+      "step": 5275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0647816750178956e-05,
+      "loss": 0.1772,
+      "step": 5276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.064602720114532e-05,
+      "loss": 0.2351,
+      "step": 5277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.064423765211167e-05,
+      "loss": 0.3269,
+      "step": 5278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.064244810307802e-05,
+      "loss": 0.2593,
+      "step": 5279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0640658554044383e-05,
+      "loss": 0.2517,
+      "step": 5280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.063886900501074e-05,
+      "loss": 0.2374,
+      "step": 5281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0637079455977094e-05,
+      "loss": 0.3157,
+      "step": 5282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.063528990694345e-05,
+      "loss": 0.2114,
+      "step": 5283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.063350035790981e-05,
+      "loss": 0.3065,
+      "step": 5284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0631710808876166e-05,
+      "loss": 0.2054,
+      "step": 5285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062992125984252e-05,
+      "loss": 0.392,
+      "step": 5286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0628131710808877e-05,
+      "loss": 0.2122,
+      "step": 5287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062634216177523e-05,
+      "loss": 0.2485,
+      "step": 5288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062455261274159e-05,
+      "loss": 0.1668,
+      "step": 5289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.062276306370795e-05,
+      "loss": 0.3494,
+      "step": 5290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0620973514674304e-05,
+      "loss": 0.1609,
+      "step": 5291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061918396564066e-05,
+      "loss": 0.2659,
+      "step": 5292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061739441660702e-05,
+      "loss": 0.3029,
+      "step": 5293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0615604867573376e-05,
+      "loss": 0.1853,
+      "step": 5294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0613815318539725e-05,
+      "loss": 0.366,
+      "step": 5295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061202576950609e-05,
+      "loss": 0.4177,
+      "step": 5296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061023622047244e-05,
+      "loss": 0.2359,
+      "step": 5297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.06084466714388e-05,
+      "loss": 0.2926,
+      "step": 5298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.060665712240516e-05,
+      "loss": 0.2124,
+      "step": 5299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0604867573371514e-05,
+      "loss": 0.266,
+      "step": 5300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.060307802433787e-05,
+      "loss": 0.3947,
+      "step": 5301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0601288475304225e-05,
+      "loss": 0.174,
+      "step": 5302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.059949892627058e-05,
+      "loss": 0.248,
+      "step": 5303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0597709377236935e-05,
+      "loss": 0.2976,
+      "step": 5304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.05959198282033e-05,
+      "loss": 0.2534,
+      "step": 5305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.059413027916965e-05,
+      "loss": 0.1394,
+      "step": 5306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.059234073013601e-05,
+      "loss": 0.222,
+      "step": 5307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.059055118110236e-05,
+      "loss": 0.3934,
+      "step": 5308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0588761632068725e-05,
+      "loss": 0.241,
+      "step": 5309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.058697208303508e-05,
+      "loss": 0.1873,
+      "step": 5310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.058518253400143e-05,
+      "loss": 0.2823,
+      "step": 5311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.058339298496779e-05,
+      "loss": 0.4332,
+      "step": 5312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0581603435934145e-05,
+      "loss": 0.3432,
+      "step": 5313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.05798138869005e-05,
+      "loss": 0.3712,
+      "step": 5314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.057802433786686e-05,
+      "loss": 0.3086,
+      "step": 5315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.057623478883322e-05,
+      "loss": 0.3298,
+      "step": 5316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.057444523979957e-05,
+      "loss": 0.3716,
+      "step": 5317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0572655690765935e-05,
+      "loss": 0.2934,
+      "step": 5318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.057086614173228e-05,
+      "loss": 0.264,
+      "step": 5319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.056907659269864e-05,
+      "loss": 0.2787,
+      "step": 5320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0567287043665e-05,
+      "loss": 0.368,
+      "step": 5321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0565497494631356e-05,
+      "loss": 0.2888,
+      "step": 5322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.056370794559771e-05,
+      "loss": 0.3533,
+      "step": 5323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0561918396564066e-05,
+      "loss": 0.3074,
+      "step": 5324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.056012884753043e-05,
+      "loss": 0.3531,
+      "step": 5325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0558339298496777e-05,
+      "loss": 0.2615,
+      "step": 5326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.055654974946314e-05,
+      "loss": 0.3656,
+      "step": 5327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0554760200429494e-05,
+      "loss": 0.266,
+      "step": 5328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.055297065139585e-05,
+      "loss": 0.3218,
+      "step": 5329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0551181102362204e-05,
+      "loss": 0.3621,
+      "step": 5330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0549391553328566e-05,
+      "loss": 0.3538,
+      "step": 5331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.054760200429492e-05,
+      "loss": 0.3431,
+      "step": 5332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0545812455261276e-05,
+      "loss": 0.3099,
+      "step": 5333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.054402290622763e-05,
+      "loss": 0.2142,
+      "step": 5334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.054223335719399e-05,
+      "loss": 0.3612,
+      "step": 5335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.054044380816034e-05,
+      "loss": 0.3319,
+      "step": 5336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0538654259126704e-05,
+      "loss": 0.4398,
+      "step": 5337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.053686471009306e-05,
+      "loss": 0.1706,
+      "step": 5338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0535075161059414e-05,
+      "loss": 0.3459,
+      "step": 5339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0533285612025776e-05,
+      "loss": 0.3213,
+      "step": 5340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.053149606299213e-05,
+      "loss": 0.4068,
+      "step": 5341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.052970651395848e-05,
+      "loss": 0.2581,
+      "step": 5342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.052791696492484e-05,
+      "loss": 0.3967,
+      "step": 5343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.05261274158912e-05,
+      "loss": 0.422,
+      "step": 5344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.052433786685755e-05,
+      "loss": 0.3736,
+      "step": 5345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.052254831782391e-05,
+      "loss": 0.2775,
+      "step": 5346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.052075876879027e-05,
+      "loss": 0.284,
+      "step": 5347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0518969219756625e-05,
+      "loss": 0.2624,
+      "step": 5348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.051717967072298e-05,
+      "loss": 0.2815,
+      "step": 5349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0515390121689335e-05,
+      "loss": 0.2983,
+      "step": 5350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.051360057265569e-05,
+      "loss": 0.1953,
+      "step": 5351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0511811023622045e-05,
+      "loss": 0.4163,
+      "step": 5352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.051002147458841e-05,
+      "loss": 0.4555,
+      "step": 5353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050823192555476e-05,
+      "loss": 0.245,
+      "step": 5354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050644237652112e-05,
+      "loss": 0.313,
+      "step": 5355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050465282748748e-05,
+      "loss": 0.3037,
+      "step": 5356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0502863278453835e-05,
+      "loss": 0.218,
+      "step": 5357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050107372942018e-05,
+      "loss": 0.3464,
+      "step": 5358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0499284180386545e-05,
+      "loss": 0.2242,
+      "step": 5359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.04974946313529e-05,
+      "loss": 0.29,
+      "step": 5360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0495705082319256e-05,
+      "loss": 0.2502,
+      "step": 5361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.049391553328562e-05,
+      "loss": 0.2736,
+      "step": 5362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.049212598425197e-05,
+      "loss": 0.2286,
+      "step": 5363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.049033643521833e-05,
+      "loss": 0.2705,
+      "step": 5364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048854688618468e-05,
+      "loss": 0.2674,
+      "step": 5365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048675733715104e-05,
+      "loss": 0.2726,
+      "step": 5366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0484967788117394e-05,
+      "loss": 0.1946,
+      "step": 5367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048317823908375e-05,
+      "loss": 0.2286,
+      "step": 5368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048138869005011e-05,
+      "loss": 0.488,
+      "step": 5369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0479599141016466e-05,
+      "loss": 0.2536,
+      "step": 5370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.047780959198282e-05,
+      "loss": 0.2714,
+      "step": 5371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.047602004294918e-05,
+      "loss": 0.2976,
+      "step": 5372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.047423049391553e-05,
+      "loss": 0.2107,
+      "step": 5373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.047244094488189e-05,
+      "loss": 0.2401,
+      "step": 5374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.047065139584825e-05,
+      "loss": 0.2367,
+      "step": 5375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0468861846814604e-05,
+      "loss": 0.1331,
+      "step": 5376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.046707229778096e-05,
+      "loss": 0.2673,
+      "step": 5377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.046528274874732e-05,
+      "loss": 0.2269,
+      "step": 5378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0463493199713676e-05,
+      "loss": 0.1597,
+      "step": 5379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.046170365068003e-05,
+      "loss": 0.2957,
+      "step": 5380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045991410164639e-05,
+      "loss": 0.2986,
+      "step": 5381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045812455261274e-05,
+      "loss": 0.3342,
+      "step": 5382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.04563350035791e-05,
+      "loss": 0.2569,
+      "step": 5383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045454545454546e-05,
+      "loss": 0.2954,
+      "step": 5384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0452755905511814e-05,
+      "loss": 0.4685,
+      "step": 5385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045096635647817e-05,
+      "loss": 0.3199,
+      "step": 5386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0449176807444525e-05,
+      "loss": 0.1925,
+      "step": 5387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0447387258410887e-05,
+      "loss": 0.2471,
+      "step": 5388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0445597709377235e-05,
+      "loss": 0.2607,
+      "step": 5389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.04438081603436e-05,
+      "loss": 0.4852,
+      "step": 5390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.044201861130995e-05,
+      "loss": 0.1519,
+      "step": 5391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.044022906227631e-05,
+      "loss": 0.3133,
+      "step": 5392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.043843951324266e-05,
+      "loss": 0.3334,
+      "step": 5393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0436649964209024e-05,
+      "loss": 0.1604,
+      "step": 5394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.043486041517538e-05,
+      "loss": 0.1477,
+      "step": 5395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0433070866141735e-05,
+      "loss": 0.465,
+      "step": 5396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.043128131710809e-05,
+      "loss": 0.2724,
+      "step": 5397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0429491768074445e-05,
+      "loss": 0.3968,
+      "step": 5398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.04277022190408e-05,
+      "loss": 0.2989,
+      "step": 5399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.042591267000716e-05,
+      "loss": 0.3339,
+      "step": 5400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.042412312097352e-05,
+      "loss": 0.3869,
+      "step": 5401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.042233357193987e-05,
+      "loss": 0.1849,
+      "step": 5402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0420544022906235e-05,
+      "loss": 0.2044,
+      "step": 5403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.041875447387259e-05,
+      "loss": 0.2404,
+      "step": 5404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.041696492483894e-05,
+      "loss": 0.175,
+      "step": 5405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.04151753758053e-05,
+      "loss": 0.3387,
+      "step": 5406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0413385826771656e-05,
+      "loss": 0.2962,
+      "step": 5407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.041159627773801e-05,
+      "loss": 0.228,
+      "step": 5408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0409806728704366e-05,
+      "loss": 0.3116,
+      "step": 5409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040801717967073e-05,
+      "loss": 0.367,
+      "step": 5410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040622763063708e-05,
+      "loss": 0.2741,
+      "step": 5411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040443808160344e-05,
+      "loss": 0.3456,
+      "step": 5412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0402648532569793e-05,
+      "loss": 0.4405,
+      "step": 5413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040085898353615e-05,
+      "loss": 0.3254,
+      "step": 5414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0399069434502504e-05,
+      "loss": 0.325,
+      "step": 5415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0397279885468866e-05,
+      "loss": 0.2047,
+      "step": 5416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.039549033643522e-05,
+      "loss": 0.4009,
+      "step": 5417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0393700787401576e-05,
+      "loss": 0.2549,
+      "step": 5418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.039191123836794e-05,
+      "loss": 0.3617,
+      "step": 5419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.039012168933429e-05,
+      "loss": 0.3382,
+      "step": 5420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.038833214030064e-05,
+      "loss": 0.2482,
+      "step": 5421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0386542591267004e-05,
+      "loss": 0.3197,
+      "step": 5422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.038475304223336e-05,
+      "loss": 0.2618,
+      "step": 5423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0382963493199714e-05,
+      "loss": 0.2568,
+      "step": 5424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0381173944166076e-05,
+      "loss": 0.1905,
+      "step": 5425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.037938439513243e-05,
+      "loss": 0.288,
+      "step": 5426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0377594846098787e-05,
+      "loss": 0.1426,
+      "step": 5427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.037580529706514e-05,
+      "loss": 0.3246,
+      "step": 5428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.03740157480315e-05,
+      "loss": 0.314,
+      "step": 5429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.037222619899785e-05,
+      "loss": 0.2268,
+      "step": 5430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.037043664996421e-05,
+      "loss": 0.3104,
+      "step": 5431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.036864710093057e-05,
+      "loss": 0.3148,
+      "step": 5432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0366857551896924e-05,
+      "loss": 0.2873,
+      "step": 5433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.036506800286328e-05,
+      "loss": 0.3806,
+      "step": 5434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.036327845382964e-05,
+      "loss": 0.3807,
+      "step": 5435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.036148890479599e-05,
+      "loss": 0.2204,
+      "step": 5436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0359699355762345e-05,
+      "loss": 0.2859,
+      "step": 5437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.035790980672871e-05,
+      "loss": 0.1675,
+      "step": 5438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.035612025769506e-05,
+      "loss": 0.3282,
+      "step": 5439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.035433070866142e-05,
+      "loss": 0.2789,
+      "step": 5440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.035254115962778e-05,
+      "loss": 0.3335,
+      "step": 5441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0350751610594135e-05,
+      "loss": 0.1741,
+      "step": 5442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.034896206156049e-05,
+      "loss": 0.2744,
+      "step": 5443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0347172512526845e-05,
+      "loss": 0.3958,
+      "step": 5444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.03453829634932e-05,
+      "loss": 0.2003,
+      "step": 5445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0343593414459556e-05,
+      "loss": 0.4,
+      "step": 5446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.034180386542592e-05,
+      "loss": 0.2082,
+      "step": 5447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.034001431639227e-05,
+      "loss": 0.3055,
+      "step": 5448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.033822476735863e-05,
+      "loss": 0.3287,
+      "step": 5449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.033643521832498e-05,
+      "loss": 0.4578,
+      "step": 5450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0334645669291345e-05,
+      "loss": 0.3602,
+      "step": 5451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0332856120257693e-05,
+      "loss": 0.2949,
+      "step": 5452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0331066571224055e-05,
+      "loss": 0.1683,
+      "step": 5453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032927702219041e-05,
+      "loss": 0.2604,
+      "step": 5454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0327487473156766e-05,
+      "loss": 0.2328,
+      "step": 5455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032569792412312e-05,
+      "loss": 0.307,
+      "step": 5456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032390837508948e-05,
+      "loss": 0.2266,
+      "step": 5457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.032211882605584e-05,
+      "loss": 0.2596,
+      "step": 5458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0320329277022187e-05,
+      "loss": 0.295,
+      "step": 5459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.031853972798855e-05,
+      "loss": 0.178,
+      "step": 5460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0316750178954904e-05,
+      "loss": 0.2674,
+      "step": 5461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.031496062992126e-05,
+      "loss": 0.3201,
+      "step": 5462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.031317108088762e-05,
+      "loss": 0.2569,
+      "step": 5463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0311381531853976e-05,
+      "loss": 0.2895,
+      "step": 5464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030959198282033e-05,
+      "loss": 0.183,
+      "step": 5465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030780243378669e-05,
+      "loss": 0.12,
+      "step": 5466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030601288475304e-05,
+      "loss": 0.204,
+      "step": 5467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.03042233357194e-05,
+      "loss": 0.4178,
+      "step": 5468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030243378668576e-05,
+      "loss": 0.3537,
+      "step": 5469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0300644237652114e-05,
+      "loss": 0.1704,
+      "step": 5470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.029885468861847e-05,
+      "loss": 0.2742,
+      "step": 5471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0297065139584824e-05,
+      "loss": 0.2764,
+      "step": 5472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0295275590551186e-05,
+      "loss": 0.2604,
+      "step": 5473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.029348604151754e-05,
+      "loss": 0.3731,
+      "step": 5474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.02916964924839e-05,
+      "loss": 0.4221,
+      "step": 5475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.028990694345025e-05,
+      "loss": 0.2383,
+      "step": 5476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.028811739441661e-05,
+      "loss": 0.2964,
+      "step": 5477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.028632784538296e-05,
+      "loss": 0.3074,
+      "step": 5478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0284538296349324e-05,
+      "loss": 0.2297,
+      "step": 5479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.028274874731568e-05,
+      "loss": 0.2552,
+      "step": 5480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0280959198282035e-05,
+      "loss": 0.1706,
+      "step": 5481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.02791696492484e-05,
+      "loss": 0.3026,
+      "step": 5482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0277380100214745e-05,
+      "loss": 0.2766,
+      "step": 5483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.02755905511811e-05,
+      "loss": 0.3745,
+      "step": 5484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.027380100214746e-05,
+      "loss": 0.3962,
+      "step": 5485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.027201145311382e-05,
+      "loss": 0.1979,
+      "step": 5486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.027022190408017e-05,
+      "loss": 0.3941,
+      "step": 5487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0268432355046535e-05,
+      "loss": 0.3098,
+      "step": 5488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.026664280601289e-05,
+      "loss": 0.2264,
+      "step": 5489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0264853256979245e-05,
+      "loss": 0.1612,
+      "step": 5490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.02630637079456e-05,
+      "loss": 0.4282,
+      "step": 5491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0261274158911955e-05,
+      "loss": 0.365,
+      "step": 5492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025948460987831e-05,
+      "loss": 0.1765,
+      "step": 5493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0257695060844666e-05,
+      "loss": 0.2374,
+      "step": 5494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025590551181103e-05,
+      "loss": 0.2937,
+      "step": 5495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025411596277738e-05,
+      "loss": 0.2397,
+      "step": 5496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025232641374374e-05,
+      "loss": 0.2808,
+      "step": 5497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.025053686471009e-05,
+      "loss": 0.1876,
+      "step": 5498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.024874731567645e-05,
+      "loss": 0.4353,
+      "step": 5499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0246957766642804e-05,
+      "loss": 0.4768,
+      "step": 5500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0245168217609166e-05,
+      "loss": 0.2158,
+      "step": 5501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.024337866857552e-05,
+      "loss": 0.3713,
+      "step": 5502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0241589119541876e-05,
+      "loss": 0.3647,
+      "step": 5503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.023979957050824e-05,
+      "loss": 0.2893,
+      "step": 5504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.023801002147459e-05,
+      "loss": 0.371,
+      "step": 5505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.023622047244094e-05,
+      "loss": 0.2976,
+      "step": 5506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0234430923407304e-05,
+      "loss": 0.3239,
+      "step": 5507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.023264137437366e-05,
+      "loss": 0.2341,
+      "step": 5508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0230851825340014e-05,
+      "loss": 0.3367,
+      "step": 5509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0229062276306376e-05,
+      "loss": 0.331,
+      "step": 5510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.022727272727273e-05,
+      "loss": 0.2448,
+      "step": 5511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0225483178239086e-05,
+      "loss": 0.2257,
+      "step": 5512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.022369362920544e-05,
+      "loss": 0.3054,
+      "step": 5513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.02219040801718e-05,
+      "loss": 0.3539,
+      "step": 5514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.022011453113815e-05,
+      "loss": 0.3676,
+      "step": 5515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0218324982104514e-05,
+      "loss": 0.2688,
+      "step": 5516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021653543307087e-05,
+      "loss": 0.3222,
+      "step": 5517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0214745884037224e-05,
+      "loss": 0.1887,
+      "step": 5518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021295633500358e-05,
+      "loss": 0.3085,
+      "step": 5519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021116678596994e-05,
+      "loss": 0.3183,
+      "step": 5520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.02093772369363e-05,
+      "loss": 0.3572,
+      "step": 5521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0207587687902645e-05,
+      "loss": 0.5373,
+      "step": 5522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.020579813886901e-05,
+      "loss": 0.2177,
+      "step": 5523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.020400858983536e-05,
+      "loss": 0.2774,
+      "step": 5524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.020221904080172e-05,
+      "loss": 0.4094,
+      "step": 5525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.020042949176808e-05,
+      "loss": 0.4048,
+      "step": 5526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0198639942734435e-05,
+      "loss": 0.4091,
+      "step": 5527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.019685039370079e-05,
+      "loss": 0.4866,
+      "step": 5528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.019506084466715e-05,
+      "loss": 0.3285,
+      "step": 5529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.01932712956335e-05,
+      "loss": 0.4719,
+      "step": 5530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0191481746599855e-05,
+      "loss": 0.2353,
+      "step": 5531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.018969219756622e-05,
+      "loss": 0.3568,
+      "step": 5532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.018790264853257e-05,
+      "loss": 0.2949,
+      "step": 5533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.018611309949893e-05,
+      "loss": 0.2281,
+      "step": 5534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.018432355046528e-05,
+      "loss": 0.282,
+      "step": 5535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0182534001431645e-05,
+      "loss": 0.2301,
+      "step": 5536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0180744452398e-05,
+      "loss": 0.274,
+      "step": 5537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0178954903364355e-05,
+      "loss": 0.4794,
+      "step": 5538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.017716535433071e-05,
+      "loss": 0.2698,
+      "step": 5539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0175375805297066e-05,
+      "loss": 0.216,
+      "step": 5540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.017358625626342e-05,
+      "loss": 0.3464,
+      "step": 5541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.017179670722978e-05,
+      "loss": 0.3503,
+      "step": 5542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.017000715819614e-05,
+      "loss": 0.2145,
+      "step": 5543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.016821760916249e-05,
+      "loss": 0.4242,
+      "step": 5544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.016642806012885e-05,
+      "loss": 0.2968,
+      "step": 5545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0164638511095204e-05,
+      "loss": 0.4059,
+      "step": 5546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.016284896206156e-05,
+      "loss": 0.2751,
+      "step": 5547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.016105941302792e-05,
+      "loss": 0.2502,
+      "step": 5548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0159269863994276e-05,
+      "loss": 0.3982,
+      "step": 5549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.015748031496063e-05,
+      "loss": 0.3295,
+      "step": 5550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.015569076592699e-05,
+      "loss": 0.1727,
+      "step": 5551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.015390121689335e-05,
+      "loss": 0.425,
+      "step": 5552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.01521116678597e-05,
+      "loss": 0.3916,
+      "step": 5553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.015032211882606e-05,
+      "loss": 0.179,
+      "step": 5554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0148532569792414e-05,
+      "loss": 0.2188,
+      "step": 5555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.014674302075877e-05,
+      "loss": 0.2066,
+      "step": 5556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0144953471725124e-05,
+      "loss": 0.3507,
+      "step": 5557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0143163922691486e-05,
+      "loss": 0.1461,
+      "step": 5558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.014137437365784e-05,
+      "loss": 0.4035,
+      "step": 5559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0139584824624197e-05,
+      "loss": 0.207,
+      "step": 5560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.013779527559055e-05,
+      "loss": 0.1839,
+      "step": 5561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.013600572655691e-05,
+      "loss": 0.2869,
+      "step": 5562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.013421617752326e-05,
+      "loss": 0.2565,
+      "step": 5563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0132426628489624e-05,
+      "loss": 0.2968,
+      "step": 5564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.013063707945598e-05,
+      "loss": 0.2261,
+      "step": 5565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0128847530422335e-05,
+      "loss": 0.1145,
+      "step": 5566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0127057981388696e-05,
+      "loss": 0.2065,
+      "step": 5567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.012526843235505e-05,
+      "loss": 0.2188,
+      "step": 5568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.01234788833214e-05,
+      "loss": 0.375,
+      "step": 5569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.012168933428776e-05,
+      "loss": 0.3239,
+      "step": 5570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.011989978525412e-05,
+      "loss": 0.3606,
+      "step": 5571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.011811023622047e-05,
+      "loss": 0.4058,
+      "step": 5572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0116320687186834e-05,
+      "loss": 0.2976,
+      "step": 5573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.011453113815319e-05,
+      "loss": 0.3794,
+      "step": 5574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0112741589119545e-05,
+      "loss": 0.4139,
+      "step": 5575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.01109520400859e-05,
+      "loss": 0.4585,
+      "step": 5576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0109162491052255e-05,
+      "loss": 0.4199,
+      "step": 5577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.010737294201861e-05,
+      "loss": 0.2179,
+      "step": 5578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0105583392984966e-05,
+      "loss": 0.2367,
+      "step": 5579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.010379384395133e-05,
+      "loss": 0.3876,
+      "step": 5580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.010200429491768e-05,
+      "loss": 0.2041,
+      "step": 5581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.010021474588404e-05,
+      "loss": 0.284,
+      "step": 5582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.00984251968504e-05,
+      "loss": 0.2795,
+      "step": 5583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0096635647816755e-05,
+      "loss": 0.2672,
+      "step": 5584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0094846098783104e-05,
+      "loss": 0.282,
+      "step": 5585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0093056549749465e-05,
+      "loss": 0.2755,
+      "step": 5586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.009126700071582e-05,
+      "loss": 0.4203,
+      "step": 5587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0089477451682176e-05,
+      "loss": 0.2313,
+      "step": 5588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.008768790264854e-05,
+      "loss": 0.3663,
+      "step": 5589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.008589835361489e-05,
+      "loss": 0.3481,
+      "step": 5590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.008410880458125e-05,
+      "loss": 0.2808,
+      "step": 5591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0082319255547603e-05,
+      "loss": 0.2955,
+      "step": 5592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.008052970651396e-05,
+      "loss": 0.1791,
+      "step": 5593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0078740157480314e-05,
+      "loss": 0.2536,
+      "step": 5594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0076950608446676e-05,
+      "loss": 0.2399,
+      "step": 5595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.007516105941303e-05,
+      "loss": 0.2285,
+      "step": 5596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0073371510379386e-05,
+      "loss": 0.255,
+      "step": 5597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.007158196134574e-05,
+      "loss": 0.2997,
+      "step": 5598
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8969709826163654,
+      "eval_loss": 0.24980729818344116,
+      "eval_runtime": 51.5094,
+      "eval_samples_per_second": 1738.848,
+      "eval_steps_per_second": 27.18,
+      "step": 5598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.00697924123121e-05,
+      "loss": 0.4721,
+      "step": 5599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.006800286327845e-05,
+      "loss": 0.3712,
+      "step": 5600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0066213314244814e-05,
+      "loss": 0.2219,
+      "step": 5601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.006442376521117e-05,
+      "loss": 0.2459,
+      "step": 5602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0062634216177524e-05,
+      "loss": 0.1875,
+      "step": 5603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.006084466714388e-05,
+      "loss": 0.4133,
+      "step": 5604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.005905511811024e-05,
+      "loss": 0.1242,
+      "step": 5605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0057265569076596e-05,
+      "loss": 0.2314,
+      "step": 5606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.005547602004295e-05,
+      "loss": 0.3582,
+      "step": 5607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.005368647100931e-05,
+      "loss": 0.204,
+      "step": 5608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.005189692197566e-05,
+      "loss": 0.194,
+      "step": 5609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.005010737294202e-05,
+      "loss": 0.1337,
+      "step": 5610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.004831782390838e-05,
+      "loss": 0.352,
+      "step": 5611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0046528274874734e-05,
+      "loss": 0.1317,
+      "step": 5612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.004473872584109e-05,
+      "loss": 0.3115,
+      "step": 5613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.004294917680745e-05,
+      "loss": 0.2537,
+      "step": 5614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.004115962777381e-05,
+      "loss": 0.282,
+      "step": 5615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0039370078740155e-05,
+      "loss": 0.4125,
+      "step": 5616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.003758052970652e-05,
+      "loss": 0.2344,
+      "step": 5617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.003579098067287e-05,
+      "loss": 0.0889,
+      "step": 5618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.003400143163923e-05,
+      "loss": 0.312,
+      "step": 5619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.003221188260558e-05,
+      "loss": 0.2753,
+      "step": 5620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0030422333571945e-05,
+      "loss": 0.1346,
+      "step": 5621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.00286327845383e-05,
+      "loss": 0.1238,
+      "step": 5622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0026843235504655e-05,
+      "loss": 0.1924,
+      "step": 5623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.002505368647101e-05,
+      "loss": 0.2101,
+      "step": 5624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0023264137437365e-05,
+      "loss": 0.2173,
+      "step": 5625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.002147458840372e-05,
+      "loss": 0.213,
+      "step": 5626
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.001968503937008e-05,
+      "loss": 0.139,
+      "step": 5627
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.001789549033644e-05,
+      "loss": 0.1163,
+      "step": 5628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.001610594130279e-05,
+      "loss": 0.1259,
+      "step": 5629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0014316392269155e-05,
+      "loss": 0.1025,
+      "step": 5630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.00125268432355e-05,
+      "loss": 0.1093,
+      "step": 5631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.001073729420186e-05,
+      "loss": 0.1146,
+      "step": 5632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.000894774516822e-05,
+      "loss": 0.1892,
+      "step": 5633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0007158196134576e-05,
+      "loss": 0.3331,
+      "step": 5634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.000536864710093e-05,
+      "loss": 0.2247,
+      "step": 5635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.000357909806729e-05,
+      "loss": 0.1521,
+      "step": 5636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.000178954903365e-05,
+      "loss": 0.1971,
+      "step": 5637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4e-05,
+      "loss": 0.1797,
+      "step": 5638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.999821045096636e-05,
+      "loss": 0.2077,
+      "step": 5639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9996420901932714e-05,
+      "loss": 0.2687,
+      "step": 5640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.999463135289907e-05,
+      "loss": 0.2382,
+      "step": 5641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9992841803865424e-05,
+      "loss": 0.2317,
+      "step": 5642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9991052254831786e-05,
+      "loss": 0.2412,
+      "step": 5643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.998926270579814e-05,
+      "loss": 0.2416,
+      "step": 5644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9987473156764496e-05,
+      "loss": 0.3779,
+      "step": 5645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.998568360773086e-05,
+      "loss": 0.2808,
+      "step": 5646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.998389405869721e-05,
+      "loss": 0.228,
+      "step": 5647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.998210450966356e-05,
+      "loss": 0.2151,
+      "step": 5648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9980314960629924e-05,
+      "loss": 0.2675,
+      "step": 5649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.997852541159628e-05,
+      "loss": 0.1312,
+      "step": 5650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9976735862562634e-05,
+      "loss": 0.1422,
+      "step": 5651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9974946313528996e-05,
+      "loss": 0.2829,
+      "step": 5652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.997315676449535e-05,
+      "loss": 0.2561,
+      "step": 5653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.997136721546171e-05,
+      "loss": 0.2035,
+      "step": 5654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.996957766642806e-05,
+      "loss": 0.1832,
+      "step": 5655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.996778811739442e-05,
+      "loss": 0.2096,
+      "step": 5656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.996599856836077e-05,
+      "loss": 0.2418,
+      "step": 5657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9964209019327134e-05,
+      "loss": 0.1627,
+      "step": 5658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.996241947029349e-05,
+      "loss": 0.2483,
+      "step": 5659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9960629921259845e-05,
+      "loss": 0.1686,
+      "step": 5660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.99588403722262e-05,
+      "loss": 0.2389,
+      "step": 5661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.995705082319256e-05,
+      "loss": 0.1571,
+      "step": 5662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.995526127415891e-05,
+      "loss": 0.1602,
+      "step": 5663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.995347172512527e-05,
+      "loss": 0.2392,
+      "step": 5664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.995168217609163e-05,
+      "loss": 0.1043,
+      "step": 5665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.994989262705798e-05,
+      "loss": 0.296,
+      "step": 5666
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.994810307802434e-05,
+      "loss": 0.274,
+      "step": 5667
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.99463135289907e-05,
+      "loss": 0.1926,
+      "step": 5668
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9944523979957055e-05,
+      "loss": 0.2268,
+      "step": 5669
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.994273443092341e-05,
+      "loss": 0.1824,
+      "step": 5670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9940944881889765e-05,
+      "loss": 0.2067,
+      "step": 5671
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.993915533285612e-05,
+      "loss": 0.2198,
+      "step": 5672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9937365783822476e-05,
+      "loss": 0.1618,
+      "step": 5673
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.993557623478884e-05,
+      "loss": 0.1441,
+      "step": 5674
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.993378668575519e-05,
+      "loss": 0.2712,
+      "step": 5675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.993199713672155e-05,
+      "loss": 0.1284,
+      "step": 5676
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.993020758768791e-05,
+      "loss": 0.2947,
+      "step": 5677
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.992841803865426e-05,
+      "loss": 0.1808,
+      "step": 5678
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9926628489620614e-05,
+      "loss": 0.2025,
+      "step": 5679
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9924838940586976e-05,
+      "loss": 0.168,
+      "step": 5680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.992304939155333e-05,
+      "loss": 0.3303,
+      "step": 5681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9921259842519686e-05,
+      "loss": 0.2194,
+      "step": 5682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.991947029348604e-05,
+      "loss": 0.4059,
+      "step": 5683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.99176807444524e-05,
+      "loss": 0.1055,
+      "step": 5684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.991589119541876e-05,
+      "loss": 0.2944,
+      "step": 5685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9914101646385114e-05,
+      "loss": 0.3341,
+      "step": 5686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.991231209735147e-05,
+      "loss": 0.2019,
+      "step": 5687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9910522548317824e-05,
+      "loss": 0.202,
+      "step": 5688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.990873299928418e-05,
+      "loss": 0.2545,
+      "step": 5689
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.990694345025054e-05,
+      "loss": 0.1235,
+      "step": 5690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9905153901216896e-05,
+      "loss": 0.1953,
+      "step": 5691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.990336435218325e-05,
+      "loss": 0.1443,
+      "step": 5692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9901574803149613e-05,
+      "loss": 0.2406,
+      "step": 5693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989978525411596e-05,
+      "loss": 0.2611,
+      "step": 5694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989799570508232e-05,
+      "loss": 0.247,
+      "step": 5695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989620615604868e-05,
+      "loss": 0.2316,
+      "step": 5696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9894416607015034e-05,
+      "loss": 0.2027,
+      "step": 5697
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989262705798139e-05,
+      "loss": 0.2525,
+      "step": 5698
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989083750894775e-05,
+      "loss": 0.1768,
+      "step": 5699
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9889047959914107e-05,
+      "loss": 0.3438,
+      "step": 5700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.988725841088046e-05,
+      "loss": 0.2595,
+      "step": 5701
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.988546886184682e-05,
+      "loss": 0.2609,
+      "step": 5702
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.988367931281317e-05,
+      "loss": 0.1789,
+      "step": 5703
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.988188976377953e-05,
+      "loss": 0.2293,
+      "step": 5704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.988010021474588e-05,
+      "loss": 0.235,
+      "step": 5705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9878310665712244e-05,
+      "loss": 0.263,
+      "step": 5706
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.98765211166786e-05,
+      "loss": 0.2595,
+      "step": 5707
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9874731567644955e-05,
+      "loss": 0.1682,
+      "step": 5708
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.987294201861132e-05,
+      "loss": 0.1603,
+      "step": 5709
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9871152469577665e-05,
+      "loss": 0.2625,
+      "step": 5710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986936292054402e-05,
+      "loss": 0.0797,
+      "step": 5711
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986757337151038e-05,
+      "loss": 0.2982,
+      "step": 5712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986578382247674e-05,
+      "loss": 0.238,
+      "step": 5713
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986399427344309e-05,
+      "loss": 0.109,
+      "step": 5714
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9862204724409455e-05,
+      "loss": 0.4002,
+      "step": 5715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.986041517537581e-05,
+      "loss": 0.1389,
+      "step": 5716
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9858625626342165e-05,
+      "loss": 0.1997,
+      "step": 5717
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.985683607730852e-05,
+      "loss": 0.1661,
+      "step": 5718
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9855046528274876e-05,
+      "loss": 0.1888,
+      "step": 5719
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.985325697924123e-05,
+      "loss": 0.2379,
+      "step": 5720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.985146743020759e-05,
+      "loss": 0.1935,
+      "step": 5721
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.984967788117395e-05,
+      "loss": 0.1672,
+      "step": 5722
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.98478883321403e-05,
+      "loss": 0.2456,
+      "step": 5723
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.984609878310666e-05,
+      "loss": 0.2656,
+      "step": 5724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9844309234073013e-05,
+      "loss": 0.1842,
+      "step": 5725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.984251968503937e-05,
+      "loss": 0.1998,
+      "step": 5726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9840730136005724e-05,
+      "loss": 0.1307,
+      "step": 5727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9838940586972086e-05,
+      "loss": 0.2399,
+      "step": 5728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.983715103793844e-05,
+      "loss": 0.2861,
+      "step": 5729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9835361488904796e-05,
+      "loss": 0.3434,
+      "step": 5730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.983357193987116e-05,
+      "loss": 0.2224,
+      "step": 5731
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.983178239083751e-05,
+      "loss": 0.2137,
+      "step": 5732
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.982999284180386e-05,
+      "loss": 0.2435,
+      "step": 5733
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9828203292770224e-05,
+      "loss": 0.3059,
+      "step": 5734
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.982641374373658e-05,
+      "loss": 0.3067,
+      "step": 5735
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9824624194702934e-05,
+      "loss": 0.223,
+      "step": 5736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9822834645669296e-05,
+      "loss": 0.3078,
+      "step": 5737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.982104509663565e-05,
+      "loss": 0.2394,
+      "step": 5738
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9819255547602007e-05,
+      "loss": 0.1864,
+      "step": 5739
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981746599856837e-05,
+      "loss": 0.1444,
+      "step": 5740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981567644953472e-05,
+      "loss": 0.2072,
+      "step": 5741
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981388690050107e-05,
+      "loss": 0.3267,
+      "step": 5742
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9812097351467434e-05,
+      "loss": 0.2285,
+      "step": 5743
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981030780243379e-05,
+      "loss": 0.2534,
+      "step": 5744
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9808518253400144e-05,
+      "loss": 0.1999,
+      "step": 5745
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.98067287043665e-05,
+      "loss": 0.4296,
+      "step": 5746
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.980493915533286e-05,
+      "loss": 0.1383,
+      "step": 5747
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.980314960629922e-05,
+      "loss": 0.1892,
+      "step": 5748
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.980136005726557e-05,
+      "loss": 0.1888,
+      "step": 5749
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979957050823193e-05,
+      "loss": 0.245,
+      "step": 5750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979778095919828e-05,
+      "loss": 0.2255,
+      "step": 5751
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979599141016464e-05,
+      "loss": 0.2093,
+      "step": 5752
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9794201861131e-05,
+      "loss": 0.1797,
+      "step": 5753
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9792412312097355e-05,
+      "loss": 0.2366,
+      "step": 5754
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979062276306371e-05,
+      "loss": 0.3278,
+      "step": 5755
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.978883321403007e-05,
+      "loss": 0.2753,
+      "step": 5756
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.978704366499642e-05,
+      "loss": 0.3899,
+      "step": 5757
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9785254115962776e-05,
+      "loss": 0.1961,
+      "step": 5758
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.978346456692914e-05,
+      "loss": 0.1981,
+      "step": 5759
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.978167501789549e-05,
+      "loss": 0.1826,
+      "step": 5760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.977988546886185e-05,
+      "loss": 0.2372,
+      "step": 5761
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.977809591982821e-05,
+      "loss": 0.2717,
+      "step": 5762
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9776306370794565e-05,
+      "loss": 0.2775,
+      "step": 5763
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9774516821760913e-05,
+      "loss": 0.1731,
+      "step": 5764
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.3321,
+      "step": 5765
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.977093772369363e-05,
+      "loss": 0.2066,
+      "step": 5766
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9769148174659986e-05,
+      "loss": 0.1856,
+      "step": 5767
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.976735862562634e-05,
+      "loss": 0.3165,
+      "step": 5768
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.97655690765927e-05,
+      "loss": 0.222,
+      "step": 5769
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.976377952755906e-05,
+      "loss": 0.1553,
+      "step": 5770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.976198997852541e-05,
+      "loss": 0.3315,
+      "step": 5771
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.976020042949177e-05,
+      "loss": 0.2161,
+      "step": 5772
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9758410880458124e-05,
+      "loss": 0.2148,
+      "step": 5773
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.975662133142448e-05,
+      "loss": 0.2482,
+      "step": 5774
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.975483178239084e-05,
+      "loss": 0.1654,
+      "step": 5775
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9753042233357196e-05,
+      "loss": 0.1275,
+      "step": 5776
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.975125268432355e-05,
+      "loss": 0.225,
+      "step": 5777
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974946313528991e-05,
+      "loss": 0.1646,
+      "step": 5778
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974767358625627e-05,
+      "loss": 0.188,
+      "step": 5779
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974588403722262e-05,
+      "loss": 0.165,
+      "step": 5780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974409448818898e-05,
+      "loss": 0.1973,
+      "step": 5781
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9742304939155334e-05,
+      "loss": 0.2886,
+      "step": 5782
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.974051539012169e-05,
+      "loss": 0.1309,
+      "step": 5783
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.973872584108805e-05,
+      "loss": 0.1663,
+      "step": 5784
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9736936292054406e-05,
+      "loss": 0.2879,
+      "step": 5785
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.973514674302076e-05,
+      "loss": 0.2514,
+      "step": 5786
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.973335719398712e-05,
+      "loss": 0.241,
+      "step": 5787
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.973156764495347e-05,
+      "loss": 0.1969,
+      "step": 5788
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.972977809591983e-05,
+      "loss": 0.1907,
+      "step": 5789
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.972798854688618e-05,
+      "loss": 0.1557,
+      "step": 5790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9726198997852544e-05,
+      "loss": 0.3186,
+      "step": 5791
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.97244094488189e-05,
+      "loss": 0.2512,
+      "step": 5792
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9722619899785255e-05,
+      "loss": 0.2404,
+      "step": 5793
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.972083035075162e-05,
+      "loss": 0.2085,
+      "step": 5794
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971904080171797e-05,
+      "loss": 0.3464,
+      "step": 5795
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971725125268432e-05,
+      "loss": 0.1874,
+      "step": 5796
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971546170365068e-05,
+      "loss": 0.2046,
+      "step": 5797
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971367215461704e-05,
+      "loss": 0.2855,
+      "step": 5798
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971188260558339e-05,
+      "loss": 0.1796,
+      "step": 5799
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9710093056549755e-05,
+      "loss": 0.2098,
+      "step": 5800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970830350751611e-05,
+      "loss": 0.1557,
+      "step": 5801
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9706513958482465e-05,
+      "loss": 0.2238,
+      "step": 5802
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970472440944882e-05,
+      "loss": 0.2578,
+      "step": 5803
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9702934860415175e-05,
+      "loss": 0.2159,
+      "step": 5804
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970114531138153e-05,
+      "loss": 0.3061,
+      "step": 5805
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969935576234789e-05,
+      "loss": 0.2521,
+      "step": 5806
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969756621331425e-05,
+      "loss": 0.2274,
+      "step": 5807
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.96957766642806e-05,
+      "loss": 0.3468,
+      "step": 5808
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969398711524696e-05,
+      "loss": 0.3051,
+      "step": 5809
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969219756621332e-05,
+      "loss": 0.2476,
+      "step": 5810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969040801717967e-05,
+      "loss": 0.2293,
+      "step": 5811
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.968861846814603e-05,
+      "loss": 0.2106,
+      "step": 5812
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9686828919112386e-05,
+      "loss": 0.1556,
+      "step": 5813
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.968503937007874e-05,
+      "loss": 0.3412,
+      "step": 5814
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9683249821045096e-05,
+      "loss": 0.1839,
+      "step": 5815
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.968146027201146e-05,
+      "loss": 0.2131,
+      "step": 5816
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.967967072297781e-05,
+      "loss": 0.158,
+      "step": 5817
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.967788117394417e-05,
+      "loss": 0.1698,
+      "step": 5818
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9676091624910524e-05,
+      "loss": 0.3,
+      "step": 5819
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.967430207587688e-05,
+      "loss": 0.3803,
+      "step": 5820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9672512526843234e-05,
+      "loss": 0.2261,
+      "step": 5821
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9670722977809596e-05,
+      "loss": 0.0834,
+      "step": 5822
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.966893342877595e-05,
+      "loss": 0.2062,
+      "step": 5823
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9667143879742306e-05,
+      "loss": 0.3397,
+      "step": 5824
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.966535433070867e-05,
+      "loss": 0.1884,
+      "step": 5825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9663564781675023e-05,
+      "loss": 0.3242,
+      "step": 5826
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.966177523264137e-05,
+      "loss": 0.3335,
+      "step": 5827
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9659985683607734e-05,
+      "loss": 0.2863,
+      "step": 5828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.965819613457409e-05,
+      "loss": 0.2407,
+      "step": 5829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9656406585540444e-05,
+      "loss": 0.2446,
+      "step": 5830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.96546170365068e-05,
+      "loss": 0.2304,
+      "step": 5831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.965282748747316e-05,
+      "loss": 0.2835,
+      "step": 5832
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.965103793843952e-05,
+      "loss": 0.2308,
+      "step": 5833
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.964924838940587e-05,
+      "loss": 0.2388,
+      "step": 5834
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.964745884037223e-05,
+      "loss": 0.2798,
+      "step": 5835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.964566929133858e-05,
+      "loss": 0.3651,
+      "step": 5836
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.964387974230494e-05,
+      "loss": 0.1284,
+      "step": 5837
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.96420901932713e-05,
+      "loss": 0.1514,
+      "step": 5838
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9640300644237655e-05,
+      "loss": 0.3029,
+      "step": 5839
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.963851109520401e-05,
+      "loss": 0.1242,
+      "step": 5840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.963672154617037e-05,
+      "loss": 0.2044,
+      "step": 5841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.963493199713673e-05,
+      "loss": 0.2105,
+      "step": 5842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9633142448103075e-05,
+      "loss": 0.1925,
+      "step": 5843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.963135289906944e-05,
+      "loss": 0.1715,
+      "step": 5844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.962956335003579e-05,
+      "loss": 0.3755,
+      "step": 5845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.962777380100215e-05,
+      "loss": 0.2084,
+      "step": 5846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.962598425196851e-05,
+      "loss": 0.2686,
+      "step": 5847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9624194702934865e-05,
+      "loss": 0.1745,
+      "step": 5848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.962240515390122e-05,
+      "loss": 0.1819,
+      "step": 5849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9620615604867575e-05,
+      "loss": 0.2211,
+      "step": 5850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.961882605583393e-05,
+      "loss": 0.3719,
+      "step": 5851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9617036506800286e-05,
+      "loss": 0.2776,
+      "step": 5852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.961524695776664e-05,
+      "loss": 0.3277,
+      "step": 5853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9613457408733e-05,
+      "loss": 0.2476,
+      "step": 5854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.961166785969936e-05,
+      "loss": 0.2722,
+      "step": 5855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.960987831066571e-05,
+      "loss": 0.1784,
+      "step": 5856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9608088761632075e-05,
+      "loss": 0.283,
+      "step": 5857
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9606299212598424e-05,
+      "loss": 0.2263,
+      "step": 5858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.960450966356478e-05,
+      "loss": 0.306,
+      "step": 5859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.960272011453114e-05,
+      "loss": 0.2761,
+      "step": 5860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9600930565497496e-05,
+      "loss": 0.3363,
+      "step": 5861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.959914101646385e-05,
+      "loss": 0.2262,
+      "step": 5862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.959735146743021e-05,
+      "loss": 0.1448,
+      "step": 5863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.959556191839657e-05,
+      "loss": 0.1309,
+      "step": 5864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9593772369362923e-05,
+      "loss": 0.2477,
+      "step": 5865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.959198282032928e-05,
+      "loss": 0.1385,
+      "step": 5866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9590193271295634e-05,
+      "loss": 0.2031,
+      "step": 5867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.958840372226199e-05,
+      "loss": 0.2123,
+      "step": 5868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.958661417322835e-05,
+      "loss": 0.093,
+      "step": 5869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9584824624194706e-05,
+      "loss": 0.2787,
+      "step": 5870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.958303507516106e-05,
+      "loss": 0.1165,
+      "step": 5871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9581245526127417e-05,
+      "loss": 0.362,
+      "step": 5872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.957945597709378e-05,
+      "loss": 0.2527,
+      "step": 5873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.957766642806013e-05,
+      "loss": 0.1848,
+      "step": 5874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.957587687902649e-05,
+      "loss": 0.285,
+      "step": 5875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9574087329992844e-05,
+      "loss": 0.303,
+      "step": 5876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.95722977809592e-05,
+      "loss": 0.2245,
+      "step": 5877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9570508231925555e-05,
+      "loss": 0.159,
+      "step": 5878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9568718682891916e-05,
+      "loss": 0.1911,
+      "step": 5879
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.956692913385827e-05,
+      "loss": 0.3149,
+      "step": 5880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.956513958482463e-05,
+      "loss": 0.193,
+      "step": 5881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.956335003579098e-05,
+      "loss": 0.1531,
+      "step": 5882
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.956156048675734e-05,
+      "loss": 0.2697,
+      "step": 5883
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.955977093772369e-05,
+      "loss": 0.1583,
+      "step": 5884
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9557981388690054e-05,
+      "loss": 0.1352,
+      "step": 5885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.955619183965641e-05,
+      "loss": 0.1982,
+      "step": 5886
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9554402290622765e-05,
+      "loss": 0.2006,
+      "step": 5887
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.955261274158913e-05,
+      "loss": 0.1644,
+      "step": 5888
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.955082319255548e-05,
+      "loss": 0.2747,
+      "step": 5889
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.954903364352183e-05,
+      "loss": 0.2143,
+      "step": 5890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.954724409448819e-05,
+      "loss": 0.1309,
+      "step": 5891
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.954545454545455e-05,
+      "loss": 0.1796,
+      "step": 5892
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.95436649964209e-05,
+      "loss": 0.4953,
+      "step": 5893
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.954187544738726e-05,
+      "loss": 0.1871,
+      "step": 5894
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.954008589835362e-05,
+      "loss": 0.1946,
+      "step": 5895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9538296349319975e-05,
+      "loss": 0.2077,
+      "step": 5896
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.953650680028633e-05,
+      "loss": 0.22,
+      "step": 5897
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9534717251252685e-05,
+      "loss": 0.0813,
+      "step": 5898
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.953292770221904e-05,
+      "loss": 0.2319,
+      "step": 5899
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9531138153185396e-05,
+      "loss": 0.2315,
+      "step": 5900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.952934860415176e-05,
+      "loss": 0.2656,
+      "step": 5901
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.952755905511811e-05,
+      "loss": 0.4326,
+      "step": 5902
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.952576950608447e-05,
+      "loss": 0.2433,
+      "step": 5903
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.952397995705083e-05,
+      "loss": 0.2015,
+      "step": 5904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.952219040801718e-05,
+      "loss": 0.2205,
+      "step": 5905
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9520400858983534e-05,
+      "loss": 0.1375,
+      "step": 5906
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9518611309949896e-05,
+      "loss": 0.1657,
+      "step": 5907
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.951682176091625e-05,
+      "loss": 0.3533,
+      "step": 5908
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9515032211882606e-05,
+      "loss": 0.2764,
+      "step": 5909
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.951324266284897e-05,
+      "loss": 0.1769,
+      "step": 5910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.951145311381532e-05,
+      "loss": 0.3417,
+      "step": 5911
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.950966356478168e-05,
+      "loss": 0.202,
+      "step": 5912
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9507874015748034e-05,
+      "loss": 0.2714,
+      "step": 5913
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.950608446671439e-05,
+      "loss": 0.159,
+      "step": 5914
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9504294917680744e-05,
+      "loss": 0.1669,
+      "step": 5915
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.95025053686471e-05,
+      "loss": 0.2112,
+      "step": 5916
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.950071581961346e-05,
+      "loss": 0.2451,
+      "step": 5917
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9498926270579816e-05,
+      "loss": 0.1688,
+      "step": 5918
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.949713672154617e-05,
+      "loss": 0.1675,
+      "step": 5919
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9495347172512534e-05,
+      "loss": 0.2614,
+      "step": 5920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.949355762347888e-05,
+      "loss": 0.2457,
+      "step": 5921
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.949176807444524e-05,
+      "loss": 0.4224,
+      "step": 5922
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.94899785254116e-05,
+      "loss": 0.2674,
+      "step": 5923
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9488188976377954e-05,
+      "loss": 0.3102,
+      "step": 5924
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.948639942734431e-05,
+      "loss": 0.2384,
+      "step": 5925
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.948460987831067e-05,
+      "loss": 0.1681,
+      "step": 5926
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.948282032927703e-05,
+      "loss": 0.2363,
+      "step": 5927
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.948103078024338e-05,
+      "loss": 0.1999,
+      "step": 5928
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.947924123120974e-05,
+      "loss": 0.1716,
+      "step": 5929
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.947745168217609e-05,
+      "loss": 0.291,
+      "step": 5930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.947566213314245e-05,
+      "loss": 0.2272,
+      "step": 5931
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.947387258410881e-05,
+      "loss": 0.1166,
+      "step": 5932
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9472083035075165e-05,
+      "loss": 0.1704,
+      "step": 5933
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.947029348604152e-05,
+      "loss": 0.1289,
+      "step": 5934
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9468503937007875e-05,
+      "loss": 0.1931,
+      "step": 5935
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.946671438797424e-05,
+      "loss": 0.1699,
+      "step": 5936
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9464924838940585e-05,
+      "loss": 0.2158,
+      "step": 5937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.946313528990694e-05,
+      "loss": 0.1745,
+      "step": 5938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.94613457408733e-05,
+      "loss": 0.2673,
+      "step": 5939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.945955619183966e-05,
+      "loss": 0.1725,
+      "step": 5940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.945776664280601e-05,
+      "loss": 0.2476,
+      "step": 5941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9455977093772375e-05,
+      "loss": 0.2271,
+      "step": 5942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.945418754473873e-05,
+      "loss": 0.1249,
+      "step": 5943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.945239799570508e-05,
+      "loss": 0.3722,
+      "step": 5944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.945060844667144e-05,
+      "loss": 0.142,
+      "step": 5945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9448818897637796e-05,
+      "loss": 0.2935,
+      "step": 5946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.944702934860415e-05,
+      "loss": 0.2662,
+      "step": 5947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.944523979957051e-05,
+      "loss": 0.1583,
+      "step": 5948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.944345025053687e-05,
+      "loss": 0.1718,
+      "step": 5949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.944166070150322e-05,
+      "loss": 0.1857,
+      "step": 5950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9439871152469585e-05,
+      "loss": 0.2161,
+      "step": 5951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9438081603435934e-05,
+      "loss": 0.3072,
+      "step": 5952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.943629205440229e-05,
+      "loss": 0.1716,
+      "step": 5953
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.943450250536865e-05,
+      "loss": 0.325,
+      "step": 5954
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9432712956335006e-05,
+      "loss": 0.2229,
+      "step": 5955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.943092340730136e-05,
+      "loss": 0.2032,
+      "step": 5956
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9429133858267716e-05,
+      "loss": 0.1926,
+      "step": 5957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.942734430923408e-05,
+      "loss": 0.2636,
+      "step": 5958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9425554760200434e-05,
+      "loss": 0.2322,
+      "step": 5959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.942376521116679e-05,
+      "loss": 0.1875,
+      "step": 5960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9421975662133144e-05,
+      "loss": 0.1436,
+      "step": 5961
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.94201861130995e-05,
+      "loss": 0.2456,
+      "step": 5962
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9418396564065854e-05,
+      "loss": 0.3044,
+      "step": 5963
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9416607015032216e-05,
+      "loss": 0.2027,
+      "step": 5964
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.941481746599857e-05,
+      "loss": 0.1167,
+      "step": 5965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.941302791696493e-05,
+      "loss": 0.29,
+      "step": 5966
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.941123836793129e-05,
+      "loss": 0.2722,
+      "step": 5967
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.940944881889764e-05,
+      "loss": 0.0883,
+      "step": 5968
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.940765926986399e-05,
+      "loss": 0.1893,
+      "step": 5969
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9405869720830354e-05,
+      "loss": 0.2107,
+      "step": 5970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.940408017179671e-05,
+      "loss": 0.2064,
+      "step": 5971
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9402290622763065e-05,
+      "loss": 0.1648,
+      "step": 5972
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9400501073729427e-05,
+      "loss": 0.0891,
+      "step": 5973
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.939871152469578e-05,
+      "loss": 0.2345,
+      "step": 5974
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.939692197566214e-05,
+      "loss": 0.2455,
+      "step": 5975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.939513242662849e-05,
+      "loss": 0.1799,
+      "step": 5976
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.939334287759485e-05,
+      "loss": 0.2385,
+      "step": 5977
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.93915533285612e-05,
+      "loss": 0.1974,
+      "step": 5978
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938976377952756e-05,
+      "loss": 0.2359,
+      "step": 5979
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938797423049392e-05,
+      "loss": 0.1793,
+      "step": 5980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9386184681460275e-05,
+      "loss": 0.1511,
+      "step": 5981
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938439513242663e-05,
+      "loss": 0.1264,
+      "step": 5982
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9382605583392985e-05,
+      "loss": 0.1623,
+      "step": 5983
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938081603435934e-05,
+      "loss": 0.2396,
+      "step": 5984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9379026485325696e-05,
+      "loss": 0.1813,
+      "step": 5985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937723693629206e-05,
+      "loss": 0.3193,
+      "step": 5986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937544738725841e-05,
+      "loss": 0.2537,
+      "step": 5987
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937365783822477e-05,
+      "loss": 0.1379,
+      "step": 5988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937186828919113e-05,
+      "loss": 0.1552,
+      "step": 5989
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9370078740157485e-05,
+      "loss": 0.2495,
+      "step": 5990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9368289191123834e-05,
+      "loss": 0.2904,
+      "step": 5991
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9366499642090196e-05,
+      "loss": 0.1493,
+      "step": 5992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.936471009305655e-05,
+      "loss": 0.1465,
+      "step": 5993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9362920544022906e-05,
+      "loss": 0.1285,
+      "step": 5994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.936113099498927e-05,
+      "loss": 0.3675,
+      "step": 5995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.935934144595562e-05,
+      "loss": 0.2674,
+      "step": 5996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.935755189692198e-05,
+      "loss": 0.351,
+      "step": 5997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9355762347888334e-05,
+      "loss": 0.2104,
+      "step": 5998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.935397279885469e-05,
+      "loss": 0.1657,
+      "step": 5999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9352183249821044e-05,
+      "loss": 0.22,
+      "step": 6000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.93503937007874e-05,
+      "loss": 0.1965,
+      "step": 6001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.934860415175376e-05,
+      "loss": 0.2689,
+      "step": 6002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9346814602720116e-05,
+      "loss": 0.202,
+      "step": 6003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.934502505368647e-05,
+      "loss": 0.1762,
+      "step": 6004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9343235504652833e-05,
+      "loss": 0.1442,
+      "step": 6005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.934144595561919e-05,
+      "loss": 0.3331,
+      "step": 6006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.933965640658554e-05,
+      "loss": 0.1664,
+      "step": 6007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.93378668575519e-05,
+      "loss": 0.174,
+      "step": 6008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9336077308518254e-05,
+      "loss": 0.3168,
+      "step": 6009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.933428775948461e-05,
+      "loss": 0.1714,
+      "step": 6010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.933249821045097e-05,
+      "loss": 0.2926,
+      "step": 6011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9330708661417327e-05,
+      "loss": 0.3644,
+      "step": 6012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.932891911238368e-05,
+      "loss": 0.2155,
+      "step": 6013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.932712956335004e-05,
+      "loss": 0.2777,
+      "step": 6014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.932534001431639e-05,
+      "loss": 0.233,
+      "step": 6015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.932355046528275e-05,
+      "loss": 0.2165,
+      "step": 6016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.932176091624911e-05,
+      "loss": 0.1956,
+      "step": 6017
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9319971367215464e-05,
+      "loss": 0.1441,
+      "step": 6018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931818181818182e-05,
+      "loss": 0.2487,
+      "step": 6019
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9316392269148175e-05,
+      "loss": 0.2657,
+      "step": 6020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931460272011454e-05,
+      "loss": 0.1751,
+      "step": 6021
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931281317108089e-05,
+      "loss": 0.1776,
+      "step": 6022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931102362204725e-05,
+      "loss": 0.1486,
+      "step": 6023
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.93092340730136e-05,
+      "loss": 0.2293,
+      "step": 6024
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.930744452397996e-05,
+      "loss": 0.3323,
+      "step": 6025
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.930565497494631e-05,
+      "loss": 0.135,
+      "step": 6026
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9303865425912675e-05,
+      "loss": 0.2975,
+      "step": 6027
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.930207587687903e-05,
+      "loss": 0.1797,
+      "step": 6028
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9300286327845385e-05,
+      "loss": 0.3168,
+      "step": 6029
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929849677881174e-05,
+      "loss": 0.1437,
+      "step": 6030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9296707229778096e-05,
+      "loss": 0.186,
+      "step": 6031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929491768074445e-05,
+      "loss": 0.1511,
+      "step": 6032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929312813171081e-05,
+      "loss": 0.3029,
+      "step": 6033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929133858267717e-05,
+      "loss": 0.1539,
+      "step": 6034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.928954903364352e-05,
+      "loss": 0.1619,
+      "step": 6035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9287759484609885e-05,
+      "loss": 0.1465,
+      "step": 6036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.928596993557624e-05,
+      "loss": 0.1887,
+      "step": 6037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.928418038654259e-05,
+      "loss": 0.4515,
+      "step": 6038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.928239083750895e-05,
+      "loss": 0.272,
+      "step": 6039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9280601288475306e-05,
+      "loss": 0.202,
+      "step": 6040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.927881173944166e-05,
+      "loss": 0.2041,
+      "step": 6041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9277022190408016e-05,
+      "loss": 0.1407,
+      "step": 6042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.927523264137438e-05,
+      "loss": 0.3248,
+      "step": 6043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.927344309234073e-05,
+      "loss": 0.3008,
+      "step": 6044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.927165354330709e-05,
+      "loss": 0.3357,
+      "step": 6045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9269863994273444e-05,
+      "loss": 0.2251,
+      "step": 6046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.92680744452398e-05,
+      "loss": 0.2595,
+      "step": 6047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9266284896206154e-05,
+      "loss": 0.1433,
+      "step": 6048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9264495347172516e-05,
+      "loss": 0.1907,
+      "step": 6049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.926270579813887e-05,
+      "loss": 0.1787,
+      "step": 6050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9260916249105226e-05,
+      "loss": 0.2625,
+      "step": 6051
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.925912670007159e-05,
+      "loss": 0.1916,
+      "step": 6052
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9257337151037944e-05,
+      "loss": 0.1883,
+      "step": 6053
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.925554760200429e-05,
+      "loss": 0.2016,
+      "step": 6054
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9253758052970654e-05,
+      "loss": 0.2812,
+      "step": 6055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.925196850393701e-05,
+      "loss": 0.2889,
+      "step": 6056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9250178954903364e-05,
+      "loss": 0.2668,
+      "step": 6057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9248389405869726e-05,
+      "loss": 0.2165,
+      "step": 6058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.924659985683608e-05,
+      "loss": 0.1693,
+      "step": 6059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.924481030780244e-05,
+      "loss": 0.28,
+      "step": 6060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.924302075876879e-05,
+      "loss": 0.1639,
+      "step": 6061
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.924123120973515e-05,
+      "loss": 0.2231,
+      "step": 6062
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.92394416607015e-05,
+      "loss": 0.2573,
+      "step": 6063
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.923765211166786e-05,
+      "loss": 0.3118,
+      "step": 6064
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.923586256263422e-05,
+      "loss": 0.263,
+      "step": 6065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9234073013600575e-05,
+      "loss": 0.2615,
+      "step": 6066
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.923228346456693e-05,
+      "loss": 0.2817,
+      "step": 6067
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.923049391553329e-05,
+      "loss": 0.2575,
+      "step": 6068
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.922870436649964e-05,
+      "loss": 0.2038,
+      "step": 6069
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9226914817465995e-05,
+      "loss": 0.2197,
+      "step": 6070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.922512526843236e-05,
+      "loss": 0.2265,
+      "step": 6071
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.922333571939871e-05,
+      "loss": 0.1432,
+      "step": 6072
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.922154617036507e-05,
+      "loss": 0.2287,
+      "step": 6073
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.921975662133143e-05,
+      "loss": 0.3941,
+      "step": 6074
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9217967072297785e-05,
+      "loss": 0.1573,
+      "step": 6075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.921617752326414e-05,
+      "loss": 0.2427,
+      "step": 6076
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9214387974230495e-05,
+      "loss": 0.271,
+      "step": 6077
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.921259842519685e-05,
+      "loss": 0.1533,
+      "step": 6078
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9210808876163206e-05,
+      "loss": 0.0943,
+      "step": 6079
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920901932712957e-05,
+      "loss": 0.2808,
+      "step": 6080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920722977809592e-05,
+      "loss": 0.2445,
+      "step": 6081
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920544022906228e-05,
+      "loss": 0.2945,
+      "step": 6082
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920365068002863e-05,
+      "loss": 0.2489,
+      "step": 6083
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9201861130994995e-05,
+      "loss": 0.07,
+      "step": 6084
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9200071581961344e-05,
+      "loss": 0.1286,
+      "step": 6085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.91982820329277e-05,
+      "loss": 0.3719,
+      "step": 6086
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.919649248389406e-05,
+      "loss": 0.1473,
+      "step": 6087
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9194702934860416e-05,
+      "loss": 0.2352,
+      "step": 6088
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.919291338582677e-05,
+      "loss": 0.2349,
+      "step": 6089
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.919112383679313e-05,
+      "loss": 0.2624,
+      "step": 6090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.918933428775949e-05,
+      "loss": 0.1874,
+      "step": 6091
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9187544738725844e-05,
+      "loss": 0.2693,
+      "step": 6092
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.91857551896922e-05,
+      "loss": 0.1126,
+      "step": 6093
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9183965640658554e-05,
+      "loss": 0.1311,
+      "step": 6094
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.918217609162491e-05,
+      "loss": 0.3483,
+      "step": 6095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.918038654259127e-05,
+      "loss": 0.3059,
+      "step": 6096
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9178596993557626e-05,
+      "loss": 0.1245,
+      "step": 6097
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.917680744452398e-05,
+      "loss": 0.1698,
+      "step": 6098
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9175017895490344e-05,
+      "loss": 0.2945,
+      "step": 6099
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.91732283464567e-05,
+      "loss": 0.1927,
+      "step": 6100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.917143879742305e-05,
+      "loss": 0.2506,
+      "step": 6101
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.916964924838941e-05,
+      "loss": 0.2429,
+      "step": 6102
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9167859699355764e-05,
+      "loss": 0.2118,
+      "step": 6103
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.916607015032212e-05,
+      "loss": 0.304,
+      "step": 6104
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9164280601288475e-05,
+      "loss": 0.165,
+      "step": 6105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.916249105225484e-05,
+      "loss": 0.1389,
+      "step": 6106
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.916070150322119e-05,
+      "loss": 0.2894,
+      "step": 6107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.915891195418755e-05,
+      "loss": 0.1804,
+      "step": 6108
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.91571224051539e-05,
+      "loss": 0.2705,
+      "step": 6109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.915533285612026e-05,
+      "loss": 0.2039,
+      "step": 6110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.915354330708661e-05,
+      "loss": 0.2613,
+      "step": 6111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9151753758052975e-05,
+      "loss": 0.3561,
+      "step": 6112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.914996420901933e-05,
+      "loss": 0.2609,
+      "step": 6113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9148174659985685e-05,
+      "loss": 0.1868,
+      "step": 6114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.914638511095205e-05,
+      "loss": 0.1712,
+      "step": 6115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9144595561918395e-05,
+      "loss": 0.2126,
+      "step": 6116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.914280601288475e-05,
+      "loss": 0.1238,
+      "step": 6117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.914101646385111e-05,
+      "loss": 0.1499,
+      "step": 6118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.913922691481747e-05,
+      "loss": 0.192,
+      "step": 6119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.913743736578382e-05,
+      "loss": 0.239,
+      "step": 6120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9135647816750185e-05,
+      "loss": 0.1408,
+      "step": 6121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.913385826771654e-05,
+      "loss": 0.3568,
+      "step": 6122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9132068718682895e-05,
+      "loss": 0.3106,
+      "step": 6123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.913027916964925e-05,
+      "loss": 0.2298,
+      "step": 6124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9128489620615606e-05,
+      "loss": 0.2852,
+      "step": 6125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.912670007158196e-05,
+      "loss": 0.2654,
+      "step": 6126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9124910522548316e-05,
+      "loss": 0.209,
+      "step": 6127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.912312097351468e-05,
+      "loss": 0.1924,
+      "step": 6128
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.912133142448103e-05,
+      "loss": 0.1586,
+      "step": 6129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.911954187544739e-05,
+      "loss": 0.2211,
+      "step": 6130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.911775232641375e-05,
+      "loss": 0.1897,
+      "step": 6131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.91159627773801e-05,
+      "loss": 0.3277,
+      "step": 6132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9114173228346454e-05,
+      "loss": 0.1429,
+      "step": 6133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9112383679312816e-05,
+      "loss": 0.142,
+      "step": 6134
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.911059413027917e-05,
+      "loss": 0.2893,
+      "step": 6135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9108804581245526e-05,
+      "loss": 0.2351,
+      "step": 6136
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.910701503221189e-05,
+      "loss": 0.2143,
+      "step": 6137
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9105225483178243e-05,
+      "loss": 0.2001,
+      "step": 6138
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.91034359341446e-05,
+      "loss": 0.1657,
+      "step": 6139
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9101646385110954e-05,
+      "loss": 0.3358,
+      "step": 6140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.909985683607731e-05,
+      "loss": 0.1394,
+      "step": 6141
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9098067287043664e-05,
+      "loss": 0.2222,
+      "step": 6142
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9096277738010026e-05,
+      "loss": 0.2791,
+      "step": 6143
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.909448818897638e-05,
+      "loss": 0.1488,
+      "step": 6144
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9092698639942737e-05,
+      "loss": 0.1463,
+      "step": 6145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.909090909090909e-05,
+      "loss": 0.2521,
+      "step": 6146
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9089119541875454e-05,
+      "loss": 0.1193,
+      "step": 6147
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.90873299928418e-05,
+      "loss": 0.2562,
+      "step": 6148
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.908554044380816e-05,
+      "loss": 0.2255,
+      "step": 6149
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.908375089477452e-05,
+      "loss": 0.2203,
+      "step": 6150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9081961345740875e-05,
+      "loss": 0.1626,
+      "step": 6151
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.908017179670723e-05,
+      "loss": 0.2458,
+      "step": 6152
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.907838224767359e-05,
+      "loss": 0.3539,
+      "step": 6153
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.907659269863995e-05,
+      "loss": 0.149,
+      "step": 6154
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.90748031496063e-05,
+      "loss": 0.1605,
+      "step": 6155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.907301360057266e-05,
+      "loss": 0.2293,
+      "step": 6156
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.907122405153901e-05,
+      "loss": 0.2116,
+      "step": 6157
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.906943450250537e-05,
+      "loss": 0.168,
+      "step": 6158
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.906764495347173e-05,
+      "loss": 0.1978,
+      "step": 6159
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9065855404438085e-05,
+      "loss": 0.2929,
+      "step": 6160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.906406585540444e-05,
+      "loss": 0.198,
+      "step": 6161
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.90622763063708e-05,
+      "loss": 0.1534,
+      "step": 6162
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.906048675733715e-05,
+      "loss": 0.1979,
+      "step": 6163
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9058697208303506e-05,
+      "loss": 0.1323,
+      "step": 6164
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.905690765926987e-05,
+      "loss": 0.2047,
+      "step": 6165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.905511811023622e-05,
+      "loss": 0.3385,
+      "step": 6166
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.905332856120258e-05,
+      "loss": 0.2039,
+      "step": 6167
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.905153901216893e-05,
+      "loss": 0.2166,
+      "step": 6168
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9049749463135295e-05,
+      "loss": 0.2288,
+      "step": 6169
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.904795991410165e-05,
+      "loss": 0.2407,
+      "step": 6170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9046170365068006e-05,
+      "loss": 0.2836,
+      "step": 6171
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.904438081603436e-05,
+      "loss": 0.134,
+      "step": 6172
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9042591267000716e-05,
+      "loss": 0.1775,
+      "step": 6173
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.904080171796707e-05,
+      "loss": 0.2909,
+      "step": 6174
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.903901216893343e-05,
+      "loss": 0.2736,
+      "step": 6175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.903722261989979e-05,
+      "loss": 0.1104,
+      "step": 6176
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9035433070866143e-05,
+      "loss": 0.2798,
+      "step": 6177
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9033643521832505e-05,
+      "loss": 0.1799,
+      "step": 6178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9031853972798854e-05,
+      "loss": 0.3013,
+      "step": 6179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.903006442376521e-05,
+      "loss": 0.1674,
+      "step": 6180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.902827487473157e-05,
+      "loss": 0.2744,
+      "step": 6181
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9026485325697926e-05,
+      "loss": 0.2068,
+      "step": 6182
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.902469577666428e-05,
+      "loss": 0.1654,
+      "step": 6183
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.902290622763064e-05,
+      "loss": 0.1733,
+      "step": 6184
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9021116678597e-05,
+      "loss": 0.2151,
+      "step": 6185
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9019327129563354e-05,
+      "loss": 0.1633,
+      "step": 6186
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.901753758052971e-05,
+      "loss": 0.2423,
+      "step": 6187
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9015748031496064e-05,
+      "loss": 0.1754,
+      "step": 6188
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.901395848246242e-05,
+      "loss": 0.2502,
+      "step": 6189
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9012168933428774e-05,
+      "loss": 0.2234,
+      "step": 6190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9010379384395136e-05,
+      "loss": 0.217,
+      "step": 6191
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900858983536149e-05,
+      "loss": 0.232,
+      "step": 6192
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900680028632785e-05,
+      "loss": 0.2173,
+      "step": 6193
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900501073729421e-05,
+      "loss": 0.1784,
+      "step": 6194
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900322118826056e-05,
+      "loss": 0.176,
+      "step": 6195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900143163922691e-05,
+      "loss": 0.1883,
+      "step": 6196
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8999642090193274e-05,
+      "loss": 0.2791,
+      "step": 6197
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.899785254115963e-05,
+      "loss": 0.3011,
+      "step": 6198
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8996062992125985e-05,
+      "loss": 0.3585,
+      "step": 6199
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.899427344309235e-05,
+      "loss": 0.3279,
+      "step": 6200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.89924838940587e-05,
+      "loss": 0.344,
+      "step": 6201
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.899069434502505e-05,
+      "loss": 0.1893,
+      "step": 6202
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.898890479599141e-05,
+      "loss": 0.2396,
+      "step": 6203
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.898711524695777e-05,
+      "loss": 0.2422,
+      "step": 6204
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.898532569792412e-05,
+      "loss": 0.1877,
+      "step": 6205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8983536148890485e-05,
+      "loss": 0.3829,
+      "step": 6206
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.898174659985684e-05,
+      "loss": 0.1266,
+      "step": 6207
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8979957050823195e-05,
+      "loss": 0.1811,
+      "step": 6208
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.897816750178955e-05,
+      "loss": 0.1802,
+      "step": 6209
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8976377952755905e-05,
+      "loss": 0.1324,
+      "step": 6210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.897458840372226e-05,
+      "loss": 0.2751,
+      "step": 6211
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8972798854688616e-05,
+      "loss": 0.1443,
+      "step": 6212
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.897100930565498e-05,
+      "loss": 0.0818,
+      "step": 6213
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.896921975662133e-05,
+      "loss": 0.1295,
+      "step": 6214
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.896743020758769e-05,
+      "loss": 0.2167,
+      "step": 6215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.896564065855405e-05,
+      "loss": 0.0807,
+      "step": 6216
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8963851109520405e-05,
+      "loss": 0.1715,
+      "step": 6217
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8962061560486754e-05,
+      "loss": 0.3068,
+      "step": 6218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8960272011453116e-05,
+      "loss": 0.3267,
+      "step": 6219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.895848246241947e-05,
+      "loss": 0.261,
+      "step": 6220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8956692913385826e-05,
+      "loss": 0.2373,
+      "step": 6221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.895490336435219e-05,
+      "loss": 0.2898,
+      "step": 6222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.895311381531854e-05,
+      "loss": 0.2533,
+      "step": 6223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.89513242662849e-05,
+      "loss": 0.1812,
+      "step": 6224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8949534717251254e-05,
+      "loss": 0.3255,
+      "step": 6225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.894774516821761e-05,
+      "loss": 0.1326,
+      "step": 6226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8945955619183964e-05,
+      "loss": 0.2365,
+      "step": 6227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8944166070150326e-05,
+      "loss": 0.2678,
+      "step": 6228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.894237652111668e-05,
+      "loss": 0.2953,
+      "step": 6229
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8940586972083036e-05,
+      "loss": 0.2193,
+      "step": 6230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.893879742304939e-05,
+      "loss": 0.1732,
+      "step": 6231
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8937007874015754e-05,
+      "loss": 0.1926,
+      "step": 6232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.893521832498211e-05,
+      "loss": 0.1738,
+      "step": 6233
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8933428775948464e-05,
+      "loss": 0.1484,
+      "step": 6234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.893163922691482e-05,
+      "loss": 0.2213,
+      "step": 6235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8929849677881174e-05,
+      "loss": 0.2876,
+      "step": 6236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.892806012884753e-05,
+      "loss": 0.3103,
+      "step": 6237
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.892627057981389e-05,
+      "loss": 0.0904,
+      "step": 6238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.892448103078025e-05,
+      "loss": 0.3417,
+      "step": 6239
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.89226914817466e-05,
+      "loss": 0.2438,
+      "step": 6240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8920901932712964e-05,
+      "loss": 0.2748,
+      "step": 6241
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.891911238367931e-05,
+      "loss": 0.1324,
+      "step": 6242
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.891732283464567e-05,
+      "loss": 0.1966,
+      "step": 6243
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.891553328561203e-05,
+      "loss": 0.1633,
+      "step": 6244
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8913743736578385e-05,
+      "loss": 0.1332,
+      "step": 6245
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.891195418754474e-05,
+      "loss": 0.1628,
+      "step": 6246
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.89101646385111e-05,
+      "loss": 0.197,
+      "step": 6247
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.890837508947746e-05,
+      "loss": 0.2574,
+      "step": 6248
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8906585540443805e-05,
+      "loss": 0.3396,
+      "step": 6249
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.890479599141017e-05,
+      "loss": 0.3698,
+      "step": 6250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.890300644237652e-05,
+      "loss": 0.3381,
+      "step": 6251
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.890121689334288e-05,
+      "loss": 0.1786,
+      "step": 6252
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.889942734430923e-05,
+      "loss": 0.115,
+      "step": 6253
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8897637795275595e-05,
+      "loss": 0.1639,
+      "step": 6254
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.889584824624195e-05,
+      "loss": 0.2541,
+      "step": 6255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8894058697208305e-05,
+      "loss": 0.2426,
+      "step": 6256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.889226914817466e-05,
+      "loss": 0.3403,
+      "step": 6257
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8890479599141016e-05,
+      "loss": 0.2895,
+      "step": 6258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.888869005010737e-05,
+      "loss": 0.2666,
+      "step": 6259
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.888690050107373e-05,
+      "loss": 0.2514,
+      "step": 6260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.888511095204009e-05,
+      "loss": 0.1829,
+      "step": 6261
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.888332140300644e-05,
+      "loss": 0.2301,
+      "step": 6262
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8881531853972805e-05,
+      "loss": 0.2367,
+      "step": 6263
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.887974230493916e-05,
+      "loss": 0.2697,
+      "step": 6264
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.887795275590551e-05,
+      "loss": 0.2404,
+      "step": 6265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.887616320687187e-05,
+      "loss": 0.2488,
+      "step": 6266
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8874373657838226e-05,
+      "loss": 0.2907,
+      "step": 6267
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.887258410880458e-05,
+      "loss": 0.305,
+      "step": 6268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.887079455977094e-05,
+      "loss": 0.1149,
+      "step": 6269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.88690050107373e-05,
+      "loss": 0.1251,
+      "step": 6270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8867215461703654e-05,
+      "loss": 0.3492,
+      "step": 6271
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.886542591267001e-05,
+      "loss": 0.1941,
+      "step": 6272
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8863636363636364e-05,
+      "loss": 0.2352,
+      "step": 6273
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.886184681460272e-05,
+      "loss": 0.1345,
+      "step": 6274
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8860057265569074e-05,
+      "loss": 0.1734,
+      "step": 6275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8858267716535436e-05,
+      "loss": 0.2213,
+      "step": 6276
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.885647816750179e-05,
+      "loss": 0.2734,
+      "step": 6277
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.885468861846815e-05,
+      "loss": 0.222,
+      "step": 6278
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.885289906943451e-05,
+      "loss": 0.1456,
+      "step": 6279
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8851109520400864e-05,
+      "loss": 0.2121,
+      "step": 6280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.884931997136721e-05,
+      "loss": 0.1162,
+      "step": 6281
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8847530422333574e-05,
+      "loss": 0.2606,
+      "step": 6282
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.884574087329993e-05,
+      "loss": 0.2062,
+      "step": 6283
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8843951324266285e-05,
+      "loss": 0.1338,
+      "step": 6284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8842161775232647e-05,
+      "loss": 0.3157,
+      "step": 6285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8840372226199e-05,
+      "loss": 0.1731,
+      "step": 6286
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.883858267716536e-05,
+      "loss": 0.1233,
+      "step": 6287
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.883679312813171e-05,
+      "loss": 0.2526,
+      "step": 6288
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.883500357909807e-05,
+      "loss": 0.2429,
+      "step": 6289
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.883321403006442e-05,
+      "loss": 0.1666,
+      "step": 6290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8831424481030785e-05,
+      "loss": 0.4473,
+      "step": 6291
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.882963493199714e-05,
+      "loss": 0.1958,
+      "step": 6292
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8827845382963495e-05,
+      "loss": 0.163,
+      "step": 6293
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.882605583392985e-05,
+      "loss": 0.2394,
+      "step": 6294
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.882426628489621e-05,
+      "loss": 0.1899,
+      "step": 6295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.882247673586256e-05,
+      "loss": 0.2441,
+      "step": 6296
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8820687186828916e-05,
+      "loss": 0.2032,
+      "step": 6297
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881889763779528e-05,
+      "loss": 0.1687,
+      "step": 6298
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881710808876163e-05,
+      "loss": 0.2626,
+      "step": 6299
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881531853972799e-05,
+      "loss": 0.1932,
+      "step": 6300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881352899069435e-05,
+      "loss": 0.2203,
+      "step": 6301
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8811739441660705e-05,
+      "loss": 0.2237,
+      "step": 6302
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880994989262706e-05,
+      "loss": 0.1708,
+      "step": 6303
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8808160343593416e-05,
+      "loss": 0.3131,
+      "step": 6304
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880637079455977e-05,
+      "loss": 0.27,
+      "step": 6305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8804581245526126e-05,
+      "loss": 0.2011,
+      "step": 6306
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880279169649249e-05,
+      "loss": 0.2347,
+      "step": 6307
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880100214745884e-05,
+      "loss": 0.2517,
+      "step": 6308
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.87992125984252e-05,
+      "loss": 0.2118,
+      "step": 6309
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.879742304939156e-05,
+      "loss": 0.1283,
+      "step": 6310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8795633500357915e-05,
+      "loss": 0.2281,
+      "step": 6311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8793843951324264e-05,
+      "loss": 0.1897,
+      "step": 6312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8792054402290626e-05,
+      "loss": 0.1708,
+      "step": 6313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.879026485325698e-05,
+      "loss": 0.2325,
+      "step": 6314
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8788475304223336e-05,
+      "loss": 0.2169,
+      "step": 6315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.878668575518969e-05,
+      "loss": 0.2324,
+      "step": 6316
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8784896206156053e-05,
+      "loss": 0.1347,
+      "step": 6317
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.878310665712241e-05,
+      "loss": 0.2341,
+      "step": 6318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8781317108088764e-05,
+      "loss": 0.3013,
+      "step": 6319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.877952755905512e-05,
+      "loss": 0.2509,
+      "step": 6320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8777738010021474e-05,
+      "loss": 0.1772,
+      "step": 6321
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.877594846098783e-05,
+      "loss": 0.2134,
+      "step": 6322
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.877415891195419e-05,
+      "loss": 0.205,
+      "step": 6323
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8772369362920547e-05,
+      "loss": 0.2005,
+      "step": 6324
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.87705798138869e-05,
+      "loss": 0.2336,
+      "step": 6325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8768790264853264e-05,
+      "loss": 0.1969,
+      "step": 6326
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.876700071581962e-05,
+      "loss": 0.1844,
+      "step": 6327
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.876521116678597e-05,
+      "loss": 0.2927,
+      "step": 6328
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.876342161775233e-05,
+      "loss": 0.1884,
+      "step": 6329
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8761632068718684e-05,
+      "loss": 0.0921,
+      "step": 6330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.875984251968504e-05,
+      "loss": 0.2527,
+      "step": 6331
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.87580529706514e-05,
+      "loss": 0.2111,
+      "step": 6332
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.875626342161776e-05,
+      "loss": 0.211,
+      "step": 6333
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.875447387258411e-05,
+      "loss": 0.2136,
+      "step": 6334
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.875268432355047e-05,
+      "loss": 0.2487,
+      "step": 6335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.875089477451682e-05,
+      "loss": 0.0987,
+      "step": 6336
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.874910522548318e-05,
+      "loss": 0.2253,
+      "step": 6337
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.874731567644953e-05,
+      "loss": 0.1358,
+      "step": 6338
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8745526127415895e-05,
+      "loss": 0.1485,
+      "step": 6339
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.874373657838225e-05,
+      "loss": 0.1417,
+      "step": 6340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8741947029348605e-05,
+      "loss": 0.2588,
+      "step": 6341
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.874015748031497e-05,
+      "loss": 0.2008,
+      "step": 6342
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8738367931281316e-05,
+      "loss": 0.1593,
+      "step": 6343
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.873657838224767e-05,
+      "loss": 0.2062,
+      "step": 6344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.873478883321403e-05,
+      "loss": 0.2462,
+      "step": 6345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.873299928418039e-05,
+      "loss": 0.1752,
+      "step": 6346
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.873120973514674e-05,
+      "loss": 0.2159,
+      "step": 6347
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8729420186113105e-05,
+      "loss": 0.283,
+      "step": 6348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.872763063707946e-05,
+      "loss": 0.1747,
+      "step": 6349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8725841088045815e-05,
+      "loss": 0.1286,
+      "step": 6350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.872405153901217e-05,
+      "loss": 0.1938,
+      "step": 6351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8722261989978526e-05,
+      "loss": 0.158,
+      "step": 6352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.872047244094488e-05,
+      "loss": 0.2397,
+      "step": 6353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871868289191124e-05,
+      "loss": 0.1998,
+      "step": 6354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.87168933428776e-05,
+      "loss": 0.2836,
+      "step": 6355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871510379384395e-05,
+      "loss": 0.1559,
+      "step": 6356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871331424481031e-05,
+      "loss": 0.1616,
+      "step": 6357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871152469577667e-05,
+      "loss": 0.302,
+      "step": 6358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.870973514674302e-05,
+      "loss": 0.3583,
+      "step": 6359
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8707945597709374e-05,
+      "loss": 0.2873,
+      "step": 6360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8706156048675736e-05,
+      "loss": 0.2025,
+      "step": 6361
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.870436649964209e-05,
+      "loss": 0.1436,
+      "step": 6362
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8702576950608446e-05,
+      "loss": 0.2348,
+      "step": 6363
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.870078740157481e-05,
+      "loss": 0.2449,
+      "step": 6364
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8698997852541164e-05,
+      "loss": 0.2934,
+      "step": 6365
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.869720830350752e-05,
+      "loss": 0.208,
+      "step": 6366
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8695418754473874e-05,
+      "loss": 0.2716,
+      "step": 6367
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.869362920544023e-05,
+      "loss": 0.3332,
+      "step": 6368
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8691839656406584e-05,
+      "loss": 0.3559,
+      "step": 6369
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8690050107372946e-05,
+      "loss": 0.1031,
+      "step": 6370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.86882605583393e-05,
+      "loss": 0.1431,
+      "step": 6371
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.868647100930566e-05,
+      "loss": 0.1872,
+      "step": 6372
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.868468146027201e-05,
+      "loss": 0.2534,
+      "step": 6373
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8682891911238374e-05,
+      "loss": 0.2377,
+      "step": 6374
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.868110236220472e-05,
+      "loss": 0.1661,
+      "step": 6375
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8679312813171084e-05,
+      "loss": 0.2458,
+      "step": 6376
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.867752326413744e-05,
+      "loss": 0.2364,
+      "step": 6377
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8675733715103795e-05,
+      "loss": 0.2103,
+      "step": 6378
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.867394416607015e-05,
+      "loss": 0.1429,
+      "step": 6379
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.867215461703651e-05,
+      "loss": 0.1012,
+      "step": 6380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.867036506800287e-05,
+      "loss": 0.167,
+      "step": 6381
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.866857551896922e-05,
+      "loss": 0.1817,
+      "step": 6382
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.866678596993558e-05,
+      "loss": 0.1567,
+      "step": 6383
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.866499642090193e-05,
+      "loss": 0.1866,
+      "step": 6384
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.866320687186829e-05,
+      "loss": 0.3088,
+      "step": 6385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.866141732283465e-05,
+      "loss": 0.3205,
+      "step": 6386
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8659627773801005e-05,
+      "loss": 0.1492,
+      "step": 6387
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.865783822476736e-05,
+      "loss": 0.2759,
+      "step": 6388
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.865604867573372e-05,
+      "loss": 0.4853,
+      "step": 6389
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.865425912670007e-05,
+      "loss": 0.1348,
+      "step": 6390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8652469577666426e-05,
+      "loss": 0.1483,
+      "step": 6391
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.865068002863279e-05,
+      "loss": 0.2343,
+      "step": 6392
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.864889047959914e-05,
+      "loss": 0.192,
+      "step": 6393
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.86471009305655e-05,
+      "loss": 0.3158,
+      "step": 6394
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.864531138153186e-05,
+      "loss": 0.1996,
+      "step": 6395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8643521832498215e-05,
+      "loss": 0.2114,
+      "step": 6396
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.864173228346457e-05,
+      "loss": 0.1659,
+      "step": 6397
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8639942734430926e-05,
+      "loss": 0.2316,
+      "step": 6398
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.863815318539728e-05,
+      "loss": 0.2199,
+      "step": 6399
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.2371,
+      "step": 6400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.863457408732999e-05,
+      "loss": 0.1616,
+      "step": 6401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.863278453829635e-05,
+      "loss": 0.2032,
+      "step": 6402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.863099498926271e-05,
+      "loss": 0.2525,
+      "step": 6403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8629205440229064e-05,
+      "loss": 0.1738,
+      "step": 6404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8627415891195426e-05,
+      "loss": 0.1342,
+      "step": 6405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8625626342161774e-05,
+      "loss": 0.2347,
+      "step": 6406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.862383679312813e-05,
+      "loss": 0.1411,
+      "step": 6407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.862204724409449e-05,
+      "loss": 0.1839,
+      "step": 6408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8620257695060846e-05,
+      "loss": 0.247,
+      "step": 6409
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86184681460272e-05,
+      "loss": 0.0831,
+      "step": 6410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8616678596993564e-05,
+      "loss": 0.1511,
+      "step": 6411
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.861488904795992e-05,
+      "loss": 0.1709,
+      "step": 6412
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8613099498926274e-05,
+      "loss": 0.1824,
+      "step": 6413
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.861130994989263e-05,
+      "loss": 0.2991,
+      "step": 6414
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8609520400858984e-05,
+      "loss": 0.1999,
+      "step": 6415
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.860773085182534e-05,
+      "loss": 0.2201,
+      "step": 6416
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86059413027917e-05,
+      "loss": 0.2863,
+      "step": 6417
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.860415175375806e-05,
+      "loss": 0.0723,
+      "step": 6418
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.860236220472441e-05,
+      "loss": 0.1565,
+      "step": 6419
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.860057265569077e-05,
+      "loss": 0.148,
+      "step": 6420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.859878310665712e-05,
+      "loss": 0.2065,
+      "step": 6421
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.859699355762348e-05,
+      "loss": 0.4905,
+      "step": 6422
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.859520400858983e-05,
+      "loss": 0.2907,
+      "step": 6423
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8593414459556195e-05,
+      "loss": 0.3161,
+      "step": 6424
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.859162491052255e-05,
+      "loss": 0.3642,
+      "step": 6425
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8589835361488905e-05,
+      "loss": 0.2446,
+      "step": 6426
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858804581245527e-05,
+      "loss": 0.3003,
+      "step": 6427
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858625626342162e-05,
+      "loss": 0.2144,
+      "step": 6428
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858446671438797e-05,
+      "loss": 0.1575,
+      "step": 6429
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858267716535433e-05,
+      "loss": 0.1361,
+      "step": 6430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858088761632069e-05,
+      "loss": 0.1628,
+      "step": 6431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.857909806728704e-05,
+      "loss": 0.2913,
+      "step": 6432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8577308518253405e-05,
+      "loss": 0.1401,
+      "step": 6433
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.857551896921976e-05,
+      "loss": 0.2046,
+      "step": 6434
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8573729420186115e-05,
+      "loss": 0.1757,
+      "step": 6435
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.857193987115247e-05,
+      "loss": 0.1409,
+      "step": 6436
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8570150322118826e-05,
+      "loss": 0.2411,
+      "step": 6437
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.856836077308518e-05,
+      "loss": 0.1919,
+      "step": 6438
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.856657122405154e-05,
+      "loss": 0.1983,
+      "step": 6439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.85647816750179e-05,
+      "loss": 0.2411,
+      "step": 6440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.856299212598425e-05,
+      "loss": 0.2509,
+      "step": 6441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.856120257695061e-05,
+      "loss": 0.2333,
+      "step": 6442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.855941302791697e-05,
+      "loss": 0.2526,
+      "step": 6443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8557623478883326e-05,
+      "loss": 0.1847,
+      "step": 6444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.855583392984968e-05,
+      "loss": 0.1312,
+      "step": 6445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8554044380816036e-05,
+      "loss": 0.3056,
+      "step": 6446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.855225483178239e-05,
+      "loss": 0.1557,
+      "step": 6447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8550465282748746e-05,
+      "loss": 0.2297,
+      "step": 6448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854867573371511e-05,
+      "loss": 0.2535,
+      "step": 6449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8546886184681463e-05,
+      "loss": 0.267,
+      "step": 6450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854509663564782e-05,
+      "loss": 0.153,
+      "step": 6451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854330708661418e-05,
+      "loss": 0.1428,
+      "step": 6452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.854151753758053e-05,
+      "loss": 0.4211,
+      "step": 6453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8539727988546884e-05,
+      "loss": 0.1857,
+      "step": 6454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8537938439513246e-05,
+      "loss": 0.1517,
+      "step": 6455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.85361488904796e-05,
+      "loss": 0.1802,
+      "step": 6456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8534359341445957e-05,
+      "loss": 0.1175,
+      "step": 6457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.853256979241232e-05,
+      "loss": 0.1371,
+      "step": 6458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8530780243378674e-05,
+      "loss": 0.2442,
+      "step": 6459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.852899069434503e-05,
+      "loss": 0.1875,
+      "step": 6460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8527201145311384e-05,
+      "loss": 0.2548,
+      "step": 6461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.852541159627774e-05,
+      "loss": 0.2601,
+      "step": 6462
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8523622047244095e-05,
+      "loss": 0.2932,
+      "step": 6463
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.852183249821045e-05,
+      "loss": 0.2552,
+      "step": 6464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.852004294917681e-05,
+      "loss": 0.2773,
+      "step": 6465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851825340014317e-05,
+      "loss": 0.2251,
+      "step": 6466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851646385110952e-05,
+      "loss": 0.2638,
+      "step": 6467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851467430207588e-05,
+      "loss": 0.2441,
+      "step": 6468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851288475304223e-05,
+      "loss": 0.2065,
+      "step": 6469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851109520400859e-05,
+      "loss": 0.2819,
+      "step": 6470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850930565497495e-05,
+      "loss": 0.2114,
+      "step": 6471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8507516105941305e-05,
+      "loss": 0.286,
+      "step": 6472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850572655690766e-05,
+      "loss": 0.2444,
+      "step": 6473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850393700787402e-05,
+      "loss": 0.1801,
+      "step": 6474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850214745884038e-05,
+      "loss": 0.1979,
+      "step": 6475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8500357909806726e-05,
+      "loss": 0.1595,
+      "step": 6476
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.849856836077309e-05,
+      "loss": 0.23,
+      "step": 6477
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.849677881173944e-05,
+      "loss": 0.1407,
+      "step": 6478
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.84949892627058e-05,
+      "loss": 0.2808,
+      "step": 6479
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.849319971367216e-05,
+      "loss": 0.2573,
+      "step": 6480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8491410164638515e-05,
+      "loss": 0.2016,
+      "step": 6481
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.848962061560487e-05,
+      "loss": 0.1875,
+      "step": 6482
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8487831066571225e-05,
+      "loss": 0.2904,
+      "step": 6483
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.848604151753758e-05,
+      "loss": 0.1155,
+      "step": 6484
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8484251968503936e-05,
+      "loss": 0.2645,
+      "step": 6485
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.848246241947029e-05,
+      "loss": 0.2007,
+      "step": 6486
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.848067287043665e-05,
+      "loss": 0.2037,
+      "step": 6487
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.847888332140301e-05,
+      "loss": 0.0884,
+      "step": 6488
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8477093772369363e-05,
+      "loss": 0.3273,
+      "step": 6489
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8475304223335725e-05,
+      "loss": 0.1655,
+      "step": 6490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.847351467430208e-05,
+      "loss": 0.1897,
+      "step": 6491
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.847172512526843e-05,
+      "loss": 0.3859,
+      "step": 6492
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.846993557623479e-05,
+      "loss": 0.2989,
+      "step": 6493
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8468146027201146e-05,
+      "loss": 0.1655,
+      "step": 6494
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.84663564781675e-05,
+      "loss": 0.1488,
+      "step": 6495
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.846456692913386e-05,
+      "loss": 0.1712,
+      "step": 6496
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.846277738010022e-05,
+      "loss": 0.1934,
+      "step": 6497
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8460987831066574e-05,
+      "loss": 0.115,
+      "step": 6498
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.845919828203293e-05,
+      "loss": 0.2442,
+      "step": 6499
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8457408732999284e-05,
+      "loss": 0.246,
+      "step": 6500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.845561918396564e-05,
+      "loss": 0.2349,
+      "step": 6501
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8453829634932e-05,
+      "loss": 0.2921,
+      "step": 6502
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8452040085898356e-05,
+      "loss": 0.2887,
+      "step": 6503
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.845025053686471e-05,
+      "loss": 0.235,
+      "step": 6504
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.844846098783107e-05,
+      "loss": 0.1758,
+      "step": 6505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.844667143879743e-05,
+      "loss": 0.3573,
+      "step": 6506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8444881889763784e-05,
+      "loss": 0.2056,
+      "step": 6507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.844309234073013e-05,
+      "loss": 0.1832,
+      "step": 6508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8441302791696494e-05,
+      "loss": 0.209,
+      "step": 6509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.843951324266285e-05,
+      "loss": 0.408,
+      "step": 6510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8437723693629205e-05,
+      "loss": 0.253,
+      "step": 6511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.843593414459557e-05,
+      "loss": 0.1474,
+      "step": 6512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.843414459556192e-05,
+      "loss": 0.4712,
+      "step": 6513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.843235504652828e-05,
+      "loss": 0.2196,
+      "step": 6514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.843056549749463e-05,
+      "loss": 0.2567,
+      "step": 6515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.842877594846099e-05,
+      "loss": 0.3353,
+      "step": 6516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.842698639942734e-05,
+      "loss": 0.1817,
+      "step": 6517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8425196850393705e-05,
+      "loss": 0.227,
+      "step": 6518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.842340730136006e-05,
+      "loss": 0.1698,
+      "step": 6519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8421617752326415e-05,
+      "loss": 0.3139,
+      "step": 6520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.841982820329278e-05,
+      "loss": 0.1937,
+      "step": 6521
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841803865425913e-05,
+      "loss": 0.222,
+      "step": 6522
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841624910522548e-05,
+      "loss": 0.241,
+      "step": 6523
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841445955619184e-05,
+      "loss": 0.2009,
+      "step": 6524
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.84126700071582e-05,
+      "loss": 0.3962,
+      "step": 6525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841088045812455e-05,
+      "loss": 0.178,
+      "step": 6526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840909090909091e-05,
+      "loss": 0.2053,
+      "step": 6527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840730136005727e-05,
+      "loss": 0.2145,
+      "step": 6528
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8405511811023625e-05,
+      "loss": 0.2449,
+      "step": 6529
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840372226198998e-05,
+      "loss": 0.2076,
+      "step": 6530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8401932712956336e-05,
+      "loss": 0.1606,
+      "step": 6531
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840014316392269e-05,
+      "loss": 0.2051,
+      "step": 6532
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8398353614889046e-05,
+      "loss": 0.2591,
+      "step": 6533
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.839656406585541e-05,
+      "loss": 0.1337,
+      "step": 6534
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.839477451682176e-05,
+      "loss": 0.1307,
+      "step": 6535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.839298496778812e-05,
+      "loss": 0.279,
+      "step": 6536
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.839119541875448e-05,
+      "loss": 0.3564,
+      "step": 6537
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8389405869720836e-05,
+      "loss": 0.2445,
+      "step": 6538
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8387616320687184e-05,
+      "loss": 0.2133,
+      "step": 6539
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8385826771653546e-05,
+      "loss": 0.2297,
+      "step": 6540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.83840372226199e-05,
+      "loss": 0.3115,
+      "step": 6541
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8382247673586256e-05,
+      "loss": 0.2225,
+      "step": 6542
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.838045812455262e-05,
+      "loss": 0.1654,
+      "step": 6543
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8378668575518974e-05,
+      "loss": 0.2876,
+      "step": 6544
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.837687902648533e-05,
+      "loss": 0.1681,
+      "step": 6545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8375089477451684e-05,
+      "loss": 0.2188,
+      "step": 6546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.837329992841804e-05,
+      "loss": 0.068,
+      "step": 6547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8371510379384394e-05,
+      "loss": 0.1734,
+      "step": 6548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.836972083035075e-05,
+      "loss": 0.2512,
+      "step": 6549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.836793128131711e-05,
+      "loss": 0.164,
+      "step": 6550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.836614173228347e-05,
+      "loss": 0.2779,
+      "step": 6551
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.836435218324982e-05,
+      "loss": 0.201,
+      "step": 6552
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8362562634216184e-05,
+      "loss": 0.3766,
+      "step": 6553
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.836077308518253e-05,
+      "loss": 0.2518,
+      "step": 6554
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.835898353614889e-05,
+      "loss": 0.2914,
+      "step": 6555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.835719398711525e-05,
+      "loss": 0.2034,
+      "step": 6556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8355404438081605e-05,
+      "loss": 0.143,
+      "step": 6557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.835361488904796e-05,
+      "loss": 0.2205,
+      "step": 6558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.835182534001432e-05,
+      "loss": 0.2547,
+      "step": 6559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.835003579098068e-05,
+      "loss": 0.2212,
+      "step": 6560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.834824624194703e-05,
+      "loss": 0.1017,
+      "step": 6561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.834645669291339e-05,
+      "loss": 0.1257,
+      "step": 6562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.834466714387974e-05,
+      "loss": 0.1863,
+      "step": 6563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.83428775948461e-05,
+      "loss": 0.1706,
+      "step": 6564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.834108804581246e-05,
+      "loss": 0.1428,
+      "step": 6565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8339298496778815e-05,
+      "loss": 0.1812,
+      "step": 6566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.833750894774517e-05,
+      "loss": 0.2701,
+      "step": 6567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8335719398711525e-05,
+      "loss": 0.3433,
+      "step": 6568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.833392984967789e-05,
+      "loss": 0.2677,
+      "step": 6569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8332140300644236e-05,
+      "loss": 0.2183,
+      "step": 6570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.833035075161059e-05,
+      "loss": 0.1376,
+      "step": 6571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.832856120257695e-05,
+      "loss": 0.1764,
+      "step": 6572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.832677165354331e-05,
+      "loss": 0.1858,
+      "step": 6573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.832498210450966e-05,
+      "loss": 0.1915,
+      "step": 6574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8323192555476025e-05,
+      "loss": 0.2856,
+      "step": 6575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.832140300644238e-05,
+      "loss": 0.3103,
+      "step": 6576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8319613457408736e-05,
+      "loss": 0.1576,
+      "step": 6577
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.831782390837509e-05,
+      "loss": 0.2994,
+      "step": 6578
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8316034359341446e-05,
+      "loss": 0.164,
+      "step": 6579
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.83142448103078e-05,
+      "loss": 0.1475,
+      "step": 6580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.831245526127416e-05,
+      "loss": 0.329,
+      "step": 6581
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.831066571224052e-05,
+      "loss": 0.1189,
+      "step": 6582
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8308876163206874e-05,
+      "loss": 0.1727,
+      "step": 6583
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.830708661417323e-05,
+      "loss": 0.1495,
+      "step": 6584
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.830529706513959e-05,
+      "loss": 0.1668,
+      "step": 6585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.830350751610594e-05,
+      "loss": 0.2228,
+      "step": 6586
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.83017179670723e-05,
+      "loss": 0.1626,
+      "step": 6587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8299928418038656e-05,
+      "loss": 0.1684,
+      "step": 6588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.829813886900501e-05,
+      "loss": 0.2738,
+      "step": 6589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.829634931997137e-05,
+      "loss": 0.2716,
+      "step": 6590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.829455977093773e-05,
+      "loss": 0.1895,
+      "step": 6591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8292770221904084e-05,
+      "loss": 0.2245,
+      "step": 6592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.829098067287044e-05,
+      "loss": 0.1964,
+      "step": 6593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8289191123836794e-05,
+      "loss": 0.2228,
+      "step": 6594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.828740157480315e-05,
+      "loss": 0.113,
+      "step": 6595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8285612025769505e-05,
+      "loss": 0.2735,
+      "step": 6596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8283822476735867e-05,
+      "loss": 0.1929,
+      "step": 6597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.828203292770222e-05,
+      "loss": 0.2761,
+      "step": 6598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.828024337866858e-05,
+      "loss": 0.265,
+      "step": 6599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.827845382963494e-05,
+      "loss": 0.2509,
+      "step": 6600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.827666428060129e-05,
+      "loss": 0.1899,
+      "step": 6601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.827487473156764e-05,
+      "loss": 0.2679,
+      "step": 6602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8273085182534004e-05,
+      "loss": 0.2421,
+      "step": 6603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.827129563350036e-05,
+      "loss": 0.1694,
+      "step": 6604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8269506084466715e-05,
+      "loss": 0.2882,
+      "step": 6605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.826771653543308e-05,
+      "loss": 0.2892,
+      "step": 6606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.826592698639943e-05,
+      "loss": 0.3464,
+      "step": 6607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.826413743736579e-05,
+      "loss": 0.3322,
+      "step": 6608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.826234788833214e-05,
+      "loss": 0.1967,
+      "step": 6609
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.82605583392985e-05,
+      "loss": 0.2287,
+      "step": 6610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.825876879026485e-05,
+      "loss": 0.3176,
+      "step": 6611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.825697924123121e-05,
+      "loss": 0.2793,
+      "step": 6612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.825518969219757e-05,
+      "loss": 0.2446,
+      "step": 6613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8253400143163925e-05,
+      "loss": 0.2385,
+      "step": 6614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.825161059413028e-05,
+      "loss": 0.2317,
+      "step": 6615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.824982104509664e-05,
+      "loss": 0.2851,
+      "step": 6616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.824803149606299e-05,
+      "loss": 0.2065,
+      "step": 6617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8246241947029346e-05,
+      "loss": 0.1864,
+      "step": 6618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.824445239799571e-05,
+      "loss": 0.2131,
+      "step": 6619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.824266284896206e-05,
+      "loss": 0.1282,
+      "step": 6620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.824087329992842e-05,
+      "loss": 0.2018,
+      "step": 6621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.823908375089478e-05,
+      "loss": 0.16,
+      "step": 6622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8237294201861135e-05,
+      "loss": 0.3488,
+      "step": 6623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.823550465282749e-05,
+      "loss": 0.1348,
+      "step": 6624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8233715103793846e-05,
+      "loss": 0.2245,
+      "step": 6625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.82319255547602e-05,
+      "loss": 0.2504,
+      "step": 6626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8230136005726556e-05,
+      "loss": 0.1899,
+      "step": 6627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.822834645669292e-05,
+      "loss": 0.1852,
+      "step": 6628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8226556907659273e-05,
+      "loss": 0.2944,
+      "step": 6629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.822476735862563e-05,
+      "loss": 0.1722,
+      "step": 6630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8222977809591984e-05,
+      "loss": 0.3637,
+      "step": 6631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8221188260558346e-05,
+      "loss": 0.2225,
+      "step": 6632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8219398711524694e-05,
+      "loss": 0.2032,
+      "step": 6633
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.821760916249105e-05,
+      "loss": 0.3369,
+      "step": 6634
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.821581961345741e-05,
+      "loss": 0.1254,
+      "step": 6635
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8214030064423767e-05,
+      "loss": 0.3434,
+      "step": 6636
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.821224051539012e-05,
+      "loss": 0.3374,
+      "step": 6637
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8210450966356484e-05,
+      "loss": 0.1168,
+      "step": 6638
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820866141732284e-05,
+      "loss": 0.1926,
+      "step": 6639
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820687186828919e-05,
+      "loss": 0.1558,
+      "step": 6640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820508231925555e-05,
+      "loss": 0.2952,
+      "step": 6641
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8203292770221904e-05,
+      "loss": 0.298,
+      "step": 6642
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820150322118826e-05,
+      "loss": 0.1048,
+      "step": 6643
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819971367215462e-05,
+      "loss": 0.1679,
+      "step": 6644
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819792412312098e-05,
+      "loss": 0.1564,
+      "step": 6645
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819613457408733e-05,
+      "loss": 0.2649,
+      "step": 6646
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819434502505369e-05,
+      "loss": 0.1833,
+      "step": 6647
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819255547602004e-05,
+      "loss": 0.1214,
+      "step": 6648
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.81907659269864e-05,
+      "loss": 0.1826,
+      "step": 6649
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.818897637795276e-05,
+      "loss": 0.146,
+      "step": 6650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8187186828919115e-05,
+      "loss": 0.348,
+      "step": 6651
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.818539727988547e-05,
+      "loss": 0.0639,
+      "step": 6652
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8183607730851825e-05,
+      "loss": 0.2153,
+      "step": 6653
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.818181818181819e-05,
+      "loss": 0.3011,
+      "step": 6654
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.818002863278454e-05,
+      "loss": 0.2202,
+      "step": 6655
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.817823908375089e-05,
+      "loss": 0.2113,
+      "step": 6656
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.817644953471725e-05,
+      "loss": 0.2095,
+      "step": 6657
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.817465998568361e-05,
+      "loss": 0.1519,
+      "step": 6658
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.817287043664996e-05,
+      "loss": 0.1611,
+      "step": 6659
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8171080887616325e-05,
+      "loss": 0.171,
+      "step": 6660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.816929133858268e-05,
+      "loss": 0.3302,
+      "step": 6661
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8167501789549035e-05,
+      "loss": 0.1898,
+      "step": 6662
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.81657122405154e-05,
+      "loss": 0.1976,
+      "step": 6663
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8163922691481746e-05,
+      "loss": 0.2527,
+      "step": 6664
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.81621331424481e-05,
+      "loss": 0.1661,
+      "step": 6665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.816034359341446e-05,
+      "loss": 0.2775,
+      "step": 6666
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.815855404438082e-05,
+      "loss": 0.135,
+      "step": 6667
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.815676449534717e-05,
+      "loss": 0.263,
+      "step": 6668
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8154974946313535e-05,
+      "loss": 0.3056,
+      "step": 6669
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.815318539727989e-05,
+      "loss": 0.1656,
+      "step": 6670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8151395848246246e-05,
+      "loss": 0.1382,
+      "step": 6671
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.81496062992126e-05,
+      "loss": 0.1246,
+      "step": 6672
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8147816750178956e-05,
+      "loss": 0.3072,
+      "step": 6673
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.814602720114531e-05,
+      "loss": 0.2034,
+      "step": 6674
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8144237652111666e-05,
+      "loss": 0.1529,
+      "step": 6675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.814244810307803e-05,
+      "loss": 0.1366,
+      "step": 6676
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8140658554044384e-05,
+      "loss": 0.3113,
+      "step": 6677
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.813886900501074e-05,
+      "loss": 0.2056,
+      "step": 6678
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.81370794559771e-05,
+      "loss": 0.1858,
+      "step": 6679
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.813528990694345e-05,
+      "loss": 0.1688,
+      "step": 6680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8133500357909804e-05,
+      "loss": 0.163,
+      "step": 6681
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8131710808876166e-05,
+      "loss": 0.1721,
+      "step": 6682
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.812992125984252e-05,
+      "loss": 0.2067,
+      "step": 6683
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.812813171080888e-05,
+      "loss": 0.2208,
+      "step": 6684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.812634216177524e-05,
+      "loss": 0.25,
+      "step": 6685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8124552612741594e-05,
+      "loss": 0.1279,
+      "step": 6686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.812276306370794e-05,
+      "loss": 0.1997,
+      "step": 6687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8120973514674304e-05,
+      "loss": 0.1267,
+      "step": 6688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.811918396564066e-05,
+      "loss": 0.2319,
+      "step": 6689
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8117394416607015e-05,
+      "loss": 0.1872,
+      "step": 6690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.811560486757338e-05,
+      "loss": 0.2166,
+      "step": 6691
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.811381531853973e-05,
+      "loss": 0.2025,
+      "step": 6692
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.811202576950609e-05,
+      "loss": 0.1938,
+      "step": 6693
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.811023622047244e-05,
+      "loss": 0.1317,
+      "step": 6694
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.81084466714388e-05,
+      "loss": 0.2178,
+      "step": 6695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.810665712240515e-05,
+      "loss": 0.2537,
+      "step": 6696
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.810486757337151e-05,
+      "loss": 0.226,
+      "step": 6697
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.810307802433787e-05,
+      "loss": 0.213,
+      "step": 6698
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8101288475304225e-05,
+      "loss": 0.2732,
+      "step": 6699
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.809949892627058e-05,
+      "loss": 0.1788,
+      "step": 6700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.809770937723694e-05,
+      "loss": 0.1308,
+      "step": 6701
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.80959198282033e-05,
+      "loss": 0.1114,
+      "step": 6702
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8094130279169646e-05,
+      "loss": 0.2591,
+      "step": 6703
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.809234073013601e-05,
+      "loss": 0.1832,
+      "step": 6704
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.809055118110236e-05,
+      "loss": 0.2699,
+      "step": 6705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.808876163206872e-05,
+      "loss": 0.1651,
+      "step": 6706
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.808697208303508e-05,
+      "loss": 0.2841,
+      "step": 6707
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8085182534001435e-05,
+      "loss": 0.1699,
+      "step": 6708
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.808339298496779e-05,
+      "loss": 0.1653,
+      "step": 6709
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8081603435934146e-05,
+      "loss": 0.218,
+      "step": 6710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.80798138869005e-05,
+      "loss": 0.2994,
+      "step": 6711
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8078024337866856e-05,
+      "loss": 0.1647,
+      "step": 6712
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.807623478883322e-05,
+      "loss": 0.2438,
+      "step": 6713
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.807444523979957e-05,
+      "loss": 0.1584,
+      "step": 6714
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.807265569076593e-05,
+      "loss": 0.2928,
+      "step": 6715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8070866141732284e-05,
+      "loss": 0.1477,
+      "step": 6716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8069076592698646e-05,
+      "loss": 0.3259,
+      "step": 6717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8067287043665e-05,
+      "loss": 0.2886,
+      "step": 6718
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.806549749463135e-05,
+      "loss": 0.2758,
+      "step": 6719
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.806370794559771e-05,
+      "loss": 0.1796,
+      "step": 6720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8061918396564066e-05,
+      "loss": 0.2435,
+      "step": 6721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.806012884753042e-05,
+      "loss": 0.3122,
+      "step": 6722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8058339298496784e-05,
+      "loss": 0.256,
+      "step": 6723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.805654974946314e-05,
+      "loss": 0.1588,
+      "step": 6724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8054760200429494e-05,
+      "loss": 0.1859,
+      "step": 6725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8052970651395856e-05,
+      "loss": 0.2465,
+      "step": 6726
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8051181102362204e-05,
+      "loss": 0.3971,
+      "step": 6727
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804939155332856e-05,
+      "loss": 0.255,
+      "step": 6728
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804760200429492e-05,
+      "loss": 0.2773,
+      "step": 6729
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804581245526128e-05,
+      "loss": 0.1858,
+      "step": 6730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804402290622763e-05,
+      "loss": 0.1822,
+      "step": 6731
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804223335719399e-05,
+      "loss": 0.1947,
+      "step": 6732
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.804044380816035e-05,
+      "loss": 0.1572,
+      "step": 6733
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.80386542591267e-05,
+      "loss": 0.1446,
+      "step": 6734
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.803686471009306e-05,
+      "loss": 0.1784,
+      "step": 6735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8035075161059415e-05,
+      "loss": 0.2135,
+      "step": 6736
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.803328561202577e-05,
+      "loss": 0.2289,
+      "step": 6737
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8031496062992125e-05,
+      "loss": 0.12,
+      "step": 6738
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.802970651395849e-05,
+      "loss": 0.1837,
+      "step": 6739
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.802791696492484e-05,
+      "loss": 0.198,
+      "step": 6740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.80261274158912e-05,
+      "loss": 0.1111,
+      "step": 6741
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.802433786685755e-05,
+      "loss": 0.1514,
+      "step": 6742
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.802254831782391e-05,
+      "loss": 0.2759,
+      "step": 6743
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.802075876879026e-05,
+      "loss": 0.2055,
+      "step": 6744
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8018969219756625e-05,
+      "loss": 0.2822,
+      "step": 6745
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.801717967072298e-05,
+      "loss": 0.2349,
+      "step": 6746
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8015390121689335e-05,
+      "loss": 0.2614,
+      "step": 6747
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.80136005726557e-05,
+      "loss": 0.1879,
+      "step": 6748
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.801181102362205e-05,
+      "loss": 0.256,
+      "step": 6749
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.80100214745884e-05,
+      "loss": 0.0988,
+      "step": 6750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.800823192555476e-05,
+      "loss": 0.2989,
+      "step": 6751
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.800644237652112e-05,
+      "loss": 0.2686,
+      "step": 6752
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.800465282748747e-05,
+      "loss": 0.2821,
+      "step": 6753
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8002863278453835e-05,
+      "loss": 0.1288,
+      "step": 6754
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.800107372942019e-05,
+      "loss": 0.2865,
+      "step": 6755
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7999284180386546e-05,
+      "loss": 0.1916,
+      "step": 6756
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.79974946313529e-05,
+      "loss": 0.2968,
+      "step": 6757
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7995705082319256e-05,
+      "loss": 0.1899,
+      "step": 6758
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.799391553328561e-05,
+      "loss": 0.2077,
+      "step": 6759
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7992125984251966e-05,
+      "loss": 0.2952,
+      "step": 6760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.799033643521833e-05,
+      "loss": 0.1554,
+      "step": 6761
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7988546886184683e-05,
+      "loss": 0.2505,
+      "step": 6762
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.798675733715104e-05,
+      "loss": 0.2466,
+      "step": 6763
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.79849677881174e-05,
+      "loss": 0.1499,
+      "step": 6764
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7983178239083756e-05,
+      "loss": 0.2996,
+      "step": 6765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7981388690050104e-05,
+      "loss": 0.2561,
+      "step": 6766
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7979599141016466e-05,
+      "loss": 0.1608,
+      "step": 6767
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.797780959198282e-05,
+      "loss": 0.2355,
+      "step": 6768
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7976020042949177e-05,
+      "loss": 0.3067,
+      "step": 6769
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.797423049391554e-05,
+      "loss": 0.1893,
+      "step": 6770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7972440944881894e-05,
+      "loss": 0.1667,
+      "step": 6771
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.797065139584825e-05,
+      "loss": 0.352,
+      "step": 6772
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7968861846814604e-05,
+      "loss": 0.3647,
+      "step": 6773
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.796707229778096e-05,
+      "loss": 0.1983,
+      "step": 6774
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7965282748747315e-05,
+      "loss": 0.1235,
+      "step": 6775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7963493199713676e-05,
+      "loss": 0.1313,
+      "step": 6776
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.796170365068003e-05,
+      "loss": 0.1527,
+      "step": 6777
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.795991410164639e-05,
+      "loss": 0.1618,
+      "step": 6778
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.795812455261274e-05,
+      "loss": 0.3615,
+      "step": 6779
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7956335003579104e-05,
+      "loss": 0.2995,
+      "step": 6780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.795454545454545e-05,
+      "loss": 0.2609,
+      "step": 6781
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.795275590551181e-05,
+      "loss": 0.274,
+      "step": 6782
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.795096635647817e-05,
+      "loss": 0.2789,
+      "step": 6783
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7949176807444525e-05,
+      "loss": 0.1789,
+      "step": 6784
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.794738725841088e-05,
+      "loss": 0.3122,
+      "step": 6785
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.794559770937724e-05,
+      "loss": 0.4045,
+      "step": 6786
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.79438081603436e-05,
+      "loss": 0.2071,
+      "step": 6787
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.794201861130995e-05,
+      "loss": 0.3282,
+      "step": 6788
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.794022906227631e-05,
+      "loss": 0.2178,
+      "step": 6789
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.793843951324266e-05,
+      "loss": 0.1517,
+      "step": 6790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.793664996420902e-05,
+      "loss": 0.1974,
+      "step": 6791
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.793486041517538e-05,
+      "loss": 0.2366,
+      "step": 6792
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7933070866141735e-05,
+      "loss": 0.199,
+      "step": 6793
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.793128131710809e-05,
+      "loss": 0.194,
+      "step": 6794
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7929491768074445e-05,
+      "loss": 0.1699,
+      "step": 6795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.792770221904081e-05,
+      "loss": 0.171,
+      "step": 6796
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7925912670007156e-05,
+      "loss": 0.3402,
+      "step": 6797
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.792412312097352e-05,
+      "loss": 0.2836,
+      "step": 6798
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.792233357193987e-05,
+      "loss": 0.2567,
+      "step": 6799
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.792054402290623e-05,
+      "loss": 0.1592,
+      "step": 6800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7918754473872583e-05,
+      "loss": 0.1943,
+      "step": 6801
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7916964924838945e-05,
+      "loss": 0.1494,
+      "step": 6802
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.79151753758053e-05,
+      "loss": 0.2701,
+      "step": 6803
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7913385826771656e-05,
+      "loss": 0.2061,
+      "step": 6804
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.791159627773801e-05,
+      "loss": 0.1354,
+      "step": 6805
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7909806728704366e-05,
+      "loss": 0.2526,
+      "step": 6806
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.790801717967072e-05,
+      "loss": 0.2364,
+      "step": 6807
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.790622763063708e-05,
+      "loss": 0.1807,
+      "step": 6808
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.790443808160344e-05,
+      "loss": 0.1755,
+      "step": 6809
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7902648532569794e-05,
+      "loss": 0.334,
+      "step": 6810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7900858983536156e-05,
+      "loss": 0.2913,
+      "step": 6811
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.789906943450251e-05,
+      "loss": 0.1679,
+      "step": 6812
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.789727988546886e-05,
+      "loss": 0.1881,
+      "step": 6813
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.789549033643522e-05,
+      "loss": 0.2325,
+      "step": 6814
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7893700787401576e-05,
+      "loss": 0.2813,
+      "step": 6815
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.789191123836793e-05,
+      "loss": 0.3033,
+      "step": 6816
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7890121689334294e-05,
+      "loss": 0.4508,
+      "step": 6817
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.788833214030065e-05,
+      "loss": 0.254,
+      "step": 6818
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7886542591267004e-05,
+      "loss": 0.1799,
+      "step": 6819
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.788475304223336e-05,
+      "loss": 0.1016,
+      "step": 6820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7882963493199714e-05,
+      "loss": 0.3759,
+      "step": 6821
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.788117394416607e-05,
+      "loss": 0.2848,
+      "step": 6822
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7879384395132425e-05,
+      "loss": 0.1499,
+      "step": 6823
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.787759484609879e-05,
+      "loss": 0.1718,
+      "step": 6824
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.787580529706514e-05,
+      "loss": 0.217,
+      "step": 6825
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.78740157480315e-05,
+      "loss": 0.2299,
+      "step": 6826
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.787222619899786e-05,
+      "loss": 0.2672,
+      "step": 6827
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.787043664996421e-05,
+      "loss": 0.1433,
+      "step": 6828
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.786864710093056e-05,
+      "loss": 0.3608,
+      "step": 6829
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7866857551896925e-05,
+      "loss": 0.2261,
+      "step": 6830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.786506800286328e-05,
+      "loss": 0.2237,
+      "step": 6831
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7863278453829635e-05,
+      "loss": 0.1257,
+      "step": 6832
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7861488904796e-05,
+      "loss": 0.1443,
+      "step": 6833
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.785969935576235e-05,
+      "loss": 0.1301,
+      "step": 6834
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.785790980672871e-05,
+      "loss": 0.1893,
+      "step": 6835
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.785612025769506e-05,
+      "loss": 0.2465,
+      "step": 6836
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.785433070866142e-05,
+      "loss": 0.2307,
+      "step": 6837
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.785254115962777e-05,
+      "loss": 0.1572,
+      "step": 6838
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7850751610594135e-05,
+      "loss": 0.2026,
+      "step": 6839
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.784896206156049e-05,
+      "loss": 0.1959,
+      "step": 6840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7847172512526845e-05,
+      "loss": 0.1367,
+      "step": 6841
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.78453829634932e-05,
+      "loss": 0.1662,
+      "step": 6842
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.784359341445956e-05,
+      "loss": 0.1829,
+      "step": 6843
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.784180386542591e-05,
+      "loss": 0.1974,
+      "step": 6844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7840014316392266e-05,
+      "loss": 0.308,
+      "step": 6845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.783822476735863e-05,
+      "loss": 0.2299,
+      "step": 6846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.783643521832498e-05,
+      "loss": 0.3495,
+      "step": 6847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.783464566929134e-05,
+      "loss": 0.1164,
+      "step": 6848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.78328561202577e-05,
+      "loss": 0.2375,
+      "step": 6849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7831066571224056e-05,
+      "loss": 0.2376,
+      "step": 6850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.782927702219041e-05,
+      "loss": 0.2264,
+      "step": 6851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7827487473156766e-05,
+      "loss": 0.1388,
+      "step": 6852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.782569792412312e-05,
+      "loss": 0.2003,
+      "step": 6853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7823908375089476e-05,
+      "loss": 0.1056,
+      "step": 6854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.782211882605584e-05,
+      "loss": 0.2739,
+      "step": 6855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7820329277022194e-05,
+      "loss": 0.175,
+      "step": 6856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.781853972798855e-05,
+      "loss": 0.3826,
+      "step": 6857
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7816750178954904e-05,
+      "loss": 0.1187,
+      "step": 6858
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7814960629921266e-05,
+      "loss": 0.1545,
+      "step": 6859
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7813171080887614e-05,
+      "loss": 0.1092,
+      "step": 6860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7811381531853976e-05,
+      "loss": 0.1553,
+      "step": 6861
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780959198282033e-05,
+      "loss": 0.1343,
+      "step": 6862
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780780243378669e-05,
+      "loss": 0.2701,
+      "step": 6863
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780601288475304e-05,
+      "loss": 0.2108,
+      "step": 6864
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7804223335719404e-05,
+      "loss": 0.2322,
+      "step": 6865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780243378668576e-05,
+      "loss": 0.2115,
+      "step": 6866
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780064423765211e-05,
+      "loss": 0.1853,
+      "step": 6867
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779885468861847e-05,
+      "loss": 0.2113,
+      "step": 6868
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7797065139584825e-05,
+      "loss": 0.1877,
+      "step": 6869
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779527559055118e-05,
+      "loss": 0.2669,
+      "step": 6870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779348604151754e-05,
+      "loss": 0.3289,
+      "step": 6871
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.77916964924839e-05,
+      "loss": 0.1905,
+      "step": 6872
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.778990694345025e-05,
+      "loss": 0.488,
+      "step": 6873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7788117394416614e-05,
+      "loss": 0.2649,
+      "step": 6874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.778632784538296e-05,
+      "loss": 0.3781,
+      "step": 6875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.778453829634932e-05,
+      "loss": 0.2167,
+      "step": 6876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.778274874731568e-05,
+      "loss": 0.2404,
+      "step": 6877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7780959198282035e-05,
+      "loss": 0.2323,
+      "step": 6878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777916964924839e-05,
+      "loss": 0.2805,
+      "step": 6879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777738010021475e-05,
+      "loss": 0.3112,
+      "step": 6880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777559055118111e-05,
+      "loss": 0.3276,
+      "step": 6881
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777380100214746e-05,
+      "loss": 0.2392,
+      "step": 6882
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777201145311382e-05,
+      "loss": 0.2748,
+      "step": 6883
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.777022190408017e-05,
+      "loss": 0.3046,
+      "step": 6884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.776843235504653e-05,
+      "loss": 0.3151,
+      "step": 6885
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.776664280601288e-05,
+      "loss": 0.2237,
+      "step": 6886
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7764853256979245e-05,
+      "loss": 0.193,
+      "step": 6887
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.77630637079456e-05,
+      "loss": 0.2439,
+      "step": 6888
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7761274158911956e-05,
+      "loss": 0.2994,
+      "step": 6889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.775948460987832e-05,
+      "loss": 0.244,
+      "step": 6890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7757695060844666e-05,
+      "loss": 0.231,
+      "step": 6891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.775590551181102e-05,
+      "loss": 0.2604,
+      "step": 6892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.775411596277738e-05,
+      "loss": 0.2209,
+      "step": 6893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.775232641374374e-05,
+      "loss": 0.118,
+      "step": 6894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7750536864710094e-05,
+      "loss": 0.2458,
+      "step": 6895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7748747315676455e-05,
+      "loss": 0.3556,
+      "step": 6896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.774695776664281e-05,
+      "loss": 0.2363,
+      "step": 6897
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7745168217609166e-05,
+      "loss": 0.1664,
+      "step": 6898
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.774337866857552e-05,
+      "loss": 0.3288,
+      "step": 6899
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7741589119541876e-05,
+      "loss": 0.2293,
+      "step": 6900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.773979957050823e-05,
+      "loss": 0.2187,
+      "step": 6901
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7738010021474593e-05,
+      "loss": 0.2269,
+      "step": 6902
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.773622047244095e-05,
+      "loss": 0.1324,
+      "step": 6903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7734430923407304e-05,
+      "loss": 0.1967,
+      "step": 6904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.773264137437366e-05,
+      "loss": 0.4363,
+      "step": 6905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7730851825340014e-05,
+      "loss": 0.3113,
+      "step": 6906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.772906227630637e-05,
+      "loss": 0.1877,
+      "step": 6907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7727272727272725e-05,
+      "loss": 0.1249,
+      "step": 6908
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7725483178239087e-05,
+      "loss": 0.2272,
+      "step": 6909
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.772369362920544e-05,
+      "loss": 0.2337,
+      "step": 6910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.77219040801718e-05,
+      "loss": 0.154,
+      "step": 6911
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.772011453113816e-05,
+      "loss": 0.2275,
+      "step": 6912
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7718324982104514e-05,
+      "loss": 0.1691,
+      "step": 6913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.771653543307086e-05,
+      "loss": 0.2296,
+      "step": 6914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7714745884037224e-05,
+      "loss": 0.2756,
+      "step": 6915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.771295633500358e-05,
+      "loss": 0.1148,
+      "step": 6916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7711166785969935e-05,
+      "loss": 0.2965,
+      "step": 6917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.77093772369363e-05,
+      "loss": 0.1659,
+      "step": 6918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770758768790265e-05,
+      "loss": 0.2101,
+      "step": 6919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770579813886901e-05,
+      "loss": 0.1553,
+      "step": 6920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770400858983536e-05,
+      "loss": 0.0935,
+      "step": 6921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770221904080172e-05,
+      "loss": 0.1159,
+      "step": 6922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770042949176807e-05,
+      "loss": 0.1475,
+      "step": 6923
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7698639942734435e-05,
+      "loss": 0.2091,
+      "step": 6924
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.769685039370079e-05,
+      "loss": 0.1455,
+      "step": 6925
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7695060844667145e-05,
+      "loss": 0.238,
+      "step": 6926
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.76932712956335e-05,
+      "loss": 0.3522,
+      "step": 6927
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.769148174659986e-05,
+      "loss": 0.1378,
+      "step": 6928
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.768969219756622e-05,
+      "loss": 0.3222,
+      "step": 6929
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7687902648532566e-05,
+      "loss": 0.1593,
+      "step": 6930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.768611309949893e-05,
+      "loss": 0.2981,
+      "step": 6931
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.768432355046528e-05,
+      "loss": 0.1133,
+      "step": 6932
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.768253400143164e-05,
+      "loss": 0.164,
+      "step": 6933
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7680744452398e-05,
+      "loss": 0.1777,
+      "step": 6934
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7678954903364355e-05,
+      "loss": 0.0656,
+      "step": 6935
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.767716535433071e-05,
+      "loss": 0.1472,
+      "step": 6936
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.767537580529707e-05,
+      "loss": 0.2516,
+      "step": 6937
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.767358625626342e-05,
+      "loss": 0.1727,
+      "step": 6938
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7671796707229776e-05,
+      "loss": 0.2625,
+      "step": 6939
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.767000715819614e-05,
+      "loss": 0.2457,
+      "step": 6940
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.766821760916249e-05,
+      "loss": 0.193,
+      "step": 6941
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.766642806012885e-05,
+      "loss": 0.0945,
+      "step": 6942
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7664638511095204e-05,
+      "loss": 0.1009,
+      "step": 6943
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7662848962061566e-05,
+      "loss": 0.2454,
+      "step": 6944
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.766105941302792e-05,
+      "loss": 0.1925,
+      "step": 6945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7659269863994276e-05,
+      "loss": 0.1467,
+      "step": 6946
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.765748031496063e-05,
+      "loss": 0.358,
+      "step": 6947
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7655690765926987e-05,
+      "loss": 0.4048,
+      "step": 6948
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.765390121689334e-05,
+      "loss": 0.182,
+      "step": 6949
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7652111667859704e-05,
+      "loss": 0.1756,
+      "step": 6950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.765032211882606e-05,
+      "loss": 0.1741,
+      "step": 6951
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7648532569792414e-05,
+      "loss": 0.1871,
+      "step": 6952
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.764674302075877e-05,
+      "loss": 0.2005,
+      "step": 6953
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7644953471725124e-05,
+      "loss": 0.2058,
+      "step": 6954
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.764316392269148e-05,
+      "loss": 0.1581,
+      "step": 6955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.764137437365784e-05,
+      "loss": 0.1974,
+      "step": 6956
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.76395848246242e-05,
+      "loss": 0.2209,
+      "step": 6957
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.763779527559055e-05,
+      "loss": 0.0981,
+      "step": 6958
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7636005726556914e-05,
+      "loss": 0.3022,
+      "step": 6959
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.763421617752327e-05,
+      "loss": 0.1835,
+      "step": 6960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.763242662848962e-05,
+      "loss": 0.2277,
+      "step": 6961
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.763063707945598e-05,
+      "loss": 0.2809,
+      "step": 6962
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7628847530422335e-05,
+      "loss": 0.1947,
+      "step": 6963
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.762705798138869e-05,
+      "loss": 0.1647,
+      "step": 6964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.762526843235505e-05,
+      "loss": 0.1971,
+      "step": 6965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.762347888332141e-05,
+      "loss": 0.1812,
+      "step": 6966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.762168933428776e-05,
+      "loss": 0.1945,
+      "step": 6967
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.761989978525412e-05,
+      "loss": 0.1442,
+      "step": 6968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.761811023622047e-05,
+      "loss": 0.2997,
+      "step": 6969
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.761632068718683e-05,
+      "loss": 0.2539,
+      "step": 6970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.761453113815318e-05,
+      "loss": 0.1861,
+      "step": 6971
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7612741589119545e-05,
+      "loss": 0.1135,
+      "step": 6972
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.76109520400859e-05,
+      "loss": 0.2648,
+      "step": 6973
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7609162491052255e-05,
+      "loss": 0.1758,
+      "step": 6974
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.760737294201862e-05,
+      "loss": 0.1924,
+      "step": 6975
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.760558339298497e-05,
+      "loss": 0.119,
+      "step": 6976
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.760379384395132e-05,
+      "loss": 0.1794,
+      "step": 6977
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.760200429491768e-05,
+      "loss": 0.1542,
+      "step": 6978
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.760021474588404e-05,
+      "loss": 0.1939,
+      "step": 6979
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.759842519685039e-05,
+      "loss": 0.2819,
+      "step": 6980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7596635647816755e-05,
+      "loss": 0.2211,
+      "step": 6981
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.759484609878311e-05,
+      "loss": 0.2362,
+      "step": 6982
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7593056549749466e-05,
+      "loss": 0.256,
+      "step": 6983
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.759126700071582e-05,
+      "loss": 0.168,
+      "step": 6984
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7589477451682176e-05,
+      "loss": 0.2356,
+      "step": 6985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.758768790264853e-05,
+      "loss": 0.1637,
+      "step": 6986
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.758589835361489e-05,
+      "loss": 0.3089,
+      "step": 6987
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.758410880458125e-05,
+      "loss": 0.1501,
+      "step": 6988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7582319255547604e-05,
+      "loss": 0.1998,
+      "step": 6989
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.758052970651396e-05,
+      "loss": 0.3115,
+      "step": 6990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.757874015748032e-05,
+      "loss": 0.2315,
+      "step": 6991
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.757695060844667e-05,
+      "loss": 0.183,
+      "step": 6992
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7575161059413024e-05,
+      "loss": 0.2131,
+      "step": 6993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7573371510379386e-05,
+      "loss": 0.1424,
+      "step": 6994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.757158196134574e-05,
+      "loss": 0.2603,
+      "step": 6995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.75697924123121e-05,
+      "loss": 0.1229,
+      "step": 6996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.756800286327846e-05,
+      "loss": 0.1991,
+      "step": 6997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7566213314244814e-05,
+      "loss": 0.1187,
+      "step": 6998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.756442376521117e-05,
+      "loss": 0.1937,
+      "step": 6999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7562634216177524e-05,
+      "loss": 0.1216,
+      "step": 7000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.756084466714388e-05,
+      "loss": 0.2371,
+      "step": 7001
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7559055118110235e-05,
+      "loss": 0.1778,
+      "step": 7002
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.75572655690766e-05,
+      "loss": 0.1506,
+      "step": 7003
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.755547602004295e-05,
+      "loss": 0.1443,
+      "step": 7004
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.755368647100931e-05,
+      "loss": 0.211,
+      "step": 7005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.755189692197566e-05,
+      "loss": 0.1637,
+      "step": 7006
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7550107372942024e-05,
+      "loss": 0.1456,
+      "step": 7007
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.754831782390837e-05,
+      "loss": 0.131,
+      "step": 7008
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7546528274874735e-05,
+      "loss": 0.418,
+      "step": 7009
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.754473872584109e-05,
+      "loss": 0.1131,
+      "step": 7010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7542949176807445e-05,
+      "loss": 0.2161,
+      "step": 7011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.75411596277738e-05,
+      "loss": 0.3953,
+      "step": 7012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753937007874016e-05,
+      "loss": 0.0902,
+      "step": 7013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753758052970652e-05,
+      "loss": 0.1175,
+      "step": 7014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753579098067287e-05,
+      "loss": 0.2931,
+      "step": 7015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753400143163923e-05,
+      "loss": 0.1359,
+      "step": 7016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753221188260558e-05,
+      "loss": 0.2571,
+      "step": 7017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.753042233357194e-05,
+      "loss": 0.2062,
+      "step": 7018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.75286327845383e-05,
+      "loss": 0.2018,
+      "step": 7019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7526843235504655e-05,
+      "loss": 0.3497,
+      "step": 7020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.752505368647101e-05,
+      "loss": 0.201,
+      "step": 7021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.752326413743737e-05,
+      "loss": 0.1519,
+      "step": 7022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.752147458840373e-05,
+      "loss": 0.2525,
+      "step": 7023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7519685039370076e-05,
+      "loss": 0.3289,
+      "step": 7024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.751789549033644e-05,
+      "loss": 0.2116,
+      "step": 7025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.751610594130279e-05,
+      "loss": 0.218,
+      "step": 7026
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.751431639226915e-05,
+      "loss": 0.2922,
+      "step": 7027
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.751252684323551e-05,
+      "loss": 0.1536,
+      "step": 7028
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7510737294201866e-05,
+      "loss": 0.1941,
+      "step": 7029
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.750894774516822e-05,
+      "loss": 0.2537,
+      "step": 7030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7507158196134576e-05,
+      "loss": 0.2151,
+      "step": 7031
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.750536864710093e-05,
+      "loss": 0.1746,
+      "step": 7032
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7503579098067286e-05,
+      "loss": 0.2404,
+      "step": 7033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.750178954903364e-05,
+      "loss": 0.1982,
+      "step": 7034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2919,
+      "step": 7035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.749821045096636e-05,
+      "loss": 0.1172,
+      "step": 7036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7496420901932714e-05,
+      "loss": 0.1771,
+      "step": 7037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7494631352899076e-05,
+      "loss": 0.2124,
+      "step": 7038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7492841803865424e-05,
+      "loss": 0.2017,
+      "step": 7039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.749105225483178e-05,
+      "loss": 0.1943,
+      "step": 7040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.748926270579814e-05,
+      "loss": 0.2854,
+      "step": 7041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.74874731567645e-05,
+      "loss": 0.2627,
+      "step": 7042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.748568360773085e-05,
+      "loss": 0.2078,
+      "step": 7043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7483894058697214e-05,
+      "loss": 0.2773,
+      "step": 7044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.748210450966357e-05,
+      "loss": 0.2318,
+      "step": 7045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7480314960629924e-05,
+      "loss": 0.16,
+      "step": 7046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.747852541159628e-05,
+      "loss": 0.1017,
+      "step": 7047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7476735862562635e-05,
+      "loss": 0.195,
+      "step": 7048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.747494631352899e-05,
+      "loss": 0.1962,
+      "step": 7049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.747315676449535e-05,
+      "loss": 0.2126,
+      "step": 7050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.747136721546171e-05,
+      "loss": 0.2197,
+      "step": 7051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.746957766642806e-05,
+      "loss": 0.1654,
+      "step": 7052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.746778811739442e-05,
+      "loss": 0.1886,
+      "step": 7053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.746599856836078e-05,
+      "loss": 0.0955,
+      "step": 7054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.746420901932713e-05,
+      "loss": 0.1719,
+      "step": 7055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.746241947029348e-05,
+      "loss": 0.1896,
+      "step": 7056
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7460629921259845e-05,
+      "loss": 0.196,
+      "step": 7057
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.74588403722262e-05,
+      "loss": 0.157,
+      "step": 7058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7457050823192555e-05,
+      "loss": 0.163,
+      "step": 7059
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.745526127415892e-05,
+      "loss": 0.2259,
+      "step": 7060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.745347172512527e-05,
+      "loss": 0.1827,
+      "step": 7061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.745168217609163e-05,
+      "loss": 0.2574,
+      "step": 7062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.744989262705798e-05,
+      "loss": 0.1909,
+      "step": 7063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.744810307802434e-05,
+      "loss": 0.1487,
+      "step": 7064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.744631352899069e-05,
+      "loss": 0.3058,
+      "step": 7065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7444523979957055e-05,
+      "loss": 0.1603,
+      "step": 7066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.744273443092341e-05,
+      "loss": 0.179,
+      "step": 7067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7440944881889766e-05,
+      "loss": 0.2679,
+      "step": 7068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.743915533285612e-05,
+      "loss": 0.3031,
+      "step": 7069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.743736578382248e-05,
+      "loss": 0.124,
+      "step": 7070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.743557623478883e-05,
+      "loss": 0.0987,
+      "step": 7071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.743378668575519e-05,
+      "loss": 0.2791,
+      "step": 7072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.743199713672155e-05,
+      "loss": 0.2918,
+      "step": 7073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7430207587687903e-05,
+      "loss": 0.2638,
+      "step": 7074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.742841803865426e-05,
+      "loss": 0.2127,
+      "step": 7075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.742662848962062e-05,
+      "loss": 0.1557,
+      "step": 7076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7424838940586976e-05,
+      "loss": 0.2528,
+      "step": 7077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.742304939155333e-05,
+      "loss": 0.2996,
+      "step": 7078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7421259842519686e-05,
+      "loss": 0.1771,
+      "step": 7079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.741947029348604e-05,
+      "loss": 0.142,
+      "step": 7080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7417680744452397e-05,
+      "loss": 0.1679,
+      "step": 7081
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.741589119541876e-05,
+      "loss": 0.1552,
+      "step": 7082
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7414101646385114e-05,
+      "loss": 0.1704,
+      "step": 7083
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.741231209735147e-05,
+      "loss": 0.1206,
+      "step": 7084
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.741052254831783e-05,
+      "loss": 0.1777,
+      "step": 7085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.740873299928418e-05,
+      "loss": 0.3523,
+      "step": 7086
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7406943450250535e-05,
+      "loss": 0.1151,
+      "step": 7087
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7405153901216896e-05,
+      "loss": 0.2173,
+      "step": 7088
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.740336435218325e-05,
+      "loss": 0.1878,
+      "step": 7089
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.740157480314961e-05,
+      "loss": 0.2172,
+      "step": 7090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.739978525411597e-05,
+      "loss": 0.0769,
+      "step": 7091
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7397995705082324e-05,
+      "loss": 0.1049,
+      "step": 7092
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.739620615604868e-05,
+      "loss": 0.164,
+      "step": 7093
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7394416607015034e-05,
+      "loss": 0.1384,
+      "step": 7094
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.739262705798139e-05,
+      "loss": 0.2267,
+      "step": 7095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7390837508947745e-05,
+      "loss": 0.2467,
+      "step": 7096
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.73890479599141e-05,
+      "loss": 0.2074,
+      "step": 7097
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.738725841088046e-05,
+      "loss": 0.1684,
+      "step": 7098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.738546886184682e-05,
+      "loss": 0.1687,
+      "step": 7099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.738367931281317e-05,
+      "loss": 0.194,
+      "step": 7100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7381889763779534e-05,
+      "loss": 0.1752,
+      "step": 7101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.738010021474588e-05,
+      "loss": 0.4389,
+      "step": 7102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.737831066571224e-05,
+      "loss": 0.1616,
+      "step": 7103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.73765211166786e-05,
+      "loss": 0.1755,
+      "step": 7104
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7374731567644955e-05,
+      "loss": 0.1453,
+      "step": 7105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.737294201861131e-05,
+      "loss": 0.2256,
+      "step": 7106
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.737115246957767e-05,
+      "loss": 0.1868,
+      "step": 7107
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736936292054403e-05,
+      "loss": 0.1418,
+      "step": 7108
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736757337151038e-05,
+      "loss": 0.1805,
+      "step": 7109
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736578382247674e-05,
+      "loss": 0.2006,
+      "step": 7110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736399427344309e-05,
+      "loss": 0.1942,
+      "step": 7111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736220472440945e-05,
+      "loss": 0.1309,
+      "step": 7112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.736041517537581e-05,
+      "loss": 0.1956,
+      "step": 7113
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7358625626342165e-05,
+      "loss": 0.2182,
+      "step": 7114
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.735683607730852e-05,
+      "loss": 0.2189,
+      "step": 7115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7355046528274876e-05,
+      "loss": 0.1673,
+      "step": 7116
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.735325697924124e-05,
+      "loss": 0.1204,
+      "step": 7117
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7351467430207586e-05,
+      "loss": 0.1448,
+      "step": 7118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.734967788117394e-05,
+      "loss": 0.2701,
+      "step": 7119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.73478883321403e-05,
+      "loss": 0.1575,
+      "step": 7120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.734609878310666e-05,
+      "loss": 0.2025,
+      "step": 7121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7344309234073014e-05,
+      "loss": 0.2375,
+      "step": 7122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7342519685039376e-05,
+      "loss": 0.1523,
+      "step": 7123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.734073013600573e-05,
+      "loss": 0.1801,
+      "step": 7124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.733894058697208e-05,
+      "loss": 0.1874,
+      "step": 7125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.733715103793844e-05,
+      "loss": 0.1399,
+      "step": 7126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7335361488904796e-05,
+      "loss": 0.2602,
+      "step": 7127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.733357193987115e-05,
+      "loss": 0.1872,
+      "step": 7128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7331782390837514e-05,
+      "loss": 0.2039,
+      "step": 7129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.732999284180387e-05,
+      "loss": 0.1049,
+      "step": 7130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7328203292770224e-05,
+      "loss": 0.2297,
+      "step": 7131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.732641374373658e-05,
+      "loss": 0.2308,
+      "step": 7132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7324624194702934e-05,
+      "loss": 0.2859,
+      "step": 7133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.732283464566929e-05,
+      "loss": 0.2892,
+      "step": 7134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.732104509663565e-05,
+      "loss": 0.1278,
+      "step": 7135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.731925554760201e-05,
+      "loss": 0.2875,
+      "step": 7136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.731746599856836e-05,
+      "loss": 0.2604,
+      "step": 7137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.731567644953472e-05,
+      "loss": 0.1736,
+      "step": 7138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.731388690050108e-05,
+      "loss": 0.1487,
+      "step": 7139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7312097351467434e-05,
+      "loss": 0.2541,
+      "step": 7140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.731030780243378e-05,
+      "loss": 0.2625,
+      "step": 7141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7308518253400145e-05,
+      "loss": 0.0905,
+      "step": 7142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.73067287043665e-05,
+      "loss": 0.1405,
+      "step": 7143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7304939155332855e-05,
+      "loss": 0.3243,
+      "step": 7144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.730314960629922e-05,
+      "loss": 0.3154,
+      "step": 7145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.730136005726557e-05,
+      "loss": 0.124,
+      "step": 7146
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729957050823193e-05,
+      "loss": 0.127,
+      "step": 7147
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729778095919829e-05,
+      "loss": 0.3554,
+      "step": 7148
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729599141016464e-05,
+      "loss": 0.2379,
+      "step": 7149
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729420186113099e-05,
+      "loss": 0.1842,
+      "step": 7150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7292412312097355e-05,
+      "loss": 0.2241,
+      "step": 7151
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729062276306371e-05,
+      "loss": 0.283,
+      "step": 7152
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7288833214030065e-05,
+      "loss": 0.1743,
+      "step": 7153
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728704366499642e-05,
+      "loss": 0.22,
+      "step": 7154
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728525411596278e-05,
+      "loss": 0.1971,
+      "step": 7155
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728346456692914e-05,
+      "loss": 0.2125,
+      "step": 7156
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728167501789549e-05,
+      "loss": 0.2965,
+      "step": 7157
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.727988546886185e-05,
+      "loss": 0.137,
+      "step": 7158
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.72780959198282e-05,
+      "loss": 0.1008,
+      "step": 7159
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.727630637079456e-05,
+      "loss": 0.0833,
+      "step": 7160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.727451682176092e-05,
+      "loss": 0.12,
+      "step": 7161
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7272727272727276e-05,
+      "loss": 0.1567,
+      "step": 7162
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.727093772369363e-05,
+      "loss": 0.1257,
+      "step": 7163
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.726914817465999e-05,
+      "loss": 0.2827,
+      "step": 7164
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.726735862562634e-05,
+      "loss": 0.2965,
+      "step": 7165
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7265569076592696e-05,
+      "loss": 0.269,
+      "step": 7166
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.726377952755906e-05,
+      "loss": 0.2287,
+      "step": 7167
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7261989978525414e-05,
+      "loss": 0.1067,
+      "step": 7168
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.726020042949177e-05,
+      "loss": 0.112,
+      "step": 7169
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.725841088045813e-05,
+      "loss": 0.1627,
+      "step": 7170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7256621331424486e-05,
+      "loss": 0.2896,
+      "step": 7171
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7254831782390834e-05,
+      "loss": 0.1969,
+      "step": 7172
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7253042233357196e-05,
+      "loss": 0.1728,
+      "step": 7173
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.725125268432355e-05,
+      "loss": 0.1946,
+      "step": 7174
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.724946313528991e-05,
+      "loss": 0.1499,
+      "step": 7175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.724767358625627e-05,
+      "loss": 0.1767,
+      "step": 7176
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7245884037222624e-05,
+      "loss": 0.2308,
+      "step": 7177
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.724409448818898e-05,
+      "loss": 0.1392,
+      "step": 7178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7242304939155334e-05,
+      "loss": 0.2575,
+      "step": 7179
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.724051539012169e-05,
+      "loss": 0.2084,
+      "step": 7180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7238725841088045e-05,
+      "loss": 0.1224,
+      "step": 7181
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.72369362920544e-05,
+      "loss": 0.16,
+      "step": 7182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.723514674302076e-05,
+      "loss": 0.2037,
+      "step": 7183
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.723335719398712e-05,
+      "loss": 0.1512,
+      "step": 7184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.723156764495347e-05,
+      "loss": 0.0772,
+      "step": 7185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7229778095919834e-05,
+      "loss": 0.1419,
+      "step": 7186
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.722798854688619e-05,
+      "loss": 0.3199,
+      "step": 7187
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.722619899785254e-05,
+      "loss": 0.1977,
+      "step": 7188
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.72244094488189e-05,
+      "loss": 0.2459,
+      "step": 7189
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7222619899785255e-05,
+      "loss": 0.3751,
+      "step": 7190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.722083035075161e-05,
+      "loss": 0.3577,
+      "step": 7191
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.721904080171797e-05,
+      "loss": 0.1925,
+      "step": 7192
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.721725125268433e-05,
+      "loss": 0.1542,
+      "step": 7193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721546170365068e-05,
+      "loss": 0.2573,
+      "step": 7194
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721367215461704e-05,
+      "loss": 0.2347,
+      "step": 7195
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721188260558339e-05,
+      "loss": 0.1498,
+      "step": 7196
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721009305654975e-05,
+      "loss": 0.1452,
+      "step": 7197
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720830350751611e-05,
+      "loss": 0.3005,
+      "step": 7198
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7206513958482465e-05,
+      "loss": 0.1165,
+      "step": 7199
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720472440944882e-05,
+      "loss": 0.1809,
+      "step": 7200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7202934860415176e-05,
+      "loss": 0.253,
+      "step": 7201
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720114531138154e-05,
+      "loss": 0.1626,
+      "step": 7202
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.719935576234789e-05,
+      "loss": 0.1338,
+      "step": 7203
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.719756621331424e-05,
+      "loss": 0.2053,
+      "step": 7204
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.71957766642806e-05,
+      "loss": 0.2814,
+      "step": 7205
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.719398711524696e-05,
+      "loss": 0.1563,
+      "step": 7206
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7192197566213314e-05,
+      "loss": 0.3109,
+      "step": 7207
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7190408017179675e-05,
+      "loss": 0.2256,
+      "step": 7208
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.718861846814603e-05,
+      "loss": 0.1754,
+      "step": 7209
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7186828919112386e-05,
+      "loss": 0.1783,
+      "step": 7210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.718503937007874e-05,
+      "loss": 0.1387,
+      "step": 7211
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7183249821045096e-05,
+      "loss": 0.136,
+      "step": 7212
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.718146027201145e-05,
+      "loss": 0.1491,
+      "step": 7213
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7179670722977813e-05,
+      "loss": 0.2064,
+      "step": 7214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.717788117394417e-05,
+      "loss": 0.2324,
+      "step": 7215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7176091624910524e-05,
+      "loss": 0.1993,
+      "step": 7216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.717430207587688e-05,
+      "loss": 0.2067,
+      "step": 7217
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.717251252684324e-05,
+      "loss": 0.2777,
+      "step": 7218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.717072297780959e-05,
+      "loss": 0.2336,
+      "step": 7219
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.716893342877595e-05,
+      "loss": 0.1614,
+      "step": 7220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7167143879742307e-05,
+      "loss": 0.2014,
+      "step": 7221
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.716535433070866e-05,
+      "loss": 0.2463,
+      "step": 7222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.716356478167502e-05,
+      "loss": 0.0831,
+      "step": 7223
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.716177523264138e-05,
+      "loss": 0.1796,
+      "step": 7224
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7159985683607734e-05,
+      "loss": 0.1359,
+      "step": 7225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.715819613457409e-05,
+      "loss": 0.2379,
+      "step": 7226
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7156406585540444e-05,
+      "loss": 0.217,
+      "step": 7227
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.71546170365068e-05,
+      "loss": 0.1043,
+      "step": 7228
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7152827487473155e-05,
+      "loss": 0.3264,
+      "step": 7229
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.715103793843952e-05,
+      "loss": 0.2045,
+      "step": 7230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.714924838940587e-05,
+      "loss": 0.1065,
+      "step": 7231
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.714745884037223e-05,
+      "loss": 0.2179,
+      "step": 7232
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.714566929133859e-05,
+      "loss": 0.1479,
+      "step": 7233
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7143879742304944e-05,
+      "loss": 0.1371,
+      "step": 7234
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.714209019327129e-05,
+      "loss": 0.1551,
+      "step": 7235
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7140300644237655e-05,
+      "loss": 0.2297,
+      "step": 7236
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.713851109520401e-05,
+      "loss": 0.1966,
+      "step": 7237
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7136721546170365e-05,
+      "loss": 0.2197,
+      "step": 7238
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.713493199713673e-05,
+      "loss": 0.2527,
+      "step": 7239
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.713314244810308e-05,
+      "loss": 0.2821,
+      "step": 7240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.713135289906944e-05,
+      "loss": 0.1771,
+      "step": 7241
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.712956335003579e-05,
+      "loss": 0.1014,
+      "step": 7242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.712777380100215e-05,
+      "loss": 0.2043,
+      "step": 7243
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.71259842519685e-05,
+      "loss": 0.1977,
+      "step": 7244
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.712419470293486e-05,
+      "loss": 0.2438,
+      "step": 7245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.712240515390122e-05,
+      "loss": 0.1198,
+      "step": 7246
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7120615604867575e-05,
+      "loss": 0.1044,
+      "step": 7247
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.711882605583393e-05,
+      "loss": 0.3885,
+      "step": 7248
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.711703650680029e-05,
+      "loss": 0.1058,
+      "step": 7249
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.711524695776665e-05,
+      "loss": 0.1901,
+      "step": 7250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7113457408732996e-05,
+      "loss": 0.1752,
+      "step": 7251
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.711166785969936e-05,
+      "loss": 0.2015,
+      "step": 7252
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710987831066571e-05,
+      "loss": 0.1257,
+      "step": 7253
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710808876163207e-05,
+      "loss": 0.2417,
+      "step": 7254
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710629921259843e-05,
+      "loss": 0.2047,
+      "step": 7255
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7104509663564786e-05,
+      "loss": 0.1353,
+      "step": 7256
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710272011453114e-05,
+      "loss": 0.254,
+      "step": 7257
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7100930565497496e-05,
+      "loss": 0.3451,
+      "step": 7258
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.709914101646385e-05,
+      "loss": 0.2135,
+      "step": 7259
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7097351467430207e-05,
+      "loss": 0.2484,
+      "step": 7260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.709556191839657e-05,
+      "loss": 0.2907,
+      "step": 7261
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7093772369362924e-05,
+      "loss": 0.1825,
+      "step": 7262
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.709198282032928e-05,
+      "loss": 0.1497,
+      "step": 7263
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7090193271295634e-05,
+      "loss": 0.201,
+      "step": 7264
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7088403722261996e-05,
+      "loss": 0.1528,
+      "step": 7265
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7086614173228344e-05,
+      "loss": 0.2646,
+      "step": 7266
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.70848246241947e-05,
+      "loss": 0.1597,
+      "step": 7267
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.708303507516106e-05,
+      "loss": 0.2735,
+      "step": 7268
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.708124552612742e-05,
+      "loss": 0.1769,
+      "step": 7269
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.707945597709377e-05,
+      "loss": 0.0966,
+      "step": 7270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7077666428060134e-05,
+      "loss": 0.1629,
+      "step": 7271
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.707587687902649e-05,
+      "loss": 0.2464,
+      "step": 7272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7074087329992844e-05,
+      "loss": 0.2245,
+      "step": 7273
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.70722977809592e-05,
+      "loss": 0.2935,
+      "step": 7274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7070508231925555e-05,
+      "loss": 0.2271,
+      "step": 7275
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.706871868289191e-05,
+      "loss": 0.1367,
+      "step": 7276
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.706692913385827e-05,
+      "loss": 0.2423,
+      "step": 7277
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.706513958482463e-05,
+      "loss": 0.1883,
+      "step": 7278
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.706335003579098e-05,
+      "loss": 0.0978,
+      "step": 7279
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.706156048675734e-05,
+      "loss": 0.2268,
+      "step": 7280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.70597709377237e-05,
+      "loss": 0.3167,
+      "step": 7281
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.705798138869005e-05,
+      "loss": 0.1731,
+      "step": 7282
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.705619183965641e-05,
+      "loss": 0.25,
+      "step": 7283
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7054402290622765e-05,
+      "loss": 0.0847,
+      "step": 7284
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.705261274158912e-05,
+      "loss": 0.1427,
+      "step": 7285
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7050823192555475e-05,
+      "loss": 0.33,
+      "step": 7286
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.704903364352184e-05,
+      "loss": 0.1033,
+      "step": 7287
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.704724409448819e-05,
+      "loss": 0.1177,
+      "step": 7288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.704545454545455e-05,
+      "loss": 0.1506,
+      "step": 7289
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.70436649964209e-05,
+      "loss": 0.1455,
+      "step": 7290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.704187544738726e-05,
+      "loss": 0.1687,
+      "step": 7291
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.704008589835361e-05,
+      "loss": 0.2056,
+      "step": 7292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7038296349319975e-05,
+      "loss": 0.2437,
+      "step": 7293
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.703650680028633e-05,
+      "loss": 0.0655,
+      "step": 7294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7034717251252686e-05,
+      "loss": 0.2746,
+      "step": 7295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.703292770221905e-05,
+      "loss": 0.1517,
+      "step": 7296
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.70311381531854e-05,
+      "loss": 0.1308,
+      "step": 7297
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.702934860415175e-05,
+      "loss": 0.3686,
+      "step": 7298
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.702755905511811e-05,
+      "loss": 0.2023,
+      "step": 7299
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.702576950608447e-05,
+      "loss": 0.2277,
+      "step": 7300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7023979957050824e-05,
+      "loss": 0.149,
+      "step": 7301
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.702219040801718e-05,
+      "loss": 0.1792,
+      "step": 7302
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.702040085898354e-05,
+      "loss": 0.1816,
+      "step": 7303
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7018611309949896e-05,
+      "loss": 0.0967,
+      "step": 7304
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.701682176091625e-05,
+      "loss": 0.1756,
+      "step": 7305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7015032211882606e-05,
+      "loss": 0.1763,
+      "step": 7306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.701324266284896e-05,
+      "loss": 0.2152,
+      "step": 7307
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.701145311381532e-05,
+      "loss": 0.1931,
+      "step": 7308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.700966356478168e-05,
+      "loss": 0.2477,
+      "step": 7309
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7007874015748034e-05,
+      "loss": 0.1179,
+      "step": 7310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.700608446671439e-05,
+      "loss": 0.249,
+      "step": 7311
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.700429491768075e-05,
+      "loss": 0.1556,
+      "step": 7312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.70025053686471e-05,
+      "loss": 0.303,
+      "step": 7313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7000715819613455e-05,
+      "loss": 0.1603,
+      "step": 7314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699892627057982e-05,
+      "loss": 0.18,
+      "step": 7315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699713672154617e-05,
+      "loss": 0.1405,
+      "step": 7316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699534717251253e-05,
+      "loss": 0.1597,
+      "step": 7317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699355762347889e-05,
+      "loss": 0.205,
+      "step": 7318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6991768074445244e-05,
+      "loss": 0.2092,
+      "step": 7319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.69899785254116e-05,
+      "loss": 0.2018,
+      "step": 7320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6988188976377955e-05,
+      "loss": 0.323,
+      "step": 7321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.698639942734431e-05,
+      "loss": 0.1521,
+      "step": 7322
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6984609878310665e-05,
+      "loss": 0.1364,
+      "step": 7323
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.698282032927703e-05,
+      "loss": 0.2038,
+      "step": 7324
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.698103078024338e-05,
+      "loss": 0.0874,
+      "step": 7325
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.697924123120974e-05,
+      "loss": 0.1586,
+      "step": 7326
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.697745168217609e-05,
+      "loss": 0.2898,
+      "step": 7327
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6975662133142454e-05,
+      "loss": 0.2342,
+      "step": 7328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.69738725841088e-05,
+      "loss": 0.1695,
+      "step": 7329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.697208303507516e-05,
+      "loss": 0.4141,
+      "step": 7330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.697029348604152e-05,
+      "loss": 0.1449,
+      "step": 7331
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6968503937007875e-05,
+      "loss": 0.27,
+      "step": 7332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.696671438797423e-05,
+      "loss": 0.0539,
+      "step": 7333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.696492483894059e-05,
+      "loss": 0.1767,
+      "step": 7334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.696313528990695e-05,
+      "loss": 0.238,
+      "step": 7335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.69613457408733e-05,
+      "loss": 0.1115,
+      "step": 7336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.695955619183966e-05,
+      "loss": 0.2925,
+      "step": 7337
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.695776664280601e-05,
+      "loss": 0.3117,
+      "step": 7338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.695597709377237e-05,
+      "loss": 0.1605,
+      "step": 7339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.695418754473873e-05,
+      "loss": 0.1938,
+      "step": 7340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6952397995705086e-05,
+      "loss": 0.1781,
+      "step": 7341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.695060844667144e-05,
+      "loss": 0.3545,
+      "step": 7342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6948818897637796e-05,
+      "loss": 0.3259,
+      "step": 7343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.694702934860415e-05,
+      "loss": 0.1839,
+      "step": 7344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6945239799570506e-05,
+      "loss": 0.2173,
+      "step": 7345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.694345025053687e-05,
+      "loss": 0.1664,
+      "step": 7346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6941660701503223e-05,
+      "loss": 0.2636,
+      "step": 7347
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.693987115246958e-05,
+      "loss": 0.3097,
+      "step": 7348
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6938081603435934e-05,
+      "loss": 0.4013,
+      "step": 7349
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6936292054402296e-05,
+      "loss": 0.2703,
+      "step": 7350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.693450250536865e-05,
+      "loss": 0.2384,
+      "step": 7351
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6932712956335e-05,
+      "loss": 0.2287,
+      "step": 7352
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.693092340730136e-05,
+      "loss": 0.1451,
+      "step": 7353
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6929133858267717e-05,
+      "loss": 0.1991,
+      "step": 7354
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.692734430923407e-05,
+      "loss": 0.1885,
+      "step": 7355
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6925554760200434e-05,
+      "loss": 0.2367,
+      "step": 7356
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.692376521116679e-05,
+      "loss": 0.1585,
+      "step": 7357
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6921975662133144e-05,
+      "loss": 0.1814,
+      "step": 7358
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6920186113099506e-05,
+      "loss": 0.1624,
+      "step": 7359
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6918396564065855e-05,
+      "loss": 0.1539,
+      "step": 7360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.691660701503221e-05,
+      "loss": 0.1451,
+      "step": 7361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.691481746599857e-05,
+      "loss": 0.1875,
+      "step": 7362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.691302791696493e-05,
+      "loss": 0.2288,
+      "step": 7363
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.691123836793128e-05,
+      "loss": 0.242,
+      "step": 7364
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.690944881889764e-05,
+      "loss": 0.1679,
+      "step": 7365
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6907659269864e-05,
+      "loss": 0.3112,
+      "step": 7366
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6905869720830354e-05,
+      "loss": 0.2957,
+      "step": 7367
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.690408017179671e-05,
+      "loss": 0.329,
+      "step": 7368
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6902290622763065e-05,
+      "loss": 0.1725,
+      "step": 7369
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.690050107372942e-05,
+      "loss": 0.1348,
+      "step": 7370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6898711524695775e-05,
+      "loss": 0.1609,
+      "step": 7371
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.689692197566214e-05,
+      "loss": 0.183,
+      "step": 7372
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.689513242662849e-05,
+      "loss": 0.0995,
+      "step": 7373
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.689334287759485e-05,
+      "loss": 0.1211,
+      "step": 7374
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.689155332856121e-05,
+      "loss": 0.1136,
+      "step": 7375
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688976377952756e-05,
+      "loss": 0.2195,
+      "step": 7376
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688797423049391e-05,
+      "loss": 0.0927,
+      "step": 7377
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6886184681460275e-05,
+      "loss": 0.0868,
+      "step": 7378
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688439513242663e-05,
+      "loss": 0.2885,
+      "step": 7379
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6882605583392986e-05,
+      "loss": 0.0871,
+      "step": 7380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688081603435935e-05,
+      "loss": 0.3685,
+      "step": 7381
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.68790264853257e-05,
+      "loss": 0.151,
+      "step": 7382
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.687723693629206e-05,
+      "loss": 0.2202,
+      "step": 7383
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.687544738725841e-05,
+      "loss": 0.2652,
+      "step": 7384
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.687365783822477e-05,
+      "loss": 0.1332,
+      "step": 7385
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6871868289191123e-05,
+      "loss": 0.0842,
+      "step": 7386
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6870078740157485e-05,
+      "loss": 0.1148,
+      "step": 7387
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.686828919112384e-05,
+      "loss": 0.1048,
+      "step": 7388
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6866499642090196e-05,
+      "loss": 0.108,
+      "step": 7389
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.686471009305655e-05,
+      "loss": 0.1557,
+      "step": 7390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6862920544022906e-05,
+      "loss": 0.2187,
+      "step": 7391
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.686113099498926e-05,
+      "loss": 0.1422,
+      "step": 7392
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6859341445955617e-05,
+      "loss": 0.2324,
+      "step": 7393
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.685755189692198e-05,
+      "loss": 0.1724,
+      "step": 7394
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6855762347888334e-05,
+      "loss": 0.2408,
+      "step": 7395
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.685397279885469e-05,
+      "loss": 0.1551,
+      "step": 7396
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.685218324982105e-05,
+      "loss": 0.1577,
+      "step": 7397
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6850393700787406e-05,
+      "loss": 0.1949,
+      "step": 7398
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6848604151753754e-05,
+      "loss": 0.1581,
+      "step": 7399
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6846814602720116e-05,
+      "loss": 0.1323,
+      "step": 7400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.684502505368647e-05,
+      "loss": 0.1863,
+      "step": 7401
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.684323550465283e-05,
+      "loss": 0.1623,
+      "step": 7402
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.684144595561919e-05,
+      "loss": 0.2401,
+      "step": 7403
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6839656406585544e-05,
+      "loss": 0.1485,
+      "step": 7404
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.68378668575519e-05,
+      "loss": 0.249,
+      "step": 7405
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6836077308518254e-05,
+      "loss": 0.234,
+      "step": 7406
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.683428775948461e-05,
+      "loss": 0.2966,
+      "step": 7407
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6832498210450965e-05,
+      "loss": 0.1019,
+      "step": 7408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.683070866141733e-05,
+      "loss": 0.3463,
+      "step": 7409
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.682891911238368e-05,
+      "loss": 0.1669,
+      "step": 7410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.682712956335004e-05,
+      "loss": 0.1877,
+      "step": 7411
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.682534001431639e-05,
+      "loss": 0.2777,
+      "step": 7412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6823550465282754e-05,
+      "loss": 0.1973,
+      "step": 7413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.682176091624911e-05,
+      "loss": 0.1837,
+      "step": 7414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.681997136721546e-05,
+      "loss": 0.2939,
+      "step": 7415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.681818181818182e-05,
+      "loss": 0.1613,
+      "step": 7416
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6816392269148175e-05,
+      "loss": 0.1447,
+      "step": 7417
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.681460272011453e-05,
+      "loss": 0.1801,
+      "step": 7418
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.681281317108089e-05,
+      "loss": 0.1704,
+      "step": 7419
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.681102362204725e-05,
+      "loss": 0.098,
+      "step": 7420
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.68092340730136e-05,
+      "loss": 0.2041,
+      "step": 7421
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6807444523979965e-05,
+      "loss": 0.1773,
+      "step": 7422
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.680565497494631e-05,
+      "loss": 0.1652,
+      "step": 7423
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.680386542591267e-05,
+      "loss": 0.2003,
+      "step": 7424
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.680207587687903e-05,
+      "loss": 0.1138,
+      "step": 7425
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6800286327845385e-05,
+      "loss": 0.1176,
+      "step": 7426
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.679849677881174e-05,
+      "loss": 0.1625,
+      "step": 7427
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6796707229778096e-05,
+      "loss": 0.1709,
+      "step": 7428
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.679491768074446e-05,
+      "loss": 0.158,
+      "step": 7429
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.679312813171081e-05,
+      "loss": 0.1388,
+      "step": 7430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.679133858267717e-05,
+      "loss": 0.1788,
+      "step": 7431
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.678954903364352e-05,
+      "loss": 0.2711,
+      "step": 7432
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.678775948460988e-05,
+      "loss": 0.1955,
+      "step": 7433
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6785969935576234e-05,
+      "loss": 0.1734,
+      "step": 7434
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6784180386542596e-05,
+      "loss": 0.1767,
+      "step": 7435
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.678239083750895e-05,
+      "loss": 0.2414,
+      "step": 7436
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6780601288475306e-05,
+      "loss": 0.2137,
+      "step": 7437
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.677881173944166e-05,
+      "loss": 0.1768,
+      "step": 7438
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6777022190408016e-05,
+      "loss": 0.3226,
+      "step": 7439
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.677523264137437e-05,
+      "loss": 0.2227,
+      "step": 7440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6773443092340734e-05,
+      "loss": 0.1918,
+      "step": 7441
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.677165354330709e-05,
+      "loss": 0.1735,
+      "step": 7442
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6769863994273444e-05,
+      "loss": 0.2894,
+      "step": 7443
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6768074445239806e-05,
+      "loss": 0.1925,
+      "step": 7444
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.676628489620616e-05,
+      "loss": 0.1619,
+      "step": 7445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.676449534717251e-05,
+      "loss": 0.0745,
+      "step": 7446
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.676270579813887e-05,
+      "loss": 0.1405,
+      "step": 7447
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.676091624910523e-05,
+      "loss": 0.1389,
+      "step": 7448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.675912670007158e-05,
+      "loss": 0.1774,
+      "step": 7449
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6757337151037944e-05,
+      "loss": 0.1624,
+      "step": 7450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.67555476020043e-05,
+      "loss": 0.2214,
+      "step": 7451
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6753758052970654e-05,
+      "loss": 0.1679,
+      "step": 7452
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.675196850393701e-05,
+      "loss": 0.1363,
+      "step": 7453
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6750178954903365e-05,
+      "loss": 0.2058,
+      "step": 7454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.674838940586972e-05,
+      "loss": 0.1543,
+      "step": 7455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6746599856836075e-05,
+      "loss": 0.1989,
+      "step": 7456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.674481030780244e-05,
+      "loss": 0.1039,
+      "step": 7457
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.674302075876879e-05,
+      "loss": 0.1403,
+      "step": 7458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.674123120973515e-05,
+      "loss": 0.2619,
+      "step": 7459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.673944166070151e-05,
+      "loss": 0.177,
+      "step": 7460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6737652111667865e-05,
+      "loss": 0.1381,
+      "step": 7461
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.673586256263421e-05,
+      "loss": 0.3375,
+      "step": 7462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6734073013600575e-05,
+      "loss": 0.1741,
+      "step": 7463
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.673228346456693e-05,
+      "loss": 0.2224,
+      "step": 7464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6730493915533285e-05,
+      "loss": 0.257,
+      "step": 7465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.672870436649965e-05,
+      "loss": 0.12,
+      "step": 7466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6726914817466e-05,
+      "loss": 0.1792,
+      "step": 7467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.672512526843236e-05,
+      "loss": 0.2683,
+      "step": 7468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.672333571939871e-05,
+      "loss": 0.2917,
+      "step": 7469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.672154617036507e-05,
+      "loss": 0.0888,
+      "step": 7470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.671975662133142e-05,
+      "loss": 0.1444,
+      "step": 7471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6717967072297785e-05,
+      "loss": 0.1712,
+      "step": 7472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.671617752326414e-05,
+      "loss": 0.2462,
+      "step": 7473
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6714387974230496e-05,
+      "loss": 0.1216,
+      "step": 7474
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.671259842519685e-05,
+      "loss": 0.1955,
+      "step": 7475
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.671080887616321e-05,
+      "loss": 0.1357,
+      "step": 7476
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670901932712956e-05,
+      "loss": 0.1628,
+      "step": 7477
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6707229778095916e-05,
+      "loss": 0.3596,
+      "step": 7478
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670544022906228e-05,
+      "loss": 0.2894,
+      "step": 7479
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6703650680028634e-05,
+      "loss": 0.2904,
+      "step": 7480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670186113099499e-05,
+      "loss": 0.2569,
+      "step": 7481
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670007158196135e-05,
+      "loss": 0.1421,
+      "step": 7482
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6698282032927706e-05,
+      "loss": 0.3087,
+      "step": 7483
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.669649248389406e-05,
+      "loss": 0.1708,
+      "step": 7484
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6694702934860416e-05,
+      "loss": 0.092,
+      "step": 7485
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.669291338582677e-05,
+      "loss": 0.2283,
+      "step": 7486
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.669112383679313e-05,
+      "loss": 0.3695,
+      "step": 7487
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.668933428775949e-05,
+      "loss": 0.1189,
+      "step": 7488
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6687544738725844e-05,
+      "loss": 0.2005,
+      "step": 7489
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.66857551896922e-05,
+      "loss": 0.1759,
+      "step": 7490
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6683965640658554e-05,
+      "loss": 0.3104,
+      "step": 7491
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6682176091624916e-05,
+      "loss": 0.2526,
+      "step": 7492
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6680386542591265e-05,
+      "loss": 0.1912,
+      "step": 7493
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6678596993557627e-05,
+      "loss": 0.1543,
+      "step": 7494
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.667680744452398e-05,
+      "loss": 0.1903,
+      "step": 7495
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.667501789549034e-05,
+      "loss": 0.3033,
+      "step": 7496
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.667322834645669e-05,
+      "loss": 0.1937,
+      "step": 7497
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6671438797423054e-05,
+      "loss": 0.1241,
+      "step": 7498
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.666964924838941e-05,
+      "loss": 0.2343,
+      "step": 7499
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6667859699355765e-05,
+      "loss": 0.1755,
+      "step": 7500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.666607015032212e-05,
+      "loss": 0.2132,
+      "step": 7501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6664280601288475e-05,
+      "loss": 0.1793,
+      "step": 7502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.666249105225483e-05,
+      "loss": 0.2122,
+      "step": 7503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.666070150322119e-05,
+      "loss": 0.2108,
+      "step": 7504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.665891195418755e-05,
+      "loss": 0.1812,
+      "step": 7505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.66571224051539e-05,
+      "loss": 0.1505,
+      "step": 7506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6655332856120264e-05,
+      "loss": 0.2605,
+      "step": 7507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.665354330708662e-05,
+      "loss": 0.3032,
+      "step": 7508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.665175375805297e-05,
+      "loss": 0.1078,
+      "step": 7509
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.664996420901933e-05,
+      "loss": 0.1848,
+      "step": 7510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6648174659985685e-05,
+      "loss": 0.1036,
+      "step": 7511
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.664638511095204e-05,
+      "loss": 0.2165,
+      "step": 7512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6644595561918396e-05,
+      "loss": 0.2136,
+      "step": 7513
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.664280601288476e-05,
+      "loss": 0.1697,
+      "step": 7514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.664101646385111e-05,
+      "loss": 0.1284,
+      "step": 7515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.663922691481747e-05,
+      "loss": 0.1233,
+      "step": 7516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.663743736578382e-05,
+      "loss": 0.2,
+      "step": 7517
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.663564781675018e-05,
+      "loss": 0.1536,
+      "step": 7518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6633858267716534e-05,
+      "loss": 0.1604,
+      "step": 7519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6632068718682895e-05,
+      "loss": 0.124,
+      "step": 7520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.663027916964925e-05,
+      "loss": 0.1817,
+      "step": 7521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6628489620615606e-05,
+      "loss": 0.2922,
+      "step": 7522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.662670007158197e-05,
+      "loss": 0.1886,
+      "step": 7523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6624910522548316e-05,
+      "loss": 0.2179,
+      "step": 7524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.662312097351467e-05,
+      "loss": 0.2338,
+      "step": 7525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6621331424481033e-05,
+      "loss": 0.1735,
+      "step": 7526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.661954187544739e-05,
+      "loss": 0.2449,
+      "step": 7527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6617752326413744e-05,
+      "loss": 0.1859,
+      "step": 7528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6615962777380106e-05,
+      "loss": 0.1911,
+      "step": 7529
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.661417322834646e-05,
+      "loss": 0.1068,
+      "step": 7530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6612383679312816e-05,
+      "loss": 0.1654,
+      "step": 7531
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.661059413027917e-05,
+      "loss": 0.1606,
+      "step": 7532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6608804581245527e-05,
+      "loss": 0.3072,
+      "step": 7533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.660701503221188e-05,
+      "loss": 0.1502,
+      "step": 7534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6605225483178244e-05,
+      "loss": 0.2169,
+      "step": 7535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.66034359341446e-05,
+      "loss": 0.1605,
+      "step": 7536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6601646385110954e-05,
+      "loss": 0.2797,
+      "step": 7537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659985683607731e-05,
+      "loss": 0.1503,
+      "step": 7538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659806728704367e-05,
+      "loss": 0.2368,
+      "step": 7539
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659627773801002e-05,
+      "loss": 0.0913,
+      "step": 7540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6594488188976375e-05,
+      "loss": 0.082,
+      "step": 7541
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659269863994274e-05,
+      "loss": 0.117,
+      "step": 7542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659090909090909e-05,
+      "loss": 0.2246,
+      "step": 7543
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658911954187545e-05,
+      "loss": 0.2226,
+      "step": 7544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658732999284181e-05,
+      "loss": 0.3419,
+      "step": 7545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6585540443808164e-05,
+      "loss": 0.2931,
+      "step": 7546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658375089477452e-05,
+      "loss": 0.0981,
+      "step": 7547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6581961345740875e-05,
+      "loss": 0.0801,
+      "step": 7548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658017179670723e-05,
+      "loss": 0.193,
+      "step": 7549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6578382247673585e-05,
+      "loss": 0.2713,
+      "step": 7550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.657659269863995e-05,
+      "loss": 0.2663,
+      "step": 7551
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.65748031496063e-05,
+      "loss": 0.2653,
+      "step": 7552
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.657301360057266e-05,
+      "loss": 0.2652,
+      "step": 7553
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.657122405153901e-05,
+      "loss": 0.3286,
+      "step": 7554
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6569434502505375e-05,
+      "loss": 0.1433,
+      "step": 7555
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.656764495347172e-05,
+      "loss": 0.1654,
+      "step": 7556
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6565855404438085e-05,
+      "loss": 0.1903,
+      "step": 7557
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.656406585540444e-05,
+      "loss": 0.3434,
+      "step": 7558
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6562276306370795e-05,
+      "loss": 0.2795,
+      "step": 7559
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.656048675733715e-05,
+      "loss": 0.2123,
+      "step": 7560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.655869720830351e-05,
+      "loss": 0.18,
+      "step": 7561
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.655690765926987e-05,
+      "loss": 0.2602,
+      "step": 7562
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6555118110236216e-05,
+      "loss": 0.1938,
+      "step": 7563
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.655332856120258e-05,
+      "loss": 0.1249,
+      "step": 7564
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.655153901216893e-05,
+      "loss": 0.3291,
+      "step": 7565
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.654974946313529e-05,
+      "loss": 0.1455,
+      "step": 7566
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.654795991410165e-05,
+      "loss": 0.1833,
+      "step": 7567
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6546170365068006e-05,
+      "loss": 0.1902,
+      "step": 7568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.654438081603436e-05,
+      "loss": 0.249,
+      "step": 7569
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.654259126700072e-05,
+      "loss": 0.2516,
+      "step": 7570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.654080171796707e-05,
+      "loss": 0.1103,
+      "step": 7571
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6539012168933426e-05,
+      "loss": 0.2593,
+      "step": 7572
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.653722261989979e-05,
+      "loss": 0.3541,
+      "step": 7573
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6535433070866144e-05,
+      "loss": 0.1292,
+      "step": 7574
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.65336435218325e-05,
+      "loss": 0.1744,
+      "step": 7575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6531853972798854e-05,
+      "loss": 0.2723,
+      "step": 7576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6530064423765216e-05,
+      "loss": 0.1714,
+      "step": 7577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.652827487473157e-05,
+      "loss": 0.2001,
+      "step": 7578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6526485325697926e-05,
+      "loss": 0.2557,
+      "step": 7579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.652469577666428e-05,
+      "loss": 0.2004,
+      "step": 7580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.652290622763064e-05,
+      "loss": 0.1771,
+      "step": 7581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.652111667859699e-05,
+      "loss": 0.1895,
+      "step": 7582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6519327129563354e-05,
+      "loss": 0.329,
+      "step": 7583
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.651753758052971e-05,
+      "loss": 0.3314,
+      "step": 7584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6515748031496064e-05,
+      "loss": 0.1509,
+      "step": 7585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6513958482462426e-05,
+      "loss": 0.1885,
+      "step": 7586
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6512168933428775e-05,
+      "loss": 0.1718,
+      "step": 7587
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.651037938439513e-05,
+      "loss": 0.1547,
+      "step": 7588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.650858983536149e-05,
+      "loss": 0.1909,
+      "step": 7589
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.650680028632785e-05,
+      "loss": 0.3272,
+      "step": 7590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.65050107372942e-05,
+      "loss": 0.1405,
+      "step": 7591
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6503221188260564e-05,
+      "loss": 0.1175,
+      "step": 7592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.650143163922692e-05,
+      "loss": 0.2439,
+      "step": 7593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6499642090193275e-05,
+      "loss": 0.23,
+      "step": 7594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.649785254115963e-05,
+      "loss": 0.2639,
+      "step": 7595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6496062992125985e-05,
+      "loss": 0.1561,
+      "step": 7596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.649427344309234e-05,
+      "loss": 0.1369,
+      "step": 7597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.64924838940587e-05,
+      "loss": 0.0905,
+      "step": 7598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.649069434502506e-05,
+      "loss": 0.2833,
+      "step": 7599
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648890479599141e-05,
+      "loss": 0.2037,
+      "step": 7600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648711524695777e-05,
+      "loss": 0.3491,
+      "step": 7601
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648532569792413e-05,
+      "loss": 0.1204,
+      "step": 7602
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648353614889048e-05,
+      "loss": 0.243,
+      "step": 7603
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648174659985683e-05,
+      "loss": 0.1699,
+      "step": 7604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6479957050823195e-05,
+      "loss": 0.2496,
+      "step": 7605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.647816750178955e-05,
+      "loss": 0.3317,
+      "step": 7606
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6476377952755906e-05,
+      "loss": 0.1434,
+      "step": 7607
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.647458840372227e-05,
+      "loss": 0.2025,
+      "step": 7608
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.647279885468862e-05,
+      "loss": 0.256,
+      "step": 7609
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.647100930565497e-05,
+      "loss": 0.2075,
+      "step": 7610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.646921975662133e-05,
+      "loss": 0.1279,
+      "step": 7611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.646743020758769e-05,
+      "loss": 0.155,
+      "step": 7612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6465640658554044e-05,
+      "loss": 0.1852,
+      "step": 7613
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6463851109520406e-05,
+      "loss": 0.3061,
+      "step": 7614
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.646206156048676e-05,
+      "loss": 0.2155,
+      "step": 7615
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6460272011453116e-05,
+      "loss": 0.1469,
+      "step": 7616
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.645848246241947e-05,
+      "loss": 0.3469,
+      "step": 7617
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6456692913385826e-05,
+      "loss": 0.3319,
+      "step": 7618
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.645490336435218e-05,
+      "loss": 0.3557,
+      "step": 7619
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6453113815318544e-05,
+      "loss": 0.2013,
+      "step": 7620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.64513242662849e-05,
+      "loss": 0.1594,
+      "step": 7621
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6449534717251254e-05,
+      "loss": 0.1911,
+      "step": 7622
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.644774516821761e-05,
+      "loss": 0.2233,
+      "step": 7623
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.644595561918397e-05,
+      "loss": 0.1644,
+      "step": 7624
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6444166070150326e-05,
+      "loss": 0.1686,
+      "step": 7625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6442376521116675e-05,
+      "loss": 0.2167,
+      "step": 7626
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.644058697208304e-05,
+      "loss": 0.2572,
+      "step": 7627
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.643879742304939e-05,
+      "loss": 0.2652,
+      "step": 7628
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.643700787401575e-05,
+      "loss": 0.2296,
+      "step": 7629
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.643521832498211e-05,
+      "loss": 0.2929,
+      "step": 7630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6433428775948464e-05,
+      "loss": 0.181,
+      "step": 7631
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.643163922691482e-05,
+      "loss": 0.1851,
+      "step": 7632
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.642984967788118e-05,
+      "loss": 0.1809,
+      "step": 7633
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.642806012884753e-05,
+      "loss": 0.3184,
+      "step": 7634
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6426270579813885e-05,
+      "loss": 0.1903,
+      "step": 7635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.642448103078025e-05,
+      "loss": 0.204,
+      "step": 7636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.64226914817466e-05,
+      "loss": 0.1551,
+      "step": 7637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.642090193271296e-05,
+      "loss": 0.0985,
+      "step": 7638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.641911238367931e-05,
+      "loss": 0.2459,
+      "step": 7639
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6417322834645674e-05,
+      "loss": 0.2551,
+      "step": 7640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.641553328561203e-05,
+      "loss": 0.269,
+      "step": 7641
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6413743736578385e-05,
+      "loss": 0.2827,
+      "step": 7642
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.641195418754474e-05,
+      "loss": 0.1678,
+      "step": 7643
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6410164638511095e-05,
+      "loss": 0.1693,
+      "step": 7644
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640837508947745e-05,
+      "loss": 0.1779,
+      "step": 7645
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640658554044381e-05,
+      "loss": 0.1976,
+      "step": 7646
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640479599141017e-05,
+      "loss": 0.223,
+      "step": 7647
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640300644237652e-05,
+      "loss": 0.1225,
+      "step": 7648
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6401216893342885e-05,
+      "loss": 0.2295,
+      "step": 7649
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639942734430923e-05,
+      "loss": 0.115,
+      "step": 7650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639763779527559e-05,
+      "loss": 0.349,
+      "step": 7651
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639584824624195e-05,
+      "loss": 0.1415,
+      "step": 7652
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6394058697208306e-05,
+      "loss": 0.4851,
+      "step": 7653
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639226914817466e-05,
+      "loss": 0.2788,
+      "step": 7654
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639047959914102e-05,
+      "loss": 0.1907,
+      "step": 7655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.638869005010738e-05,
+      "loss": 0.1048,
+      "step": 7656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6386900501073726e-05,
+      "loss": 0.1925,
+      "step": 7657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.638511095204009e-05,
+      "loss": 0.1699,
+      "step": 7658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6383321403006443e-05,
+      "loss": 0.3467,
+      "step": 7659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.63815318539728e-05,
+      "loss": 0.1647,
+      "step": 7660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.637974230493916e-05,
+      "loss": 0.354,
+      "step": 7661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6377952755905516e-05,
+      "loss": 0.2647,
+      "step": 7662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.637616320687187e-05,
+      "loss": 0.2021,
+      "step": 7663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6374373657838226e-05,
+      "loss": 0.3622,
+      "step": 7664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.637258410880458e-05,
+      "loss": 0.0816,
+      "step": 7665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6370794559770937e-05,
+      "loss": 0.1731,
+      "step": 7666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.636900501073729e-05,
+      "loss": 0.2311,
+      "step": 7667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6367215461703654e-05,
+      "loss": 0.2846,
+      "step": 7668
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.636542591267001e-05,
+      "loss": 0.168,
+      "step": 7669
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.1679,
+      "step": 7670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6361846814602726e-05,
+      "loss": 0.2546,
+      "step": 7671
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.636005726556908e-05,
+      "loss": 0.1507,
+      "step": 7672
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.635826771653543e-05,
+      "loss": 0.2964,
+      "step": 7673
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.635647816750179e-05,
+      "loss": 0.3072,
+      "step": 7674
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.635468861846815e-05,
+      "loss": 0.2274,
+      "step": 7675
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.63528990694345e-05,
+      "loss": 0.2987,
+      "step": 7676
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6351109520400864e-05,
+      "loss": 0.247,
+      "step": 7677
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.634931997136722e-05,
+      "loss": 0.1539,
+      "step": 7678
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6347530422333574e-05,
+      "loss": 0.0995,
+      "step": 7679
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.634574087329993e-05,
+      "loss": 0.2134,
+      "step": 7680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6343951324266285e-05,
+      "loss": 0.2071,
+      "step": 7681
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.634216177523264e-05,
+      "loss": 0.1701,
+      "step": 7682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6340372226199e-05,
+      "loss": 0.1253,
+      "step": 7683
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.633858267716536e-05,
+      "loss": 0.233,
+      "step": 7684
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.633679312813171e-05,
+      "loss": 0.0547,
+      "step": 7685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.633500357909807e-05,
+      "loss": 0.1965,
+      "step": 7686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.633321403006443e-05,
+      "loss": 0.1552,
+      "step": 7687
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6331424481030785e-05,
+      "loss": 0.1704,
+      "step": 7688
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.632963493199713e-05,
+      "loss": 0.1088,
+      "step": 7689
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6327845382963495e-05,
+      "loss": 0.1496,
+      "step": 7690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.632605583392985e-05,
+      "loss": 0.1194,
+      "step": 7691
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6324266284896205e-05,
+      "loss": 0.1538,
+      "step": 7692
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.632247673586257e-05,
+      "loss": 0.1614,
+      "step": 7693
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.632068718682892e-05,
+      "loss": 0.2032,
+      "step": 7694
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.631889763779528e-05,
+      "loss": 0.2441,
+      "step": 7695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.631710808876163e-05,
+      "loss": 0.3692,
+      "step": 7696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.631531853972799e-05,
+      "loss": 0.1099,
+      "step": 7697
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6313528990694343e-05,
+      "loss": 0.2587,
+      "step": 7698
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6311739441660705e-05,
+      "loss": 0.1746,
+      "step": 7699
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630994989262706e-05,
+      "loss": 0.1876,
+      "step": 7700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6308160343593416e-05,
+      "loss": 0.291,
+      "step": 7701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630637079455977e-05,
+      "loss": 0.2377,
+      "step": 7702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630458124552613e-05,
+      "loss": 0.118,
+      "step": 7703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630279169649248e-05,
+      "loss": 0.1509,
+      "step": 7704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630100214745884e-05,
+      "loss": 0.2485,
+      "step": 7705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.62992125984252e-05,
+      "loss": 0.3698,
+      "step": 7706
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6297423049391554e-05,
+      "loss": 0.1831,
+      "step": 7707
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.629563350035791e-05,
+      "loss": 0.1773,
+      "step": 7708
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.629384395132427e-05,
+      "loss": 0.1288,
+      "step": 7709
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6292054402290626e-05,
+      "loss": 0.169,
+      "step": 7710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.629026485325698e-05,
+      "loss": 0.1656,
+      "step": 7711
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6288475304223336e-05,
+      "loss": 0.2419,
+      "step": 7712
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.628668575518969e-05,
+      "loss": 0.2008,
+      "step": 7713
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.628489620615605e-05,
+      "loss": 0.22,
+      "step": 7714
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.628310665712241e-05,
+      "loss": 0.4407,
+      "step": 7715
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6281317108088764e-05,
+      "loss": 0.192,
+      "step": 7716
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.627952755905512e-05,
+      "loss": 0.1563,
+      "step": 7717
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.627773801002148e-05,
+      "loss": 0.1043,
+      "step": 7718
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6275948460987836e-05,
+      "loss": 0.1868,
+      "step": 7719
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6274158911954185e-05,
+      "loss": 0.1496,
+      "step": 7720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.627236936292055e-05,
+      "loss": 0.2103,
+      "step": 7721
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.62705798138869e-05,
+      "loss": 0.1265,
+      "step": 7722
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.626879026485326e-05,
+      "loss": 0.2586,
+      "step": 7723
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.626700071581961e-05,
+      "loss": 0.2422,
+      "step": 7724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6265211166785974e-05,
+      "loss": 0.33,
+      "step": 7725
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.626342161775233e-05,
+      "loss": 0.149,
+      "step": 7726
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6261632068718685e-05,
+      "loss": 0.161,
+      "step": 7727
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.625984251968504e-05,
+      "loss": 0.0457,
+      "step": 7728
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6258052970651395e-05,
+      "loss": 0.1317,
+      "step": 7729
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.625626342161775e-05,
+      "loss": 0.1252,
+      "step": 7730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.625447387258411e-05,
+      "loss": 0.1631,
+      "step": 7731
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.625268432355047e-05,
+      "loss": 0.182,
+      "step": 7732
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.625089477451682e-05,
+      "loss": 0.2487,
+      "step": 7733
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6249105225483185e-05,
+      "loss": 0.1935,
+      "step": 7734
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.624731567644954e-05,
+      "loss": 0.1715,
+      "step": 7735
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.624552612741589e-05,
+      "loss": 0.1855,
+      "step": 7736
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.624373657838225e-05,
+      "loss": 0.2677,
+      "step": 7737
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6241947029348605e-05,
+      "loss": 0.1088,
+      "step": 7738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.624015748031496e-05,
+      "loss": 0.2493,
+      "step": 7739
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.623836793128132e-05,
+      "loss": 0.151,
+      "step": 7740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.623657838224768e-05,
+      "loss": 0.1308,
+      "step": 7741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.623478883321403e-05,
+      "loss": 0.239,
+      "step": 7742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.623299928418039e-05,
+      "loss": 0.1525,
+      "step": 7743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.623120973514674e-05,
+      "loss": 0.0632,
+      "step": 7744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.62294201861131e-05,
+      "loss": 0.0743,
+      "step": 7745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.622763063707946e-05,
+      "loss": 0.2669,
+      "step": 7746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6225841088045816e-05,
+      "loss": 0.2576,
+      "step": 7747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.622405153901217e-05,
+      "loss": 0.1997,
+      "step": 7748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6222261989978526e-05,
+      "loss": 0.2647,
+      "step": 7749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.622047244094489e-05,
+      "loss": 0.1699,
+      "step": 7750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6218682891911236e-05,
+      "loss": 0.1584,
+      "step": 7751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.621689334287759e-05,
+      "loss": 0.3082,
+      "step": 7752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6215103793843954e-05,
+      "loss": 0.2959,
+      "step": 7753
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621331424481031e-05,
+      "loss": 0.1873,
+      "step": 7754
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6211524695776664e-05,
+      "loss": 0.112,
+      "step": 7755
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6209735146743026e-05,
+      "loss": 0.1506,
+      "step": 7756
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620794559770938e-05,
+      "loss": 0.2284,
+      "step": 7757
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6206156048675736e-05,
+      "loss": 0.2081,
+      "step": 7758
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620436649964209e-05,
+      "loss": 0.1357,
+      "step": 7759
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620257695060845e-05,
+      "loss": 0.2347,
+      "step": 7760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62007874015748e-05,
+      "loss": 0.197,
+      "step": 7761
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6198997852541164e-05,
+      "loss": 0.188,
+      "step": 7762
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619720830350752e-05,
+      "loss": 0.2275,
+      "step": 7763
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6195418754473874e-05,
+      "loss": 0.1189,
+      "step": 7764
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619362920544023e-05,
+      "loss": 0.2341,
+      "step": 7765
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619183965640659e-05,
+      "loss": 0.3011,
+      "step": 7766
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619005010737294e-05,
+      "loss": 0.1915,
+      "step": 7767
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.61882605583393e-05,
+      "loss": 0.1165,
+      "step": 7768
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618647100930566e-05,
+      "loss": 0.3118,
+      "step": 7769
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618468146027201e-05,
+      "loss": 0.1552,
+      "step": 7770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618289191123837e-05,
+      "loss": 0.1727,
+      "step": 7771
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618110236220473e-05,
+      "loss": 0.2369,
+      "step": 7772
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6179312813171085e-05,
+      "loss": 0.1529,
+      "step": 7773
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.617752326413744e-05,
+      "loss": 0.1469,
+      "step": 7774
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6175733715103795e-05,
+      "loss": 0.2338,
+      "step": 7775
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.617394416607015e-05,
+      "loss": 0.3192,
+      "step": 7776
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6172154617036505e-05,
+      "loss": 0.2126,
+      "step": 7777
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.617036506800287e-05,
+      "loss": 0.123,
+      "step": 7778
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.616857551896922e-05,
+      "loss": 0.2116,
+      "step": 7779
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.616678596993558e-05,
+      "loss": 0.3111,
+      "step": 7780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.616499642090194e-05,
+      "loss": 0.1959,
+      "step": 7781
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.616320687186829e-05,
+      "loss": 0.1509,
+      "step": 7782
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.616141732283464e-05,
+      "loss": 0.2058,
+      "step": 7783
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6159627773801005e-05,
+      "loss": 0.1758,
+      "step": 7784
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.615783822476736e-05,
+      "loss": 0.136,
+      "step": 7785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6156048675733716e-05,
+      "loss": 0.0424,
+      "step": 7786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.615425912670007e-05,
+      "loss": 0.3213,
+      "step": 7787
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.615246957766643e-05,
+      "loss": 0.1784,
+      "step": 7788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.615068002863279e-05,
+      "loss": 0.172,
+      "step": 7789
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.614889047959914e-05,
+      "loss": 0.2112,
+      "step": 7790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.61471009305655e-05,
+      "loss": 0.1868,
+      "step": 7791
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6145311381531854e-05,
+      "loss": 0.2111,
+      "step": 7792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.614352183249821e-05,
+      "loss": 0.1799,
+      "step": 7793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.614173228346457e-05,
+      "loss": 0.142,
+      "step": 7794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6139942734430926e-05,
+      "loss": 0.2772,
+      "step": 7795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.613815318539728e-05,
+      "loss": 0.2123,
+      "step": 7796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.613636363636364e-05,
+      "loss": 0.2235,
+      "step": 7797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.613457408732999e-05,
+      "loss": 0.1603,
+      "step": 7798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.613278453829635e-05,
+      "loss": 0.1996,
+      "step": 7799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.613099498926271e-05,
+      "loss": 0.188,
+      "step": 7800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6129205440229064e-05,
+      "loss": 0.308,
+      "step": 7801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.612741589119542e-05,
+      "loss": 0.3334,
+      "step": 7802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.612562634216178e-05,
+      "loss": 0.2776,
+      "step": 7803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6123836793128136e-05,
+      "loss": 0.2147,
+      "step": 7804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.612204724409449e-05,
+      "loss": 0.1866,
+      "step": 7805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6120257695060847e-05,
+      "loss": 0.2182,
+      "step": 7806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.61184681460272e-05,
+      "loss": 0.1411,
+      "step": 7807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.611667859699356e-05,
+      "loss": 0.2069,
+      "step": 7808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.611488904795992e-05,
+      "loss": 0.1537,
+      "step": 7809
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6113099498926274e-05,
+      "loss": 0.1349,
+      "step": 7810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611130994989263e-05,
+      "loss": 0.3543,
+      "step": 7811
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6109520400858985e-05,
+      "loss": 0.3846,
+      "step": 7812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6107730851825346e-05,
+      "loss": 0.1707,
+      "step": 7813
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6105941302791695e-05,
+      "loss": 0.1077,
+      "step": 7814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610415175375805e-05,
+      "loss": 0.221,
+      "step": 7815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610236220472441e-05,
+      "loss": 0.06,
+      "step": 7816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610057265569077e-05,
+      "loss": 0.1648,
+      "step": 7817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609878310665712e-05,
+      "loss": 0.146,
+      "step": 7818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6096993557623484e-05,
+      "loss": 0.2551,
+      "step": 7819
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609520400858984e-05,
+      "loss": 0.1964,
+      "step": 7820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6093414459556195e-05,
+      "loss": 0.1955,
+      "step": 7821
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609162491052255e-05,
+      "loss": 0.2277,
+      "step": 7822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6089835361488905e-05,
+      "loss": 0.271,
+      "step": 7823
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608804581245526e-05,
+      "loss": 0.2268,
+      "step": 7824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608625626342162e-05,
+      "loss": 0.2119,
+      "step": 7825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608446671438798e-05,
+      "loss": 0.0871,
+      "step": 7826
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608267716535433e-05,
+      "loss": 0.207,
+      "step": 7827
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608088761632069e-05,
+      "loss": 0.2335,
+      "step": 7828
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607909806728704e-05,
+      "loss": 0.152,
+      "step": 7829
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60773085182534e-05,
+      "loss": 0.1001,
+      "step": 7830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607551896921976e-05,
+      "loss": 0.1878,
+      "step": 7831
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6073729420186115e-05,
+      "loss": 0.1779,
+      "step": 7832
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607193987115247e-05,
+      "loss": 0.1559,
+      "step": 7833
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6070150322118826e-05,
+      "loss": 0.1868,
+      "step": 7834
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606836077308519e-05,
+      "loss": 0.2429,
+      "step": 7835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606657122405154e-05,
+      "loss": 0.1307,
+      "step": 7836
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606478167501789e-05,
+      "loss": 0.2701,
+      "step": 7837
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6062992125984253e-05,
+      "loss": 0.2868,
+      "step": 7838
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606120257695061e-05,
+      "loss": 0.2197,
+      "step": 7839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6059413027916964e-05,
+      "loss": 0.1678,
+      "step": 7840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6057623478883326e-05,
+      "loss": 0.2622,
+      "step": 7841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.605583392984968e-05,
+      "loss": 0.1876,
+      "step": 7842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6054044380816036e-05,
+      "loss": 0.1272,
+      "step": 7843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60522548317824e-05,
+      "loss": 0.1267,
+      "step": 7844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6050465282748747e-05,
+      "loss": 0.3167,
+      "step": 7845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60486757337151e-05,
+      "loss": 0.0949,
+      "step": 7846
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6046886184681464e-05,
+      "loss": 0.1845,
+      "step": 7847
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604509663564782e-05,
+      "loss": 0.264,
+      "step": 7848
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6043307086614174e-05,
+      "loss": 0.3564,
+      "step": 7849
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604151753758053e-05,
+      "loss": 0.1049,
+      "step": 7850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603972798854689e-05,
+      "loss": 0.1746,
+      "step": 7851
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6037938439513246e-05,
+      "loss": 0.1211,
+      "step": 7852
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60361488904796e-05,
+      "loss": 0.2018,
+      "step": 7853
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603435934144596e-05,
+      "loss": 0.3589,
+      "step": 7854
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603256979241231e-05,
+      "loss": 0.1678,
+      "step": 7855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603078024337867e-05,
+      "loss": 0.3143,
+      "step": 7856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602899069434503e-05,
+      "loss": 0.2531,
+      "step": 7857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6027201145311384e-05,
+      "loss": 0.1995,
+      "step": 7858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602541159627774e-05,
+      "loss": 0.0934,
+      "step": 7859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60236220472441e-05,
+      "loss": 0.1936,
+      "step": 7860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602183249821045e-05,
+      "loss": 0.312,
+      "step": 7861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6020042949176805e-05,
+      "loss": 0.0975,
+      "step": 7862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.601825340014317e-05,
+      "loss": 0.1479,
+      "step": 7863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.601646385110952e-05,
+      "loss": 0.1736,
+      "step": 7864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.601467430207588e-05,
+      "loss": 0.2189,
+      "step": 7865
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.601288475304224e-05,
+      "loss": 0.0796,
+      "step": 7866
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6011095204008595e-05,
+      "loss": 0.1822,
+      "step": 7867
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600930565497495e-05,
+      "loss": 0.1754,
+      "step": 7868
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6007516105941305e-05,
+      "loss": 0.158,
+      "step": 7869
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600572655690766e-05,
+      "loss": 0.1147,
+      "step": 7870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6003937007874015e-05,
+      "loss": 0.255,
+      "step": 7871
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600214745884037e-05,
+      "loss": 0.2131,
+      "step": 7872
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600035790980673e-05,
+      "loss": 0.3362,
+      "step": 7873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599856836077309e-05,
+      "loss": 0.2126,
+      "step": 7874
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599677881173944e-05,
+      "loss": 0.1913,
+      "step": 7875
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59949892627058e-05,
+      "loss": 0.3777,
+      "step": 7876
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599319971367215e-05,
+      "loss": 0.1291,
+      "step": 7877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599141016463851e-05,
+      "loss": 0.3011,
+      "step": 7878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598962061560487e-05,
+      "loss": 0.3137,
+      "step": 7879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5987831066571226e-05,
+      "loss": 0.1022,
+      "step": 7880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598604151753758e-05,
+      "loss": 0.1617,
+      "step": 7881
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598425196850394e-05,
+      "loss": 0.1062,
+      "step": 7882
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59824624194703e-05,
+      "loss": 0.2262,
+      "step": 7883
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5980672870436646e-05,
+      "loss": 0.1866,
+      "step": 7884
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597888332140301e-05,
+      "loss": 0.1015,
+      "step": 7885
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5977093772369364e-05,
+      "loss": 0.2394,
+      "step": 7886
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597530422333572e-05,
+      "loss": 0.2456,
+      "step": 7887
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597351467430208e-05,
+      "loss": 0.128,
+      "step": 7888
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5971725125268436e-05,
+      "loss": 0.1295,
+      "step": 7889
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596993557623479e-05,
+      "loss": 0.161,
+      "step": 7890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5968146027201146e-05,
+      "loss": 0.1738,
+      "step": 7891
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59663564781675e-05,
+      "loss": 0.1723,
+      "step": 7892
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596456692913386e-05,
+      "loss": 0.1292,
+      "step": 7893
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596277738010022e-05,
+      "loss": 0.0946,
+      "step": 7894
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5960987831066574e-05,
+      "loss": 0.1405,
+      "step": 7895
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.595919828203293e-05,
+      "loss": 0.271,
+      "step": 7896
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5957408732999284e-05,
+      "loss": 0.2593,
+      "step": 7897
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5955619183965646e-05,
+      "loss": 0.2112,
+      "step": 7898
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5953829634932e-05,
+      "loss": 0.1391,
+      "step": 7899
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.595204008589835e-05,
+      "loss": 0.1356,
+      "step": 7900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.595025053686471e-05,
+      "loss": 0.0975,
+      "step": 7901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.594846098783107e-05,
+      "loss": 0.1712,
+      "step": 7902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.594667143879742e-05,
+      "loss": 0.1711,
+      "step": 7903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5944881889763784e-05,
+      "loss": 0.0946,
+      "step": 7904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.594309234073014e-05,
+      "loss": 0.1232,
+      "step": 7905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5941302791696495e-05,
+      "loss": 0.273,
+      "step": 7906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5939513242662857e-05,
+      "loss": 0.1264,
+      "step": 7907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5937723693629205e-05,
+      "loss": 0.1505,
+      "step": 7908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593593414459556e-05,
+      "loss": 0.1764,
+      "step": 7909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593414459556192e-05,
+      "loss": 0.1733,
+      "step": 7910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593235504652828e-05,
+      "loss": 0.2024,
+      "step": 7911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593056549749463e-05,
+      "loss": 0.1294,
+      "step": 7912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592877594846099e-05,
+      "loss": 0.4539,
+      "step": 7913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592698639942735e-05,
+      "loss": 0.2518,
+      "step": 7914
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59251968503937e-05,
+      "loss": 0.2023,
+      "step": 7915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592340730136006e-05,
+      "loss": 0.3493,
+      "step": 7916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5921617752326415e-05,
+      "loss": 0.198,
+      "step": 7917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.591982820329277e-05,
+      "loss": 0.1444,
+      "step": 7918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5918038654259126e-05,
+      "loss": 0.1013,
+      "step": 7919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.591624910522549e-05,
+      "loss": 0.2267,
+      "step": 7920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.591445955619184e-05,
+      "loss": 0.2001,
+      "step": 7921
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.59126700071582e-05,
+      "loss": 0.1142,
+      "step": 7922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.591088045812455e-05,
+      "loss": 0.1674,
+      "step": 7923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.590909090909091e-05,
+      "loss": 0.1932,
+      "step": 7924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5907301360057264e-05,
+      "loss": 0.1807,
+      "step": 7925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5905511811023626e-05,
+      "loss": 0.176,
+      "step": 7926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.590372226198998e-05,
+      "loss": 0.1648,
+      "step": 7927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5901932712956336e-05,
+      "loss": 0.1954,
+      "step": 7928
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.59001431639227e-05,
+      "loss": 0.0889,
+      "step": 7929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.589835361488905e-05,
+      "loss": 0.1546,
+      "step": 7930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58965640658554e-05,
+      "loss": 0.1959,
+      "step": 7931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5894774516821764e-05,
+      "loss": 0.1504,
+      "step": 7932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.589298496778812e-05,
+      "loss": 0.0911,
+      "step": 7933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5891195418754474e-05,
+      "loss": 0.2069,
+      "step": 7934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588940586972083e-05,
+      "loss": 0.1395,
+      "step": 7935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588761632068719e-05,
+      "loss": 0.179,
+      "step": 7936
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5885826771653546e-05,
+      "loss": 0.1763,
+      "step": 7937
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58840372226199e-05,
+      "loss": 0.1319,
+      "step": 7938
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588224767358626e-05,
+      "loss": 0.254,
+      "step": 7939
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588045812455261e-05,
+      "loss": 0.1579,
+      "step": 7940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587866857551897e-05,
+      "loss": 0.1694,
+      "step": 7941
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587687902648533e-05,
+      "loss": 0.1228,
+      "step": 7942
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5875089477451684e-05,
+      "loss": 0.1599,
+      "step": 7943
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587329992841804e-05,
+      "loss": 0.2956,
+      "step": 7944
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58715103793844e-05,
+      "loss": 0.3328,
+      "step": 7945
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5869720830350757e-05,
+      "loss": 0.1415,
+      "step": 7946
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5867931281317105e-05,
+      "loss": 0.1722,
+      "step": 7947
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.586614173228347e-05,
+      "loss": 0.3592,
+      "step": 7948
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.586435218324982e-05,
+      "loss": 0.154,
+      "step": 7949
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.586256263421618e-05,
+      "loss": 0.2412,
+      "step": 7950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.586077308518254e-05,
+      "loss": 0.1434,
+      "step": 7951
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5858983536148894e-05,
+      "loss": 0.2845,
+      "step": 7952
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585719398711525e-05,
+      "loss": 0.3809,
+      "step": 7953
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5855404438081605e-05,
+      "loss": 0.1998,
+      "step": 7954
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585361488904796e-05,
+      "loss": 0.1267,
+      "step": 7955
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5851825340014315e-05,
+      "loss": 0.2765,
+      "step": 7956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585003579098068e-05,
+      "loss": 0.1296,
+      "step": 7957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584824624194703e-05,
+      "loss": 0.1452,
+      "step": 7958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584645669291339e-05,
+      "loss": 0.1941,
+      "step": 7959
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584466714387974e-05,
+      "loss": 0.1079,
+      "step": 7960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5842877594846105e-05,
+      "loss": 0.3051,
+      "step": 7961
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584108804581245e-05,
+      "loss": 0.1691,
+      "step": 7962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583929849677881e-05,
+      "loss": 0.2511,
+      "step": 7963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583750894774517e-05,
+      "loss": 0.2472,
+      "step": 7964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5835719398711526e-05,
+      "loss": 0.188,
+      "step": 7965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583392984967788e-05,
+      "loss": 0.2749,
+      "step": 7966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583214030064424e-05,
+      "loss": 0.1904,
+      "step": 7967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58303507516106e-05,
+      "loss": 0.2667,
+      "step": 7968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582856120257695e-05,
+      "loss": 0.3025,
+      "step": 7969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582677165354331e-05,
+      "loss": 0.1593,
+      "step": 7970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5824982104509663e-05,
+      "loss": 0.1274,
+      "step": 7971
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582319255547602e-05,
+      "loss": 0.3446,
+      "step": 7972
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582140300644238e-05,
+      "loss": 0.1659,
+      "step": 7973
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5819613457408736e-05,
+      "loss": 0.1747,
+      "step": 7974
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581782390837509e-05,
+      "loss": 0.1748,
+      "step": 7975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5816034359341446e-05,
+      "loss": 0.269,
+      "step": 7976
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581424481030781e-05,
+      "loss": 0.197,
+      "step": 7977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5812455261274157e-05,
+      "loss": 0.1347,
+      "step": 7978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.581066571224052e-05,
+      "loss": 0.1907,
+      "step": 7979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5808876163206874e-05,
+      "loss": 0.0962,
+      "step": 7980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.580708661417323e-05,
+      "loss": 0.1038,
+      "step": 7981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5805297065139584e-05,
+      "loss": 0.0844,
+      "step": 7982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5803507516105946e-05,
+      "loss": 0.1445,
+      "step": 7983
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.58017179670723e-05,
+      "loss": 0.1364,
+      "step": 7984
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5799928418038656e-05,
+      "loss": 0.1053,
+      "step": 7985
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.579813886900501e-05,
+      "loss": 0.3248,
+      "step": 7986
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.579634931997137e-05,
+      "loss": 0.2298,
+      "step": 7987
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.579455977093772e-05,
+      "loss": 0.3379,
+      "step": 7988
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5792770221904084e-05,
+      "loss": 0.1814,
+      "step": 7989
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.579098067287044e-05,
+      "loss": 0.2764,
+      "step": 7990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5789191123836794e-05,
+      "loss": 0.165,
+      "step": 7991
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5787401574803156e-05,
+      "loss": 0.1382,
+      "step": 7992
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578561202576951e-05,
+      "loss": 0.1388,
+      "step": 7993
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578382247673586e-05,
+      "loss": 0.1296,
+      "step": 7994
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578203292770222e-05,
+      "loss": 0.194,
+      "step": 7995
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578024337866858e-05,
+      "loss": 0.1959,
+      "step": 7996
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.577845382963493e-05,
+      "loss": 0.305,
+      "step": 7997
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.577666428060129e-05,
+      "loss": 0.1941,
+      "step": 7998
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.577487473156765e-05,
+      "loss": 0.1323,
+      "step": 7999
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5773085182534005e-05,
+      "loss": 0.2676,
+      "step": 8000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.577129563350036e-05,
+      "loss": 0.181,
+      "step": 8001
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5769506084466715e-05,
+      "loss": 0.1891,
+      "step": 8002
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.576771653543307e-05,
+      "loss": 0.1566,
+      "step": 8003
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5765926986399425e-05,
+      "loss": 0.1426,
+      "step": 8004
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.576413743736579e-05,
+      "loss": 0.2072,
+      "step": 8005
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.576234788833214e-05,
+      "loss": 0.1488,
+      "step": 8006
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.57605583392985e-05,
+      "loss": 0.1039,
+      "step": 8007
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.575876879026486e-05,
+      "loss": 0.1879,
+      "step": 8008
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.575697924123121e-05,
+      "loss": 0.1441,
+      "step": 8009
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5755189692197563e-05,
+      "loss": 0.2463,
+      "step": 8010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5753400143163925e-05,
+      "loss": 0.2583,
+      "step": 8011
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.575161059413028e-05,
+      "loss": 0.1598,
+      "step": 8012
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5749821045096636e-05,
+      "loss": 0.1673,
+      "step": 8013
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5748031496063e-05,
+      "loss": 0.1934,
+      "step": 8014
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.574624194702935e-05,
+      "loss": 0.1154,
+      "step": 8015
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.574445239799571e-05,
+      "loss": 0.1145,
+      "step": 8016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.574266284896206e-05,
+      "loss": 0.1419,
+      "step": 8017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.574087329992842e-05,
+      "loss": 0.2452,
+      "step": 8018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5739083750894774e-05,
+      "loss": 0.1584,
+      "step": 8019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5737294201861136e-05,
+      "loss": 0.2126,
+      "step": 8020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573550465282749e-05,
+      "loss": 0.2279,
+      "step": 8021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5733715103793846e-05,
+      "loss": 0.2642,
+      "step": 8022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.57319255547602e-05,
+      "loss": 0.253,
+      "step": 8023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573013600572656e-05,
+      "loss": 0.2169,
+      "step": 8024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572834645669291e-05,
+      "loss": 0.4667,
+      "step": 8025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572655690765927e-05,
+      "loss": 0.1602,
+      "step": 8026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572476735862563e-05,
+      "loss": 0.2905,
+      "step": 8027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5722977809591984e-05,
+      "loss": 0.175,
+      "step": 8028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572118826055834e-05,
+      "loss": 0.1548,
+      "step": 8029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.57193987115247e-05,
+      "loss": 0.1652,
+      "step": 8030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5717609162491056e-05,
+      "loss": 0.2849,
+      "step": 8031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.571581961345741e-05,
+      "loss": 0.1607,
+      "step": 8032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.571403006442377e-05,
+      "loss": 0.151,
+      "step": 8033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.571224051539012e-05,
+      "loss": 0.1495,
+      "step": 8034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.571045096635648e-05,
+      "loss": 0.3648,
+      "step": 8035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.570866141732284e-05,
+      "loss": 0.2229,
+      "step": 8036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5706871868289194e-05,
+      "loss": 0.1533,
+      "step": 8037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.570508231925555e-05,
+      "loss": 0.185,
+      "step": 8038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5703292770221905e-05,
+      "loss": 0.1832,
+      "step": 8039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.570150322118827e-05,
+      "loss": 0.143,
+      "step": 8040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5699713672154615e-05,
+      "loss": 0.2457,
+      "step": 8041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569792412312098e-05,
+      "loss": 0.1172,
+      "step": 8042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569613457408733e-05,
+      "loss": 0.2946,
+      "step": 8043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569434502505369e-05,
+      "loss": 0.1357,
+      "step": 8044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569255547602004e-05,
+      "loss": 0.2354,
+      "step": 8045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5690765926986405e-05,
+      "loss": 0.1609,
+      "step": 8046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568897637795276e-05,
+      "loss": 0.2389,
+      "step": 8047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568718682891911e-05,
+      "loss": 0.1836,
+      "step": 8048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568539727988547e-05,
+      "loss": 0.3161,
+      "step": 8049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5683607730851825e-05,
+      "loss": 0.1486,
+      "step": 8050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568181818181818e-05,
+      "loss": 0.159,
+      "step": 8051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568002863278454e-05,
+      "loss": 0.2931,
+      "step": 8052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.56782390837509e-05,
+      "loss": 0.1177,
+      "step": 8053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.567644953471725e-05,
+      "loss": 0.3107,
+      "step": 8054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5674659985683615e-05,
+      "loss": 0.1167,
+      "step": 8055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.567287043664996e-05,
+      "loss": 0.171,
+      "step": 8056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.567108088761632e-05,
+      "loss": 0.1974,
+      "step": 8057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.566929133858268e-05,
+      "loss": 0.253,
+      "step": 8058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5667501789549036e-05,
+      "loss": 0.1523,
+      "step": 8059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.566571224051539e-05,
+      "loss": 0.1343,
+      "step": 8060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5663922691481746e-05,
+      "loss": 0.1943,
+      "step": 8061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.566213314244811e-05,
+      "loss": 0.1527,
+      "step": 8062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.566034359341446e-05,
+      "loss": 0.1745,
+      "step": 8063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.565855404438082e-05,
+      "loss": 0.2161,
+      "step": 8064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5656764495347174e-05,
+      "loss": 0.2982,
+      "step": 8065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.565497494631353e-05,
+      "loss": 0.2051,
+      "step": 8066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5653185397279884e-05,
+      "loss": 0.2869,
+      "step": 8067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5651395848246246e-05,
+      "loss": 0.1641,
+      "step": 8068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.56496062992126e-05,
+      "loss": 0.1569,
+      "step": 8069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5647816750178956e-05,
+      "loss": 0.3229,
+      "step": 8070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.564602720114532e-05,
+      "loss": 0.2672,
+      "step": 8071
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.564423765211167e-05,
+      "loss": 0.3406,
+      "step": 8072
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.564244810307802e-05,
+      "loss": 0.2726,
+      "step": 8073
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5640658554044384e-05,
+      "loss": 0.1587,
+      "step": 8074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.563886900501074e-05,
+      "loss": 0.1994,
+      "step": 8075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5637079455977094e-05,
+      "loss": 0.1803,
+      "step": 8076
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5635289906943456e-05,
+      "loss": 0.3163,
+      "step": 8077
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.563350035790981e-05,
+      "loss": 0.277,
+      "step": 8078
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5631710808876167e-05,
+      "loss": 0.2004,
+      "step": 8079
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562992125984252e-05,
+      "loss": 0.2108,
+      "step": 8080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562813171080888e-05,
+      "loss": 0.2569,
+      "step": 8081
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562634216177523e-05,
+      "loss": 0.1793,
+      "step": 8082
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562455261274159e-05,
+      "loss": 0.1957,
+      "step": 8083
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562276306370795e-05,
+      "loss": 0.1667,
+      "step": 8084
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5620973514674305e-05,
+      "loss": 0.1592,
+      "step": 8085
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.561918396564066e-05,
+      "loss": 0.2054,
+      "step": 8086
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.561739441660702e-05,
+      "loss": 0.2955,
+      "step": 8087
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.561560486757337e-05,
+      "loss": 0.1984,
+      "step": 8088
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5613815318539725e-05,
+      "loss": 0.1948,
+      "step": 8089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.561202576950609e-05,
+      "loss": 0.0967,
+      "step": 8090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.561023622047244e-05,
+      "loss": 0.3193,
+      "step": 8091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.56084466714388e-05,
+      "loss": 0.2004,
+      "step": 8092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.560665712240516e-05,
+      "loss": 0.2326,
+      "step": 8093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5604867573371515e-05,
+      "loss": 0.137,
+      "step": 8094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.560307802433786e-05,
+      "loss": 0.1918,
+      "step": 8095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5601288475304225e-05,
+      "loss": 0.2665,
+      "step": 8096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.559949892627058e-05,
+      "loss": 0.1367,
+      "step": 8097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5597709377236936e-05,
+      "loss": 0.232,
+      "step": 8098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.55959198282033e-05,
+      "loss": 0.196,
+      "step": 8099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.559413027916965e-05,
+      "loss": 0.268,
+      "step": 8100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.559234073013601e-05,
+      "loss": 0.2343,
+      "step": 8101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.559055118110236e-05,
+      "loss": 0.2522,
+      "step": 8102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.558876163206872e-05,
+      "loss": 0.1458,
+      "step": 8103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5586972083035074e-05,
+      "loss": 0.1888,
+      "step": 8104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5585182534001435e-05,
+      "loss": 0.1576,
+      "step": 8105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.558339298496779e-05,
+      "loss": 0.17,
+      "step": 8106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5581603435934146e-05,
+      "loss": 0.1975,
+      "step": 8107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.55798138869005e-05,
+      "loss": 0.2007,
+      "step": 8108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.557802433786686e-05,
+      "loss": 0.2134,
+      "step": 8109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.557623478883322e-05,
+      "loss": 0.1318,
+      "step": 8110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.557444523979957e-05,
+      "loss": 0.2185,
+      "step": 8111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.557265569076593e-05,
+      "loss": 0.2941,
+      "step": 8112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5570866141732284e-05,
+      "loss": 0.2194,
+      "step": 8113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.556907659269864e-05,
+      "loss": 0.1731,
+      "step": 8114
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5567287043665e-05,
+      "loss": 0.3159,
+      "step": 8115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5565497494631356e-05,
+      "loss": 0.2335,
+      "step": 8116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.556370794559771e-05,
+      "loss": 0.1882,
+      "step": 8117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.556191839656407e-05,
+      "loss": 0.1855,
+      "step": 8118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.556012884753042e-05,
+      "loss": 0.1594,
+      "step": 8119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.555833929849678e-05,
+      "loss": 0.1385,
+      "step": 8120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.555654974946314e-05,
+      "loss": 0.294,
+      "step": 8121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5554760200429494e-05,
+      "loss": 0.2888,
+      "step": 8122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.555297065139585e-05,
+      "loss": 0.2763,
+      "step": 8123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5551181102362204e-05,
+      "loss": 0.3938,
+      "step": 8124
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5549391553328566e-05,
+      "loss": 0.157,
+      "step": 8125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554760200429492e-05,
+      "loss": 0.2043,
+      "step": 8126
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554581245526128e-05,
+      "loss": 0.1975,
+      "step": 8127
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554402290622763e-05,
+      "loss": 0.1,
+      "step": 8128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554223335719399e-05,
+      "loss": 0.1256,
+      "step": 8129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.554044380816034e-05,
+      "loss": 0.1928,
+      "step": 8130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5538654259126704e-05,
+      "loss": 0.206,
+      "step": 8131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.553686471009306e-05,
+      "loss": 0.1451,
+      "step": 8132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5535075161059415e-05,
+      "loss": 0.1609,
+      "step": 8133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.553328561202577e-05,
+      "loss": 0.1321,
+      "step": 8134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5531496062992125e-05,
+      "loss": 0.2252,
+      "step": 8135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.552970651395848e-05,
+      "loss": 0.2458,
+      "step": 8136
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.552791696492484e-05,
+      "loss": 0.1366,
+      "step": 8137
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.55261274158912e-05,
+      "loss": 0.1618,
+      "step": 8138
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.552433786685755e-05,
+      "loss": 0.1958,
+      "step": 8139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5522548317823915e-05,
+      "loss": 0.1802,
+      "step": 8140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.552075876879027e-05,
+      "loss": 0.2338,
+      "step": 8141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.551896921975662e-05,
+      "loss": 0.1843,
+      "step": 8142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.551717967072298e-05,
+      "loss": 0.1468,
+      "step": 8143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5515390121689335e-05,
+      "loss": 0.328,
+      "step": 8144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.551360057265569e-05,
+      "loss": 0.1818,
+      "step": 8145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5511811023622046e-05,
+      "loss": 0.1883,
+      "step": 8146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.551002147458841e-05,
+      "loss": 0.2093,
+      "step": 8147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.550823192555476e-05,
+      "loss": 0.2715,
+      "step": 8148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.550644237652112e-05,
+      "loss": 0.1456,
+      "step": 8149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5504652827487473e-05,
+      "loss": 0.1855,
+      "step": 8150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.550286327845383e-05,
+      "loss": 0.1048,
+      "step": 8151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5501073729420184e-05,
+      "loss": 0.2695,
+      "step": 8152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5499284180386546e-05,
+      "loss": 0.2427,
+      "step": 8153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54974946313529e-05,
+      "loss": 0.3743,
+      "step": 8154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5495705082319256e-05,
+      "loss": 0.3211,
+      "step": 8155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549391553328562e-05,
+      "loss": 0.1009,
+      "step": 8156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549212598425197e-05,
+      "loss": 0.1751,
+      "step": 8157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549033643521832e-05,
+      "loss": 0.2071,
+      "step": 8158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5488546886184684e-05,
+      "loss": 0.2069,
+      "step": 8159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.548675733715104e-05,
+      "loss": 0.1437,
+      "step": 8160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5484967788117394e-05,
+      "loss": 0.1397,
+      "step": 8161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5483178239083756e-05,
+      "loss": 0.178,
+      "step": 8162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.548138869005011e-05,
+      "loss": 0.1322,
+      "step": 8163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5479599141016466e-05,
+      "loss": 0.2098,
+      "step": 8164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547780959198282e-05,
+      "loss": 0.2662,
+      "step": 8165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547602004294918e-05,
+      "loss": 0.2092,
+      "step": 8166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547423049391553e-05,
+      "loss": 0.2148,
+      "step": 8167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5472440944881894e-05,
+      "loss": 0.1666,
+      "step": 8168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547065139584825e-05,
+      "loss": 0.1212,
+      "step": 8169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5468861846814604e-05,
+      "loss": 0.2102,
+      "step": 8170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.546707229778096e-05,
+      "loss": 0.2123,
+      "step": 8171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.546528274874732e-05,
+      "loss": 0.2159,
+      "step": 8172
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.546349319971368e-05,
+      "loss": 0.1534,
+      "step": 8173
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5461703650680025e-05,
+      "loss": 0.1984,
+      "step": 8174
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.545991410164639e-05,
+      "loss": 0.1799,
+      "step": 8175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.545812455261274e-05,
+      "loss": 0.1754,
+      "step": 8176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54563350035791e-05,
+      "loss": 0.3243,
+      "step": 8177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.545454545454546e-05,
+      "loss": 0.2091,
+      "step": 8178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5452755905511815e-05,
+      "loss": 0.218,
+      "step": 8179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.545096635647817e-05,
+      "loss": 0.213,
+      "step": 8180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5449176807444525e-05,
+      "loss": 0.1272,
+      "step": 8181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.544738725841088e-05,
+      "loss": 0.065,
+      "step": 8182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5445597709377235e-05,
+      "loss": 0.1196,
+      "step": 8183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54438081603436e-05,
+      "loss": 0.1472,
+      "step": 8184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.544201861130995e-05,
+      "loss": 0.2199,
+      "step": 8185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.544022906227631e-05,
+      "loss": 0.2558,
+      "step": 8186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543843951324266e-05,
+      "loss": 0.1387,
+      "step": 8187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5436649964209025e-05,
+      "loss": 0.1664,
+      "step": 8188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543486041517537e-05,
+      "loss": 0.2013,
+      "step": 8189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5433070866141735e-05,
+      "loss": 0.3265,
+      "step": 8190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543128131710809e-05,
+      "loss": 0.1138,
+      "step": 8191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5429491768074446e-05,
+      "loss": 0.2242,
+      "step": 8192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54277022190408e-05,
+      "loss": 0.1882,
+      "step": 8193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542591267000716e-05,
+      "loss": 0.1628,
+      "step": 8194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542412312097352e-05,
+      "loss": 0.1421,
+      "step": 8195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542233357193987e-05,
+      "loss": 0.2374,
+      "step": 8196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.542054402290623e-05,
+      "loss": 0.2391,
+      "step": 8197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5418754473872584e-05,
+      "loss": 0.2224,
+      "step": 8198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.541696492483894e-05,
+      "loss": 0.1643,
+      "step": 8199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.54151753758053e-05,
+      "loss": 0.2414,
+      "step": 8200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5413385826771656e-05,
+      "loss": 0.1825,
+      "step": 8201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.541159627773801e-05,
+      "loss": 0.1945,
+      "step": 8202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540980672870437e-05,
+      "loss": 0.167,
+      "step": 8203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540801717967073e-05,
+      "loss": 0.3289,
+      "step": 8204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540622763063708e-05,
+      "loss": 0.1254,
+      "step": 8205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540443808160344e-05,
+      "loss": 0.1355,
+      "step": 8206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5402648532569794e-05,
+      "loss": 0.1955,
+      "step": 8207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540085898353615e-05,
+      "loss": 0.2187,
+      "step": 8208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5399069434502504e-05,
+      "loss": 0.1468,
+      "step": 8209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5397279885468866e-05,
+      "loss": 0.2134,
+      "step": 8210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.539549033643522e-05,
+      "loss": 0.2555,
+      "step": 8211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.539370078740158e-05,
+      "loss": 0.2493,
+      "step": 8212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.539191123836793e-05,
+      "loss": 0.2916,
+      "step": 8213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.539012168933429e-05,
+      "loss": 0.2553,
+      "step": 8214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.538833214030064e-05,
+      "loss": 0.25,
+      "step": 8215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5386542591267004e-05,
+      "loss": 0.2377,
+      "step": 8216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.538475304223336e-05,
+      "loss": 0.2649,
+      "step": 8217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5382963493199715e-05,
+      "loss": 0.1912,
+      "step": 8218
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5381173944166077e-05,
+      "loss": 0.1696,
+      "step": 8219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537938439513243e-05,
+      "loss": 0.1901,
+      "step": 8220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537759484609878e-05,
+      "loss": 0.1925,
+      "step": 8221
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537580529706514e-05,
+      "loss": 0.2258,
+      "step": 8222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.53740157480315e-05,
+      "loss": 0.2698,
+      "step": 8223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537222619899785e-05,
+      "loss": 0.1395,
+      "step": 8224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5370436649964215e-05,
+      "loss": 0.2275,
+      "step": 8225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.536864710093057e-05,
+      "loss": 0.141,
+      "step": 8226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5366857551896925e-05,
+      "loss": 0.2828,
+      "step": 8227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.536506800286328e-05,
+      "loss": 0.1086,
+      "step": 8228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5363278453829635e-05,
+      "loss": 0.1406,
+      "step": 8229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.536148890479599e-05,
+      "loss": 0.1417,
+      "step": 8230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5359699355762346e-05,
+      "loss": 0.1948,
+      "step": 8231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.535790980672871e-05,
+      "loss": 0.2005,
+      "step": 8232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.535612025769506e-05,
+      "loss": 0.1927,
+      "step": 8233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.535433070866142e-05,
+      "loss": 0.2356,
+      "step": 8234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.535254115962778e-05,
+      "loss": 0.1466,
+      "step": 8235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.535075161059413e-05,
+      "loss": 0.1693,
+      "step": 8236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5348962061560484e-05,
+      "loss": 0.1475,
+      "step": 8237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5347172512526846e-05,
+      "loss": 0.1927,
+      "step": 8238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.53453829634932e-05,
+      "loss": 0.134,
+      "step": 8239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5343593414459556e-05,
+      "loss": 0.1762,
+      "step": 8240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.534180386542592e-05,
+      "loss": 0.198,
+      "step": 8241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.534001431639227e-05,
+      "loss": 0.1948,
+      "step": 8242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.533822476735863e-05,
+      "loss": 0.1919,
+      "step": 8243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5336435218324983e-05,
+      "loss": 0.1707,
+      "step": 8244
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.533464566929134e-05,
+      "loss": 0.18,
+      "step": 8245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5332856120257694e-05,
+      "loss": 0.2495,
+      "step": 8246
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5331066571224056e-05,
+      "loss": 0.1372,
+      "step": 8247
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.532927702219041e-05,
+      "loss": 0.2232,
+      "step": 8248
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5327487473156766e-05,
+      "loss": 0.257,
+      "step": 8249
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.532569792412312e-05,
+      "loss": 0.1156,
+      "step": 8250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5323908375089483e-05,
+      "loss": 0.1272,
+      "step": 8251
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.532211882605583e-05,
+      "loss": 0.2548,
+      "step": 8252
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5320329277022194e-05,
+      "loss": 0.2601,
+      "step": 8253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.531853972798855e-05,
+      "loss": 0.243,
+      "step": 8254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5316750178954904e-05,
+      "loss": 0.2804,
+      "step": 8255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.531496062992126e-05,
+      "loss": 0.1573,
+      "step": 8256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.531317108088762e-05,
+      "loss": 0.1031,
+      "step": 8257
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5311381531853977e-05,
+      "loss": 0.2095,
+      "step": 8258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530959198282033e-05,
+      "loss": 0.14,
+      "step": 8259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530780243378669e-05,
+      "loss": 0.3194,
+      "step": 8260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530601288475304e-05,
+      "loss": 0.1795,
+      "step": 8261
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.53042233357194e-05,
+      "loss": 0.2048,
+      "step": 8262
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530243378668576e-05,
+      "loss": 0.1452,
+      "step": 8263
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5300644237652114e-05,
+      "loss": 0.3759,
+      "step": 8264
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.529885468861847e-05,
+      "loss": 0.0924,
+      "step": 8265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.529706513958483e-05,
+      "loss": 0.2244,
+      "step": 8266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.529527559055118e-05,
+      "loss": 0.2857,
+      "step": 8267
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5293486041517535e-05,
+      "loss": 0.2382,
+      "step": 8268
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.52916964924839e-05,
+      "loss": 0.2618,
+      "step": 8269
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528990694345025e-05,
+      "loss": 0.3166,
+      "step": 8270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528811739441661e-05,
+      "loss": 0.2662,
+      "step": 8271
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528632784538296e-05,
+      "loss": 0.1828,
+      "step": 8272
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5284538296349325e-05,
+      "loss": 0.1516,
+      "step": 8273
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528274874731568e-05,
+      "loss": 0.2191,
+      "step": 8274
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5280959198282035e-05,
+      "loss": 0.1254,
+      "step": 8275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.527916964924839e-05,
+      "loss": 0.2053,
+      "step": 8276
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5277380100214746e-05,
+      "loss": 0.1858,
+      "step": 8277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.52755905511811e-05,
+      "loss": 0.1444,
+      "step": 8278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.527380100214746e-05,
+      "loss": 0.1976,
+      "step": 8279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.527201145311382e-05,
+      "loss": 0.2701,
+      "step": 8280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.527022190408017e-05,
+      "loss": 0.2859,
+      "step": 8281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5268432355046535e-05,
+      "loss": 0.0677,
+      "step": 8282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5266642806012883e-05,
+      "loss": 0.2004,
+      "step": 8283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.526485325697924e-05,
+      "loss": 0.3208,
+      "step": 8284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.52630637079456e-05,
+      "loss": 0.2333,
+      "step": 8285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5261274158911956e-05,
+      "loss": 0.1168,
+      "step": 8286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.525948460987831e-05,
+      "loss": 0.1466,
+      "step": 8287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.525769506084467e-05,
+      "loss": 0.1534,
+      "step": 8288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.525590551181103e-05,
+      "loss": 0.1703,
+      "step": 8289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.525411596277738e-05,
+      "loss": 0.1421,
+      "step": 8290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.525232641374374e-05,
+      "loss": 0.2257,
+      "step": 8291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5250536864710094e-05,
+      "loss": 0.1187,
+      "step": 8292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.524874731567645e-05,
+      "loss": 0.2087,
+      "step": 8293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5246957766642804e-05,
+      "loss": 0.2307,
+      "step": 8294
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5245168217609166e-05,
+      "loss": 0.1123,
+      "step": 8295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.524337866857552e-05,
+      "loss": 0.1721,
+      "step": 8296
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5241589119541876e-05,
+      "loss": 0.2217,
+      "step": 8297
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.523979957050824e-05,
+      "loss": 0.1136,
+      "step": 8298
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.523801002147459e-05,
+      "loss": 0.2011,
+      "step": 8299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.523622047244094e-05,
+      "loss": 0.105,
+      "step": 8300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5234430923407304e-05,
+      "loss": 0.1672,
+      "step": 8301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.523264137437366e-05,
+      "loss": 0.1042,
+      "step": 8302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5230851825340014e-05,
+      "loss": 0.2023,
+      "step": 8303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5229062276306376e-05,
+      "loss": 0.1588,
+      "step": 8304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.522727272727273e-05,
+      "loss": 0.1358,
+      "step": 8305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.522548317823909e-05,
+      "loss": 0.1055,
+      "step": 8306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.522369362920544e-05,
+      "loss": 0.2594,
+      "step": 8307
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.52219040801718e-05,
+      "loss": 0.1891,
+      "step": 8308
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.522011453113815e-05,
+      "loss": 0.1953,
+      "step": 8309
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5218324982104514e-05,
+      "loss": 0.2295,
+      "step": 8310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.521653543307087e-05,
+      "loss": 0.1776,
+      "step": 8311
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5214745884037225e-05,
+      "loss": 0.1278,
+      "step": 8312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.521295633500358e-05,
+      "loss": 0.2218,
+      "step": 8313
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5211166785969935e-05,
+      "loss": 0.2603,
+      "step": 8314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520937723693629e-05,
+      "loss": 0.0932,
+      "step": 8315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520758768790265e-05,
+      "loss": 0.2073,
+      "step": 8316
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520579813886901e-05,
+      "loss": 0.1881,
+      "step": 8317
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520400858983536e-05,
+      "loss": 0.147,
+      "step": 8318
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520221904080172e-05,
+      "loss": 0.2035,
+      "step": 8319
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520042949176808e-05,
+      "loss": 0.2304,
+      "step": 8320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5198639942734435e-05,
+      "loss": 0.2518,
+      "step": 8321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5196850393700783e-05,
+      "loss": 0.1084,
+      "step": 8322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5195060844667145e-05,
+      "loss": 0.1326,
+      "step": 8323
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51932712956335e-05,
+      "loss": 0.2581,
+      "step": 8324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5191481746599856e-05,
+      "loss": 0.0939,
+      "step": 8325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518969219756622e-05,
+      "loss": 0.2237,
+      "step": 8326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518790264853257e-05,
+      "loss": 0.3043,
+      "step": 8327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518611309949893e-05,
+      "loss": 0.2281,
+      "step": 8328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518432355046529e-05,
+      "loss": 0.1706,
+      "step": 8329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518253400143164e-05,
+      "loss": 0.0972,
+      "step": 8330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5180744452397994e-05,
+      "loss": 0.229,
+      "step": 8331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5178954903364356e-05,
+      "loss": 0.1158,
+      "step": 8332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517716535433071e-05,
+      "loss": 0.1655,
+      "step": 8333
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5175375805297066e-05,
+      "loss": 0.1553,
+      "step": 8334
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517358625626342e-05,
+      "loss": 0.331,
+      "step": 8335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517179670722978e-05,
+      "loss": 0.2287,
+      "step": 8336
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517000715819614e-05,
+      "loss": 0.1695,
+      "step": 8337
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5168217609162494e-05,
+      "loss": 0.2044,
+      "step": 8338
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.516642806012885e-05,
+      "loss": 0.25,
+      "step": 8339
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5164638511095204e-05,
+      "loss": 0.2007,
+      "step": 8340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.516284896206156e-05,
+      "loss": 0.236,
+      "step": 8341
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.516105941302792e-05,
+      "loss": 0.1208,
+      "step": 8342
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5159269863994276e-05,
+      "loss": 0.1779,
+      "step": 8343
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.515748031496063e-05,
+      "loss": 0.1853,
+      "step": 8344
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5155690765926994e-05,
+      "loss": 0.2813,
+      "step": 8345
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.515390121689334e-05,
+      "loss": 0.2278,
+      "step": 8346
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51521116678597e-05,
+      "loss": 0.2738,
+      "step": 8347
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.515032211882606e-05,
+      "loss": 0.1356,
+      "step": 8348
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5148532569792414e-05,
+      "loss": 0.2621,
+      "step": 8349
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.514674302075877e-05,
+      "loss": 0.3086,
+      "step": 8350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.514495347172513e-05,
+      "loss": 0.1181,
+      "step": 8351
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.514316392269149e-05,
+      "loss": 0.4053,
+      "step": 8352
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5141374373657835e-05,
+      "loss": 0.3259,
+      "step": 8353
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51395848246242e-05,
+      "loss": 0.163,
+      "step": 8354
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.513779527559055e-05,
+      "loss": 0.2001,
+      "step": 8355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.513600572655691e-05,
+      "loss": 0.1743,
+      "step": 8356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.513421617752326e-05,
+      "loss": 0.2103,
+      "step": 8357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5132426628489625e-05,
+      "loss": 0.2922,
+      "step": 8358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.513063707945598e-05,
+      "loss": 0.1413,
+      "step": 8359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5128847530422335e-05,
+      "loss": 0.3308,
+      "step": 8360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.512705798138869e-05,
+      "loss": 0.2124,
+      "step": 8361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5125268432355045e-05,
+      "loss": 0.1774,
+      "step": 8362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51234788833214e-05,
+      "loss": 0.1372,
+      "step": 8363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.512168933428776e-05,
+      "loss": 0.1646,
+      "step": 8364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.511989978525412e-05,
+      "loss": 0.1226,
+      "step": 8365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.511811023622047e-05,
+      "loss": 0.2424,
+      "step": 8366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5116320687186835e-05,
+      "loss": 0.1833,
+      "step": 8367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.511453113815319e-05,
+      "loss": 0.0975,
+      "step": 8368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.511274158911954e-05,
+      "loss": 0.1929,
+      "step": 8369
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.51109520400859e-05,
+      "loss": 0.2449,
+      "step": 8370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5109162491052256e-05,
+      "loss": 0.2432,
+      "step": 8371
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510737294201861e-05,
+      "loss": 0.2901,
+      "step": 8372
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510558339298497e-05,
+      "loss": 0.1756,
+      "step": 8373
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510379384395133e-05,
+      "loss": 0.1193,
+      "step": 8374
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510200429491768e-05,
+      "loss": 0.2538,
+      "step": 8375
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510021474588404e-05,
+      "loss": 0.1791,
+      "step": 8376
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5098425196850394e-05,
+      "loss": 0.1527,
+      "step": 8377
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.509663564781675e-05,
+      "loss": 0.0904,
+      "step": 8378
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.509484609878311e-05,
+      "loss": 0.2673,
+      "step": 8379
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5093056549749466e-05,
+      "loss": 0.1498,
+      "step": 8380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.509126700071582e-05,
+      "loss": 0.2312,
+      "step": 8381
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5089477451682176e-05,
+      "loss": 0.1136,
+      "step": 8382
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.508768790264854e-05,
+      "loss": 0.2083,
+      "step": 8383
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5085898353614893e-05,
+      "loss": 0.1534,
+      "step": 8384
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.508410880458124e-05,
+      "loss": 0.1711,
+      "step": 8385
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5082319255547604e-05,
+      "loss": 0.2839,
+      "step": 8386
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.508052970651396e-05,
+      "loss": 0.1528,
+      "step": 8387
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5078740157480314e-05,
+      "loss": 0.1743,
+      "step": 8388
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5076950608446676e-05,
+      "loss": 0.2198,
+      "step": 8389
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.507516105941303e-05,
+      "loss": 0.2361,
+      "step": 8390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5073371510379387e-05,
+      "loss": 0.3079,
+      "step": 8391
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.507158196134575e-05,
+      "loss": 0.1938,
+      "step": 8392
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.50697924123121e-05,
+      "loss": 0.1578,
+      "step": 8393
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.506800286327845e-05,
+      "loss": 0.1777,
+      "step": 8394
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5066213314244814e-05,
+      "loss": 0.2513,
+      "step": 8395
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.506442376521117e-05,
+      "loss": 0.19,
+      "step": 8396
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5062634216177525e-05,
+      "loss": 0.3425,
+      "step": 8397
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.506084466714388e-05,
+      "loss": 0.1915,
+      "step": 8398
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505905511811024e-05,
+      "loss": 0.0876,
+      "step": 8399
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505726556907659e-05,
+      "loss": 0.2027,
+      "step": 8400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505547602004295e-05,
+      "loss": 0.103,
+      "step": 8401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505368647100931e-05,
+      "loss": 0.1687,
+      "step": 8402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505189692197566e-05,
+      "loss": 0.1728,
+      "step": 8403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.505010737294202e-05,
+      "loss": 0.2571,
+      "step": 8404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.504831782390838e-05,
+      "loss": 0.1816,
+      "step": 8405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5046528274874735e-05,
+      "loss": 0.1748,
+      "step": 8406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.504473872584109e-05,
+      "loss": 0.1233,
+      "step": 8407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5042949176807445e-05,
+      "loss": 0.1061,
+      "step": 8408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.50411596277738e-05,
+      "loss": 0.1976,
+      "step": 8409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5039370078740156e-05,
+      "loss": 0.3097,
+      "step": 8410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.503758052970652e-05,
+      "loss": 0.2172,
+      "step": 8411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.503579098067287e-05,
+      "loss": 0.2337,
+      "step": 8412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.503400143163923e-05,
+      "loss": 0.1344,
+      "step": 8413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.503221188260559e-05,
+      "loss": 0.1179,
+      "step": 8414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5030422333571945e-05,
+      "loss": 0.1721,
+      "step": 8415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5028632784538294e-05,
+      "loss": 0.1761,
+      "step": 8416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5026843235504655e-05,
+      "loss": 0.1174,
+      "step": 8417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.502505368647101e-05,
+      "loss": 0.3184,
+      "step": 8418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5023264137437366e-05,
+      "loss": 0.1982,
+      "step": 8419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.502147458840372e-05,
+      "loss": 0.2018,
+      "step": 8420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.501968503937008e-05,
+      "loss": 0.1819,
+      "step": 8421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.501789549033644e-05,
+      "loss": 0.1519,
+      "step": 8422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5016105941302793e-05,
+      "loss": 0.2541,
+      "step": 8423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.501431639226915e-05,
+      "loss": 0.2279,
+      "step": 8424
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5012526843235504e-05,
+      "loss": 0.2087,
+      "step": 8425
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.501073729420186e-05,
+      "loss": 0.1475,
+      "step": 8426
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.500894774516822e-05,
+      "loss": 0.2724,
+      "step": 8427
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5007158196134576e-05,
+      "loss": 0.156,
+      "step": 8428
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.500536864710093e-05,
+      "loss": 0.333,
+      "step": 8429
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.500357909806729e-05,
+      "loss": 0.1925,
+      "step": 8430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.500178954903365e-05,
+      "loss": 0.295,
+      "step": 8431
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5e-05,
+      "loss": 0.1289,
+      "step": 8432
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499821045096636e-05,
+      "loss": 0.1498,
+      "step": 8433
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4996420901932714e-05,
+      "loss": 0.207,
+      "step": 8434
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499463135289907e-05,
+      "loss": 0.2201,
+      "step": 8435
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499284180386543e-05,
+      "loss": 0.1498,
+      "step": 8436
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4991052254831786e-05,
+      "loss": 0.1697,
+      "step": 8437
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.498926270579814e-05,
+      "loss": 0.225,
+      "step": 8438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.49874731567645e-05,
+      "loss": 0.2713,
+      "step": 8439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.498568360773085e-05,
+      "loss": 0.2058,
+      "step": 8440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.498389405869721e-05,
+      "loss": 0.1925,
+      "step": 8441
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.498210450966356e-05,
+      "loss": 0.1147,
+      "step": 8442
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4980314960629924e-05,
+      "loss": 0.1529,
+      "step": 8443
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.497852541159628e-05,
+      "loss": 0.158,
+      "step": 8444
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4976735862562635e-05,
+      "loss": 0.2599,
+      "step": 8445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4974946313529e-05,
+      "loss": 0.327,
+      "step": 8446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4973156764495345e-05,
+      "loss": 0.2512,
+      "step": 8447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.49713672154617e-05,
+      "loss": 0.2834,
+      "step": 8448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.496957766642806e-05,
+      "loss": 0.2581,
+      "step": 8449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.496778811739442e-05,
+      "loss": 0.1842,
+      "step": 8450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.496599856836077e-05,
+      "loss": 0.1566,
+      "step": 8451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4964209019327135e-05,
+      "loss": 0.17,
+      "step": 8452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.496241947029349e-05,
+      "loss": 0.2287,
+      "step": 8453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4960629921259845e-05,
+      "loss": 0.1735,
+      "step": 8454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.49588403722262e-05,
+      "loss": 0.1616,
+      "step": 8455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4957050823192555e-05,
+      "loss": 0.2701,
+      "step": 8456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.495526127415891e-05,
+      "loss": 0.2686,
+      "step": 8457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.495347172512527e-05,
+      "loss": 0.3209,
+      "step": 8458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.495168217609163e-05,
+      "loss": 0.193,
+      "step": 8459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.494989262705798e-05,
+      "loss": 0.219,
+      "step": 8460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.494810307802434e-05,
+      "loss": 0.1432,
+      "step": 8461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.49463135289907e-05,
+      "loss": 0.235,
+      "step": 8462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.494452397995705e-05,
+      "loss": 0.2323,
+      "step": 8463
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.494273443092341e-05,
+      "loss": 0.2109,
+      "step": 8464
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4940944881889766e-05,
+      "loss": 0.1617,
+      "step": 8465
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.493915533285612e-05,
+      "loss": 0.1942,
+      "step": 8466
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4937365783822476e-05,
+      "loss": 0.238,
+      "step": 8467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.493557623478884e-05,
+      "loss": 0.1179,
+      "step": 8468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.493378668575519e-05,
+      "loss": 0.1899,
+      "step": 8469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.493199713672155e-05,
+      "loss": 0.1895,
+      "step": 8470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4930207587687904e-05,
+      "loss": 0.2656,
+      "step": 8471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.492841803865426e-05,
+      "loss": 0.1448,
+      "step": 8472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4926628489620614e-05,
+      "loss": 0.37,
+      "step": 8473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4924838940586976e-05,
+      "loss": 0.0841,
+      "step": 8474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.492304939155333e-05,
+      "loss": 0.1601,
+      "step": 8475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4921259842519686e-05,
+      "loss": 0.1884,
+      "step": 8476
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.491947029348605e-05,
+      "loss": 0.3934,
+      "step": 8477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4917680744452404e-05,
+      "loss": 0.2516,
+      "step": 8478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.491589119541875e-05,
+      "loss": 0.1786,
+      "step": 8479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4914101646385114e-05,
+      "loss": 0.3692,
+      "step": 8480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.491231209735147e-05,
+      "loss": 0.2234,
+      "step": 8481
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4910522548317824e-05,
+      "loss": 0.2594,
+      "step": 8482
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.490873299928418e-05,
+      "loss": 0.156,
+      "step": 8483
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.490694345025054e-05,
+      "loss": 0.2178,
+      "step": 8484
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.49051539012169e-05,
+      "loss": 0.2465,
+      "step": 8485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.490336435218325e-05,
+      "loss": 0.1957,
+      "step": 8486
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.490157480314961e-05,
+      "loss": 0.1848,
+      "step": 8487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489978525411596e-05,
+      "loss": 0.1173,
+      "step": 8488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489799570508232e-05,
+      "loss": 0.1944,
+      "step": 8489
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489620615604868e-05,
+      "loss": 0.0724,
+      "step": 8490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4894416607015035e-05,
+      "loss": 0.1328,
+      "step": 8491
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489262705798139e-05,
+      "loss": 0.2441,
+      "step": 8492
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489083750894775e-05,
+      "loss": 0.2118,
+      "step": 8493
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.48890479599141e-05,
+      "loss": 0.2232,
+      "step": 8494
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4887258410880455e-05,
+      "loss": 0.1787,
+      "step": 8495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488546886184682e-05,
+      "loss": 0.1655,
+      "step": 8496
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488367931281317e-05,
+      "loss": 0.2739,
+      "step": 8497
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488188976377953e-05,
+      "loss": 0.4013,
+      "step": 8498
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488010021474589e-05,
+      "loss": 0.1678,
+      "step": 8499
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4878310665712245e-05,
+      "loss": 0.1221,
+      "step": 8500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.48765211166786e-05,
+      "loss": 0.1894,
+      "step": 8501
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4874731567644955e-05,
+      "loss": 0.1702,
+      "step": 8502
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.487294201861131e-05,
+      "loss": 0.227,
+      "step": 8503
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4871152469577666e-05,
+      "loss": 0.3024,
+      "step": 8504
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.486936292054402e-05,
+      "loss": 0.1518,
+      "step": 8505
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.486757337151038e-05,
+      "loss": 0.2037,
+      "step": 8506
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.486578382247674e-05,
+      "loss": 0.1577,
+      "step": 8507
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.486399427344309e-05,
+      "loss": 0.1205,
+      "step": 8508
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4862204724409455e-05,
+      "loss": 0.1592,
+      "step": 8509
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4860415175375804e-05,
+      "loss": 0.3696,
+      "step": 8510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.485862562634216e-05,
+      "loss": 0.2696,
+      "step": 8511
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.485683607730852e-05,
+      "loss": 0.2143,
+      "step": 8512
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4855046528274876e-05,
+      "loss": 0.1784,
+      "step": 8513
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.485325697924123e-05,
+      "loss": 0.2103,
+      "step": 8514
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.485146743020759e-05,
+      "loss": 0.2517,
+      "step": 8515
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.484967788117395e-05,
+      "loss": 0.092,
+      "step": 8516
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4847888332140304e-05,
+      "loss": 0.2323,
+      "step": 8517
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.484609878310666e-05,
+      "loss": 0.192,
+      "step": 8518
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4844309234073014e-05,
+      "loss": 0.0847,
+      "step": 8519
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.484251968503937e-05,
+      "loss": 0.2219,
+      "step": 8520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.484073013600573e-05,
+      "loss": 0.2526,
+      "step": 8521
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4838940586972086e-05,
+      "loss": 0.1562,
+      "step": 8522
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.483715103793844e-05,
+      "loss": 0.1942,
+      "step": 8523
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.48353614889048e-05,
+      "loss": 0.2878,
+      "step": 8524
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.483357193987116e-05,
+      "loss": 0.1819,
+      "step": 8525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.483178239083751e-05,
+      "loss": 0.2389,
+      "step": 8526
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.482999284180387e-05,
+      "loss": 0.1447,
+      "step": 8527
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4828203292770224e-05,
+      "loss": 0.1729,
+      "step": 8528
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.482641374373658e-05,
+      "loss": 0.0973,
+      "step": 8529
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4824624194702935e-05,
+      "loss": 0.2632,
+      "step": 8530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4822834645669297e-05,
+      "loss": 0.1827,
+      "step": 8531
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.482104509663565e-05,
+      "loss": 0.1945,
+      "step": 8532
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4819255547602e-05,
+      "loss": 0.2818,
+      "step": 8533
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.481746599856836e-05,
+      "loss": 0.2188,
+      "step": 8534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.481567644953472e-05,
+      "loss": 0.1374,
+      "step": 8535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.481388690050107e-05,
+      "loss": 0.1354,
+      "step": 8536
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4812097351467434e-05,
+      "loss": 0.212,
+      "step": 8537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.481030780243379e-05,
+      "loss": 0.1692,
+      "step": 8538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4808518253400145e-05,
+      "loss": 0.1107,
+      "step": 8539
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.480672870436651e-05,
+      "loss": 0.2419,
+      "step": 8540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4804939155332855e-05,
+      "loss": 0.1824,
+      "step": 8541
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.480314960629921e-05,
+      "loss": 0.2619,
+      "step": 8542
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.480136005726557e-05,
+      "loss": 0.1771,
+      "step": 8543
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479957050823193e-05,
+      "loss": 0.1441,
+      "step": 8544
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479778095919828e-05,
+      "loss": 0.268,
+      "step": 8545
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479599141016464e-05,
+      "loss": 0.2059,
+      "step": 8546
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4794201861131e-05,
+      "loss": 0.1554,
+      "step": 8547
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4792412312097355e-05,
+      "loss": 0.4182,
+      "step": 8548
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479062276306371e-05,
+      "loss": 0.1016,
+      "step": 8549
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4788833214030066e-05,
+      "loss": 0.1684,
+      "step": 8550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.478704366499642e-05,
+      "loss": 0.3374,
+      "step": 8551
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4785254115962776e-05,
+      "loss": 0.203,
+      "step": 8552
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.478346456692914e-05,
+      "loss": 0.1459,
+      "step": 8553
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.478167501789549e-05,
+      "loss": 0.3014,
+      "step": 8554
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477988546886185e-05,
+      "loss": 0.1162,
+      "step": 8555
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477809591982821e-05,
+      "loss": 0.1874,
+      "step": 8556
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477630637079456e-05,
+      "loss": 0.3592,
+      "step": 8557
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4774516821760914e-05,
+      "loss": 0.1582,
+      "step": 8558
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4772727272727276e-05,
+      "loss": 0.1858,
+      "step": 8559
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477093772369363e-05,
+      "loss": 0.1404,
+      "step": 8560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4769148174659986e-05,
+      "loss": 0.0722,
+      "step": 8561
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.476735862562635e-05,
+      "loss": 0.1872,
+      "step": 8562
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4765569076592703e-05,
+      "loss": 0.2903,
+      "step": 8563
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.476377952755906e-05,
+      "loss": 0.1475,
+      "step": 8564
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4761989978525414e-05,
+      "loss": 0.2918,
+      "step": 8565
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.476020042949177e-05,
+      "loss": 0.1587,
+      "step": 8566
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4758410880458124e-05,
+      "loss": 0.105,
+      "step": 8567
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.475662133142448e-05,
+      "loss": 0.1495,
+      "step": 8568
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.475483178239084e-05,
+      "loss": 0.1766,
+      "step": 8569
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4753042233357197e-05,
+      "loss": 0.1985,
+      "step": 8570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.475125268432355e-05,
+      "loss": 0.1312,
+      "step": 8571
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4749463135289914e-05,
+      "loss": 0.1289,
+      "step": 8572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.474767358625626e-05,
+      "loss": 0.2524,
+      "step": 8573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.474588403722262e-05,
+      "loss": 0.2482,
+      "step": 8574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.474409448818898e-05,
+      "loss": 0.2918,
+      "step": 8575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4742304939155334e-05,
+      "loss": 0.2315,
+      "step": 8576
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.474051539012169e-05,
+      "loss": 0.1209,
+      "step": 8577
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.473872584108805e-05,
+      "loss": 0.1871,
+      "step": 8578
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.473693629205441e-05,
+      "loss": 0.1216,
+      "step": 8579
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4735146743020755e-05,
+      "loss": 0.1615,
+      "step": 8580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.473335719398712e-05,
+      "loss": 0.1631,
+      "step": 8581
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.473156764495347e-05,
+      "loss": 0.1466,
+      "step": 8582
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.472977809591983e-05,
+      "loss": 0.1569,
+      "step": 8583
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.472798854688619e-05,
+      "loss": 0.1373,
+      "step": 8584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4726198997852545e-05,
+      "loss": 0.2571,
+      "step": 8585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.47244094488189e-05,
+      "loss": 0.1683,
+      "step": 8586
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4722619899785255e-05,
+      "loss": 0.1979,
+      "step": 8587
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.472083035075161e-05,
+      "loss": 0.2391,
+      "step": 8588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4719040801717966e-05,
+      "loss": 0.2389,
+      "step": 8589
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.471725125268433e-05,
+      "loss": 0.1502,
+      "step": 8590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.471546170365068e-05,
+      "loss": 0.1449,
+      "step": 8591
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.471367215461704e-05,
+      "loss": 0.1711,
+      "step": 8592
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.471188260558339e-05,
+      "loss": 0.1267,
+      "step": 8593
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4710093056549755e-05,
+      "loss": 0.1027,
+      "step": 8594
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.470830350751611e-05,
+      "loss": 0.3706,
+      "step": 8595
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.470651395848246e-05,
+      "loss": 0.156,
+      "step": 8596
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.470472440944882e-05,
+      "loss": 0.2448,
+      "step": 8597
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4702934860415176e-05,
+      "loss": 0.2075,
+      "step": 8598
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.470114531138153e-05,
+      "loss": 0.2223,
+      "step": 8599
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469935576234789e-05,
+      "loss": 0.1947,
+      "step": 8600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469756621331425e-05,
+      "loss": 0.1691,
+      "step": 8601
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.46957766642806e-05,
+      "loss": 0.1625,
+      "step": 8602
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4693987115246965e-05,
+      "loss": 0.1852,
+      "step": 8603
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4692197566213314e-05,
+      "loss": 0.1889,
+      "step": 8604
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469040801717967e-05,
+      "loss": 0.2207,
+      "step": 8605
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.468861846814603e-05,
+      "loss": 0.1605,
+      "step": 8606
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4686828919112386e-05,
+      "loss": 0.0743,
+      "step": 8607
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.468503937007874e-05,
+      "loss": 0.1987,
+      "step": 8608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4683249821045096e-05,
+      "loss": 0.0852,
+      "step": 8609
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.468146027201146e-05,
+      "loss": 0.1646,
+      "step": 8610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4679670722977814e-05,
+      "loss": 0.0847,
+      "step": 8611
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.467788117394417e-05,
+      "loss": 0.1511,
+      "step": 8612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4676091624910524e-05,
+      "loss": 0.1895,
+      "step": 8613
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.467430207587688e-05,
+      "loss": 0.0915,
+      "step": 8614
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4672512526843234e-05,
+      "loss": 0.1721,
+      "step": 8615
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4670722977809596e-05,
+      "loss": 0.2065,
+      "step": 8616
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.466893342877595e-05,
+      "loss": 0.1308,
+      "step": 8617
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.466714387974231e-05,
+      "loss": 0.1198,
+      "step": 8618
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.466535433070866e-05,
+      "loss": 0.137,
+      "step": 8619
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.466356478167502e-05,
+      "loss": 0.2825,
+      "step": 8620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.466177523264137e-05,
+      "loss": 0.2308,
+      "step": 8621
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4659985683607734e-05,
+      "loss": 0.3719,
+      "step": 8622
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.465819613457409e-05,
+      "loss": 0.2713,
+      "step": 8623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4656406585540445e-05,
+      "loss": 0.2249,
+      "step": 8624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.465461703650681e-05,
+      "loss": 0.2711,
+      "step": 8625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.465282748747316e-05,
+      "loss": 0.3272,
+      "step": 8626
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.465103793843951e-05,
+      "loss": 0.1998,
+      "step": 8627
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.464924838940587e-05,
+      "loss": 0.2536,
+      "step": 8628
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.464745884037223e-05,
+      "loss": 0.0939,
+      "step": 8629
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.464566929133858e-05,
+      "loss": 0.0906,
+      "step": 8630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.464387974230494e-05,
+      "loss": 0.1817,
+      "step": 8631
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.46420901932713e-05,
+      "loss": 0.1491,
+      "step": 8632
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4640300644237655e-05,
+      "loss": 0.1383,
+      "step": 8633
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.463851109520401e-05,
+      "loss": 0.1655,
+      "step": 8634
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4636721546170365e-05,
+      "loss": 0.1303,
+      "step": 8635
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.463493199713672e-05,
+      "loss": 0.2153,
+      "step": 8636
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4633142448103076e-05,
+      "loss": 0.2582,
+      "step": 8637
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.463135289906944e-05,
+      "loss": 0.1453,
+      "step": 8638
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.462956335003579e-05,
+      "loss": 0.1709,
+      "step": 8639
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.462777380100215e-05,
+      "loss": 0.272,
+      "step": 8640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.462598425196851e-05,
+      "loss": 0.0924,
+      "step": 8641
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4624194702934865e-05,
+      "loss": 0.1972,
+      "step": 8642
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4622405153901214e-05,
+      "loss": 0.3699,
+      "step": 8643
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4620615604867576e-05,
+      "loss": 0.1623,
+      "step": 8644
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.461882605583393e-05,
+      "loss": 0.0908,
+      "step": 8645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4617036506800286e-05,
+      "loss": 0.2276,
+      "step": 8646
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.461524695776665e-05,
+      "loss": 0.1764,
+      "step": 8647
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4613457408733e-05,
+      "loss": 0.2286,
+      "step": 8648
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.461166785969936e-05,
+      "loss": 0.2149,
+      "step": 8649
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4609878310665714e-05,
+      "loss": 0.1348,
+      "step": 8650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460808876163207e-05,
+      "loss": 0.1916,
+      "step": 8651
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4606299212598424e-05,
+      "loss": 0.2194,
+      "step": 8652
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460450966356478e-05,
+      "loss": 0.2035,
+      "step": 8653
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460272011453114e-05,
+      "loss": 0.293,
+      "step": 8654
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4600930565497496e-05,
+      "loss": 0.1656,
+      "step": 8655
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459914101646385e-05,
+      "loss": 0.1685,
+      "step": 8656
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4597351467430213e-05,
+      "loss": 0.2191,
+      "step": 8657
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459556191839657e-05,
+      "loss": 0.2087,
+      "step": 8658
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459377236936292e-05,
+      "loss": 0.0978,
+      "step": 8659
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459198282032928e-05,
+      "loss": 0.1442,
+      "step": 8660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4590193271295634e-05,
+      "loss": 0.1699,
+      "step": 8661
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458840372226199e-05,
+      "loss": 0.1551,
+      "step": 8662
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458661417322835e-05,
+      "loss": 0.1968,
+      "step": 8663
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458482462419471e-05,
+      "loss": 0.0896,
+      "step": 8664
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458303507516106e-05,
+      "loss": 0.2496,
+      "step": 8665
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458124552612742e-05,
+      "loss": 0.2494,
+      "step": 8666
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.457945597709377e-05,
+      "loss": 0.1684,
+      "step": 8667
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.457766642806013e-05,
+      "loss": 0.1238,
+      "step": 8668
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.457587687902649e-05,
+      "loss": 0.1395,
+      "step": 8669
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4574087329992845e-05,
+      "loss": 0.211,
+      "step": 8670
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.45722977809592e-05,
+      "loss": 0.1772,
+      "step": 8671
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4570508231925555e-05,
+      "loss": 0.1327,
+      "step": 8672
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.456871868289192e-05,
+      "loss": 0.1823,
+      "step": 8673
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4566929133858265e-05,
+      "loss": 0.264,
+      "step": 8674
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.456513958482463e-05,
+      "loss": 0.3504,
+      "step": 8675
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.456335003579098e-05,
+      "loss": 0.1961,
+      "step": 8676
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.456156048675734e-05,
+      "loss": 0.2345,
+      "step": 8677
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.455977093772369e-05,
+      "loss": 0.1726,
+      "step": 8678
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4557981388690055e-05,
+      "loss": 0.2144,
+      "step": 8679
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.455619183965641e-05,
+      "loss": 0.1835,
+      "step": 8680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4554402290622765e-05,
+      "loss": 0.2096,
+      "step": 8681
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.455261274158912e-05,
+      "loss": 0.1266,
+      "step": 8682
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4550823192555476e-05,
+      "loss": 0.2357,
+      "step": 8683
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.454903364352183e-05,
+      "loss": 0.1701,
+      "step": 8684
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.454724409448819e-05,
+      "loss": 0.189,
+      "step": 8685
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.454545454545455e-05,
+      "loss": 0.111,
+      "step": 8686
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.45436649964209e-05,
+      "loss": 0.1604,
+      "step": 8687
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4541875447387265e-05,
+      "loss": 0.1892,
+      "step": 8688
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.454008589835362e-05,
+      "loss": 0.2673,
+      "step": 8689
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.453829634931997e-05,
+      "loss": 0.1582,
+      "step": 8690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.453650680028633e-05,
+      "loss": 0.1923,
+      "step": 8691
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4534717251252686e-05,
+      "loss": 0.1773,
+      "step": 8692
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.453292770221904e-05,
+      "loss": 0.2741,
+      "step": 8693
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4531138153185396e-05,
+      "loss": 0.0773,
+      "step": 8694
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.452934860415176e-05,
+      "loss": 0.0924,
+      "step": 8695
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4527559055118113e-05,
+      "loss": 0.1853,
+      "step": 8696
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.452576950608447e-05,
+      "loss": 0.2268,
+      "step": 8697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4523979957050824e-05,
+      "loss": 0.1459,
+      "step": 8698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.452219040801718e-05,
+      "loss": 0.1644,
+      "step": 8699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4520400858983534e-05,
+      "loss": 0.147,
+      "step": 8700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4518611309949896e-05,
+      "loss": 0.1899,
+      "step": 8701
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.451682176091625e-05,
+      "loss": 0.1095,
+      "step": 8702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4515032211882607e-05,
+      "loss": 0.1334,
+      "step": 8703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.451324266284897e-05,
+      "loss": 0.0556,
+      "step": 8704
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.451145311381532e-05,
+      "loss": 0.2532,
+      "step": 8705
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.450966356478167e-05,
+      "loss": 0.1917,
+      "step": 8706
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4507874015748034e-05,
+      "loss": 0.2121,
+      "step": 8707
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.450608446671439e-05,
+      "loss": 0.1721,
+      "step": 8708
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4504294917680745e-05,
+      "loss": 0.1323,
+      "step": 8709
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4502505368647106e-05,
+      "loss": 0.1059,
+      "step": 8710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.450071581961346e-05,
+      "loss": 0.159,
+      "step": 8711
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449892627057982e-05,
+      "loss": 0.0961,
+      "step": 8712
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449713672154617e-05,
+      "loss": 0.2336,
+      "step": 8713
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449534717251253e-05,
+      "loss": 0.1866,
+      "step": 8714
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449355762347888e-05,
+      "loss": 0.2172,
+      "step": 8715
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449176807444524e-05,
+      "loss": 0.2232,
+      "step": 8716
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.44899785254116e-05,
+      "loss": 0.11,
+      "step": 8717
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4488188976377955e-05,
+      "loss": 0.1303,
+      "step": 8718
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.448639942734431e-05,
+      "loss": 0.1319,
+      "step": 8719
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.448460987831067e-05,
+      "loss": 0.1937,
+      "step": 8720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.448282032927702e-05,
+      "loss": 0.2875,
+      "step": 8721
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4481030780243376e-05,
+      "loss": 0.128,
+      "step": 8722
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447924123120974e-05,
+      "loss": 0.1207,
+      "step": 8723
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447745168217609e-05,
+      "loss": 0.1821,
+      "step": 8724
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447566213314245e-05,
+      "loss": 0.2249,
+      "step": 8725
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447387258410881e-05,
+      "loss": 0.2114,
+      "step": 8726
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4472083035075165e-05,
+      "loss": 0.1406,
+      "step": 8727
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447029348604152e-05,
+      "loss": 0.1307,
+      "step": 8728
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4468503937007875e-05,
+      "loss": 0.1534,
+      "step": 8729
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.446671438797423e-05,
+      "loss": 0.121,
+      "step": 8730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4464924838940586e-05,
+      "loss": 0.2274,
+      "step": 8731
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.446313528990695e-05,
+      "loss": 0.1542,
+      "step": 8732
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.44613457408733e-05,
+      "loss": 0.1357,
+      "step": 8733
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.445955619183966e-05,
+      "loss": 0.2185,
+      "step": 8734
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4457766642806013e-05,
+      "loss": 0.1567,
+      "step": 8735
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4455977093772375e-05,
+      "loss": 0.3448,
+      "step": 8736
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4454187544738724e-05,
+      "loss": 0.1458,
+      "step": 8737
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4452397995705086e-05,
+      "loss": 0.2341,
+      "step": 8738
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.445060844667144e-05,
+      "loss": 0.2186,
+      "step": 8739
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4448818897637796e-05,
+      "loss": 0.191,
+      "step": 8740
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.444702934860415e-05,
+      "loss": 0.3052,
+      "step": 8741
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.444523979957051e-05,
+      "loss": 0.0954,
+      "step": 8742
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.444345025053687e-05,
+      "loss": 0.2661,
+      "step": 8743
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4441660701503224e-05,
+      "loss": 0.0704,
+      "step": 8744
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.443987115246958e-05,
+      "loss": 0.1237,
+      "step": 8745
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4438081603435934e-05,
+      "loss": 0.1048,
+      "step": 8746
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.443629205440229e-05,
+      "loss": 0.1569,
+      "step": 8747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.443450250536865e-05,
+      "loss": 0.1843,
+      "step": 8748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4432712956335006e-05,
+      "loss": 0.1417,
+      "step": 8749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.443092340730136e-05,
+      "loss": 0.1859,
+      "step": 8750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4429133858267724e-05,
+      "loss": 0.1787,
+      "step": 8751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.442734430923407e-05,
+      "loss": 0.1499,
+      "step": 8752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.442555476020043e-05,
+      "loss": 0.1823,
+      "step": 8753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.442376521116679e-05,
+      "loss": 0.1702,
+      "step": 8754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4421975662133144e-05,
+      "loss": 0.0612,
+      "step": 8755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.44201861130995e-05,
+      "loss": 0.155,
+      "step": 8756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4418396564065855e-05,
+      "loss": 0.1063,
+      "step": 8757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.441660701503222e-05,
+      "loss": 0.1322,
+      "step": 8758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.441481746599857e-05,
+      "loss": 0.2311,
+      "step": 8759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.441302791696493e-05,
+      "loss": 0.128,
+      "step": 8760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.441123836793128e-05,
+      "loss": 0.1901,
+      "step": 8761
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440944881889764e-05,
+      "loss": 0.2907,
+      "step": 8762
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440765926986399e-05,
+      "loss": 0.1086,
+      "step": 8763
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4405869720830355e-05,
+      "loss": 0.2739,
+      "step": 8764
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440408017179671e-05,
+      "loss": 0.1955,
+      "step": 8765
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4402290622763065e-05,
+      "loss": 0.3458,
+      "step": 8766
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440050107372943e-05,
+      "loss": 0.1783,
+      "step": 8767
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4398711524695775e-05,
+      "loss": 0.1179,
+      "step": 8768
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.439692197566213e-05,
+      "loss": 0.125,
+      "step": 8769
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.439513242662849e-05,
+      "loss": 0.0797,
+      "step": 8770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.439334287759485e-05,
+      "loss": 0.1758,
+      "step": 8771
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.43915533285612e-05,
+      "loss": 0.1718,
+      "step": 8772
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4389763779527565e-05,
+      "loss": 0.1652,
+      "step": 8773
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.438797423049392e-05,
+      "loss": 0.1268,
+      "step": 8774
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4386184681460275e-05,
+      "loss": 0.1688,
+      "step": 8775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.438439513242663e-05,
+      "loss": 0.1563,
+      "step": 8776
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4382605583392986e-05,
+      "loss": 0.1994,
+      "step": 8777
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.438081603435934e-05,
+      "loss": 0.1232,
+      "step": 8778
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4379026485325696e-05,
+      "loss": 0.1089,
+      "step": 8779
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437723693629206e-05,
+      "loss": 0.3283,
+      "step": 8780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437544738725841e-05,
+      "loss": 0.069,
+      "step": 8781
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437365783822477e-05,
+      "loss": 0.0614,
+      "step": 8782
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437186828919113e-05,
+      "loss": 0.3529,
+      "step": 8783
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437007874015748e-05,
+      "loss": 0.2133,
+      "step": 8784
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4368289191123834e-05,
+      "loss": 0.2311,
+      "step": 8785
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4366499642090196e-05,
+      "loss": 0.2176,
+      "step": 8786
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.436471009305655e-05,
+      "loss": 0.1261,
+      "step": 8787
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4362920544022906e-05,
+      "loss": 0.3203,
+      "step": 8788
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.436113099498927e-05,
+      "loss": 0.1513,
+      "step": 8789
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4359341445955624e-05,
+      "loss": 0.3105,
+      "step": 8790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.435755189692198e-05,
+      "loss": 0.1793,
+      "step": 8791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4355762347888334e-05,
+      "loss": 0.1578,
+      "step": 8792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.435397279885469e-05,
+      "loss": 0.2353,
+      "step": 8793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4352183249821044e-05,
+      "loss": 0.126,
+      "step": 8794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4350393700787406e-05,
+      "loss": 0.0832,
+      "step": 8795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434860415175376e-05,
+      "loss": 0.1714,
+      "step": 8796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434681460272012e-05,
+      "loss": 0.2664,
+      "step": 8797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434502505368647e-05,
+      "loss": 0.1875,
+      "step": 8798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434323550465283e-05,
+      "loss": 0.2919,
+      "step": 8799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434144595561918e-05,
+      "loss": 0.2631,
+      "step": 8800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.433965640658554e-05,
+      "loss": 0.1265,
+      "step": 8801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.43378668575519e-05,
+      "loss": 0.2847,
+      "step": 8802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4336077308518255e-05,
+      "loss": 0.1254,
+      "step": 8803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.433428775948461e-05,
+      "loss": 0.1422,
+      "step": 8804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.433249821045097e-05,
+      "loss": 0.271,
+      "step": 8805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.433070866141733e-05,
+      "loss": 0.1893,
+      "step": 8806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4328919112383675e-05,
+      "loss": 0.1394,
+      "step": 8807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.432712956335004e-05,
+      "loss": 0.1795,
+      "step": 8808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.432534001431639e-05,
+      "loss": 0.2282,
+      "step": 8809
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.432355046528275e-05,
+      "loss": 0.1752,
+      "step": 8810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.432176091624911e-05,
+      "loss": 0.1014,
+      "step": 8811
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4319971367215465e-05,
+      "loss": 0.1764,
+      "step": 8812
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.431818181818182e-05,
+      "loss": 0.1915,
+      "step": 8813
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.431639226914818e-05,
+      "loss": 0.1611,
+      "step": 8814
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.431460272011453e-05,
+      "loss": 0.1528,
+      "step": 8815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4312813171080886e-05,
+      "loss": 0.241,
+      "step": 8816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.431102362204725e-05,
+      "loss": 0.1987,
+      "step": 8817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.43092340730136e-05,
+      "loss": 0.1747,
+      "step": 8818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.430744452397996e-05,
+      "loss": 0.2273,
+      "step": 8819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.430565497494631e-05,
+      "loss": 0.1078,
+      "step": 8820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4303865425912675e-05,
+      "loss": 0.2296,
+      "step": 8821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.430207587687903e-05,
+      "loss": 0.233,
+      "step": 8822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4300286327845386e-05,
+      "loss": 0.1388,
+      "step": 8823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.429849677881174e-05,
+      "loss": 0.2744,
+      "step": 8824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4296707229778096e-05,
+      "loss": 0.1027,
+      "step": 8825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.429491768074445e-05,
+      "loss": 0.164,
+      "step": 8826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.429312813171081e-05,
+      "loss": 0.2508,
+      "step": 8827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.429133858267717e-05,
+      "loss": 0.1444,
+      "step": 8828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4289549033643524e-05,
+      "loss": 0.1944,
+      "step": 8829
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4287759484609885e-05,
+      "loss": 0.1671,
+      "step": 8830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4285969935576234e-05,
+      "loss": 0.1726,
+      "step": 8831
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.428418038654259e-05,
+      "loss": 0.0698,
+      "step": 8832
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.428239083750895e-05,
+      "loss": 0.1113,
+      "step": 8833
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4280601288475306e-05,
+      "loss": 0.2654,
+      "step": 8834
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.427881173944166e-05,
+      "loss": 0.1277,
+      "step": 8835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4277022190408023e-05,
+      "loss": 0.2137,
+      "step": 8836
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.427523264137438e-05,
+      "loss": 0.2989,
+      "step": 8837
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.427344309234073e-05,
+      "loss": 0.1038,
+      "step": 8838
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.427165354330709e-05,
+      "loss": 0.0546,
+      "step": 8839
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4269863994273444e-05,
+      "loss": 0.2461,
+      "step": 8840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.42680744452398e-05,
+      "loss": 0.1685,
+      "step": 8841
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4266284896206155e-05,
+      "loss": 0.1585,
+      "step": 8842
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4264495347172517e-05,
+      "loss": 0.1022,
+      "step": 8843
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.426270579813887e-05,
+      "loss": 0.2096,
+      "step": 8844
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.426091624910523e-05,
+      "loss": 0.1834,
+      "step": 8845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.425912670007158e-05,
+      "loss": 0.2746,
+      "step": 8846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.425733715103794e-05,
+      "loss": 0.3174,
+      "step": 8847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.425554760200429e-05,
+      "loss": 0.1447,
+      "step": 8848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4253758052970654e-05,
+      "loss": 0.3356,
+      "step": 8849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.425196850393701e-05,
+      "loss": 0.1954,
+      "step": 8850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4250178954903365e-05,
+      "loss": 0.282,
+      "step": 8851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.424838940586973e-05,
+      "loss": 0.2354,
+      "step": 8852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.424659985683608e-05,
+      "loss": 0.1362,
+      "step": 8853
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.424481030780243e-05,
+      "loss": 0.2147,
+      "step": 8854
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.424302075876879e-05,
+      "loss": 0.2732,
+      "step": 8855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.424123120973515e-05,
+      "loss": 0.3413,
+      "step": 8856
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.42394416607015e-05,
+      "loss": 0.2257,
+      "step": 8857
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4237652111667865e-05,
+      "loss": 0.1843,
+      "step": 8858
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.423586256263422e-05,
+      "loss": 0.1576,
+      "step": 8859
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4234073013600575e-05,
+      "loss": 0.2363,
+      "step": 8860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.423228346456693e-05,
+      "loss": 0.0683,
+      "step": 8861
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4230493915533286e-05,
+      "loss": 0.1028,
+      "step": 8862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.422870436649964e-05,
+      "loss": 0.0725,
+      "step": 8863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4226914817465996e-05,
+      "loss": 0.2002,
+      "step": 8864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.422512526843236e-05,
+      "loss": 0.2198,
+      "step": 8865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.422333571939871e-05,
+      "loss": 0.1085,
+      "step": 8866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.422154617036507e-05,
+      "loss": 0.1516,
+      "step": 8867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.421975662133143e-05,
+      "loss": 0.2853,
+      "step": 8868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4217967072297785e-05,
+      "loss": 0.2453,
+      "step": 8869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4216177523264134e-05,
+      "loss": 0.2356,
+      "step": 8870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4214387974230496e-05,
+      "loss": 0.1292,
+      "step": 8871
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.421259842519685e-05,
+      "loss": 0.263,
+      "step": 8872
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4210808876163206e-05,
+      "loss": 0.3275,
+      "step": 8873
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.420901932712957e-05,
+      "loss": 0.2538,
+      "step": 8874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.420722977809592e-05,
+      "loss": 0.1983,
+      "step": 8875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.420544022906228e-05,
+      "loss": 0.1615,
+      "step": 8876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4203650680028634e-05,
+      "loss": 0.2036,
+      "step": 8877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.420186113099499e-05,
+      "loss": 0.1574,
+      "step": 8878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4200071581961344e-05,
+      "loss": 0.1459,
+      "step": 8879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4198282032927706e-05,
+      "loss": 0.1747,
+      "step": 8880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.419649248389406e-05,
+      "loss": 0.2186,
+      "step": 8881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4194702934860417e-05,
+      "loss": 0.3264,
+      "step": 8882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.419291338582677e-05,
+      "loss": 0.2114,
+      "step": 8883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4191123836793134e-05,
+      "loss": 0.1415,
+      "step": 8884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.418933428775948e-05,
+      "loss": 0.1778,
+      "step": 8885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4187544738725844e-05,
+      "loss": 0.095,
+      "step": 8886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.41857551896922e-05,
+      "loss": 0.2674,
+      "step": 8887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4183965640658554e-05,
+      "loss": 0.1358,
+      "step": 8888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.418217609162491e-05,
+      "loss": 0.0963,
+      "step": 8889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.418038654259127e-05,
+      "loss": 0.0456,
+      "step": 8890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417859699355763e-05,
+      "loss": 0.1192,
+      "step": 8891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417680744452398e-05,
+      "loss": 0.2605,
+      "step": 8892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417501789549034e-05,
+      "loss": 0.1775,
+      "step": 8893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417322834645669e-05,
+      "loss": 0.1354,
+      "step": 8894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417143879742305e-05,
+      "loss": 0.1772,
+      "step": 8895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.416964924838941e-05,
+      "loss": 0.2478,
+      "step": 8896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4167859699355765e-05,
+      "loss": 0.1093,
+      "step": 8897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.416607015032212e-05,
+      "loss": 0.1382,
+      "step": 8898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.416428060128848e-05,
+      "loss": 0.3387,
+      "step": 8899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.416249105225484e-05,
+      "loss": 0.1118,
+      "step": 8900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4160701503221186e-05,
+      "loss": 0.1119,
+      "step": 8901
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.415891195418755e-05,
+      "loss": 0.1851,
+      "step": 8902
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.41571224051539e-05,
+      "loss": 0.1261,
+      "step": 8903
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.415533285612026e-05,
+      "loss": 0.1833,
+      "step": 8904
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.415354330708661e-05,
+      "loss": 0.1104,
+      "step": 8905
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4151753758052975e-05,
+      "loss": 0.1893,
+      "step": 8906
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.414996420901933e-05,
+      "loss": 0.155,
+      "step": 8907
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4148174659985685e-05,
+      "loss": 0.1351,
+      "step": 8908
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.414638511095204e-05,
+      "loss": 0.2546,
+      "step": 8909
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4144595561918396e-05,
+      "loss": 0.2152,
+      "step": 8910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.414280601288475e-05,
+      "loss": 0.1888,
+      "step": 8911
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.414101646385111e-05,
+      "loss": 0.0986,
+      "step": 8912
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.413922691481747e-05,
+      "loss": 0.2954,
+      "step": 8913
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.413743736578382e-05,
+      "loss": 0.2455,
+      "step": 8914
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4135647816750185e-05,
+      "loss": 0.2599,
+      "step": 8915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.413385826771654e-05,
+      "loss": 0.1177,
+      "step": 8916
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.413206871868289e-05,
+      "loss": 0.1731,
+      "step": 8917
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.413027916964925e-05,
+      "loss": 0.1491,
+      "step": 8918
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4128489620615606e-05,
+      "loss": 0.3045,
+      "step": 8919
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.412670007158196e-05,
+      "loss": 0.1532,
+      "step": 8920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.412491052254832e-05,
+      "loss": 0.2481,
+      "step": 8921
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.412312097351468e-05,
+      "loss": 0.1924,
+      "step": 8922
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4121331424481034e-05,
+      "loss": 0.1365,
+      "step": 8923
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.411954187544739e-05,
+      "loss": 0.1564,
+      "step": 8924
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4117752326413744e-05,
+      "loss": 0.2216,
+      "step": 8925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.41159627773801e-05,
+      "loss": 0.2287,
+      "step": 8926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4114173228346454e-05,
+      "loss": 0.1849,
+      "step": 8927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4112383679312816e-05,
+      "loss": 0.289,
+      "step": 8928
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.411059413027917e-05,
+      "loss": 0.1404,
+      "step": 8929
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.410880458124553e-05,
+      "loss": 0.165,
+      "step": 8930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.410701503221189e-05,
+      "loss": 0.1582,
+      "step": 8931
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.410522548317824e-05,
+      "loss": 0.1303,
+      "step": 8932
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.410343593414459e-05,
+      "loss": 0.2634,
+      "step": 8933
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4101646385110954e-05,
+      "loss": 0.1348,
+      "step": 8934
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409985683607731e-05,
+      "loss": 0.3333,
+      "step": 8935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4098067287043665e-05,
+      "loss": 0.134,
+      "step": 8936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409627773801003e-05,
+      "loss": 0.1944,
+      "step": 8937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409448818897638e-05,
+      "loss": 0.241,
+      "step": 8938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409269863994274e-05,
+      "loss": 0.2331,
+      "step": 8939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.1656,
+      "step": 8940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.408911954187545e-05,
+      "loss": 0.0786,
+      "step": 8941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.40873299928418e-05,
+      "loss": 0.1698,
+      "step": 8942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4085540443808165e-05,
+      "loss": 0.1335,
+      "step": 8943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.408375089477452e-05,
+      "loss": 0.0961,
+      "step": 8944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4081961345740875e-05,
+      "loss": 0.153,
+      "step": 8945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.408017179670723e-05,
+      "loss": 0.2121,
+      "step": 8946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.407838224767359e-05,
+      "loss": 0.3,
+      "step": 8947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.407659269863994e-05,
+      "loss": 0.2011,
+      "step": 8948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.40748031496063e-05,
+      "loss": 0.2736,
+      "step": 8949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.407301360057266e-05,
+      "loss": 0.1873,
+      "step": 8950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.407122405153901e-05,
+      "loss": 0.2033,
+      "step": 8951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.406943450250537e-05,
+      "loss": 0.1331,
+      "step": 8952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.406764495347173e-05,
+      "loss": 0.1973,
+      "step": 8953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4065855404438085e-05,
+      "loss": 0.3037,
+      "step": 8954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.406406585540444e-05,
+      "loss": 0.1139,
+      "step": 8955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4062276306370796e-05,
+      "loss": 0.0792,
+      "step": 8956
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.406048675733715e-05,
+      "loss": 0.0953,
+      "step": 8957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4058697208303506e-05,
+      "loss": 0.1342,
+      "step": 8958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.405690765926987e-05,
+      "loss": 0.149,
+      "step": 8959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.405511811023622e-05,
+      "loss": 0.1313,
+      "step": 8960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.405332856120258e-05,
+      "loss": 0.098,
+      "step": 8961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.405153901216894e-05,
+      "loss": 0.1062,
+      "step": 8962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4049749463135296e-05,
+      "loss": 0.1793,
+      "step": 8963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4047959914101644e-05,
+      "loss": 0.2165,
+      "step": 8964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4046170365068006e-05,
+      "loss": 0.1162,
+      "step": 8965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.404438081603436e-05,
+      "loss": 0.1612,
+      "step": 8966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4042591267000716e-05,
+      "loss": 0.1116,
+      "step": 8967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.404080171796707e-05,
+      "loss": 0.1646,
+      "step": 8968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4039012168933433e-05,
+      "loss": 0.2407,
+      "step": 8969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.403722261989979e-05,
+      "loss": 0.2314,
+      "step": 8970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4035433070866144e-05,
+      "loss": 0.2237,
+      "step": 8971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.40336435218325e-05,
+      "loss": 0.1517,
+      "step": 8972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4031853972798854e-05,
+      "loss": 0.1828,
+      "step": 8973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.403006442376521e-05,
+      "loss": 0.1203,
+      "step": 8974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.402827487473157e-05,
+      "loss": 0.1424,
+      "step": 8975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.402648532569793e-05,
+      "loss": 0.1285,
+      "step": 8976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.402469577666428e-05,
+      "loss": 0.1638,
+      "step": 8977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4022906227630644e-05,
+      "loss": 0.0631,
+      "step": 8978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.402111667859699e-05,
+      "loss": 0.2238,
+      "step": 8979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.401932712956335e-05,
+      "loss": 0.1546,
+      "step": 8980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.401753758052971e-05,
+      "loss": 0.0743,
+      "step": 8981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4015748031496065e-05,
+      "loss": 0.2408,
+      "step": 8982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.401395848246242e-05,
+      "loss": 0.1501,
+      "step": 8983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.401216893342878e-05,
+      "loss": 0.2025,
+      "step": 8984
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.401037938439514e-05,
+      "loss": 0.0561,
+      "step": 8985
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400858983536149e-05,
+      "loss": 0.1648,
+      "step": 8986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400680028632785e-05,
+      "loss": 0.2195,
+      "step": 8987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.40050107372942e-05,
+      "loss": 0.254,
+      "step": 8988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400322118826056e-05,
+      "loss": 0.3248,
+      "step": 8989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400143163922691e-05,
+      "loss": 0.0971,
+      "step": 8990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3999642090193275e-05,
+      "loss": 0.1716,
+      "step": 8991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.399785254115963e-05,
+      "loss": 0.13,
+      "step": 8992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3996062992125985e-05,
+      "loss": 0.1486,
+      "step": 8993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.399427344309235e-05,
+      "loss": 0.144,
+      "step": 8994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3992483894058696e-05,
+      "loss": 0.1968,
+      "step": 8995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.399069434502505e-05,
+      "loss": 0.3413,
+      "step": 8996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398890479599141e-05,
+      "loss": 0.1887,
+      "step": 8997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398711524695777e-05,
+      "loss": 0.1905,
+      "step": 8998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398532569792412e-05,
+      "loss": 0.1468,
+      "step": 8999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3983536148890485e-05,
+      "loss": 0.2628,
+      "step": 9000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398174659985684e-05,
+      "loss": 0.1215,
+      "step": 9001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3979957050823196e-05,
+      "loss": 0.1361,
+      "step": 9002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.397816750178955e-05,
+      "loss": 0.2343,
+      "step": 9003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3976377952755906e-05,
+      "loss": 0.0562,
+      "step": 9004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.397458840372226e-05,
+      "loss": 0.1851,
+      "step": 9005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.397279885468862e-05,
+      "loss": 0.4444,
+      "step": 9006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.397100930565498e-05,
+      "loss": 0.1748,
+      "step": 9007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3969219756621333e-05,
+      "loss": 0.0642,
+      "step": 9008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.396743020758769e-05,
+      "loss": 0.1356,
+      "step": 9009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.396564065855405e-05,
+      "loss": 0.0983,
+      "step": 9010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.39638511095204e-05,
+      "loss": 0.1518,
+      "step": 9011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3962061560486754e-05,
+      "loss": 0.3537,
+      "step": 9012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3960272011453116e-05,
+      "loss": 0.1023,
+      "step": 9013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.395848246241947e-05,
+      "loss": 0.1384,
+      "step": 9014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3956692913385827e-05,
+      "loss": 0.2353,
+      "step": 9015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.395490336435219e-05,
+      "loss": 0.165,
+      "step": 9016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3953113815318544e-05,
+      "loss": 0.2092,
+      "step": 9017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.395132426628489e-05,
+      "loss": 0.095,
+      "step": 9018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3949534717251254e-05,
+      "loss": 0.3037,
+      "step": 9019
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.394774516821761e-05,
+      "loss": 0.128,
+      "step": 9020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3945955619183965e-05,
+      "loss": 0.1951,
+      "step": 9021
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3944166070150326e-05,
+      "loss": 0.2667,
+      "step": 9022
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.394237652111668e-05,
+      "loss": 0.1158,
+      "step": 9023
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.394058697208304e-05,
+      "loss": 0.1517,
+      "step": 9024
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.39387974230494e-05,
+      "loss": 0.1217,
+      "step": 9025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.393700787401575e-05,
+      "loss": 0.1011,
+      "step": 9026
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.39352183249821e-05,
+      "loss": 0.277,
+      "step": 9027
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3933428775948464e-05,
+      "loss": 0.2751,
+      "step": 9028
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.393163922691482e-05,
+      "loss": 0.1775,
+      "step": 9029
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3929849677881175e-05,
+      "loss": 0.1041,
+      "step": 9030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.392806012884753e-05,
+      "loss": 0.1625,
+      "step": 9031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.392627057981389e-05,
+      "loss": 0.1536,
+      "step": 9032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.392448103078025e-05,
+      "loss": 0.1021,
+      "step": 9033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.39226914817466e-05,
+      "loss": 0.0753,
+      "step": 9034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.392090193271296e-05,
+      "loss": 0.2753,
+      "step": 9035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.391911238367931e-05,
+      "loss": 0.1271,
+      "step": 9036
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.391732283464567e-05,
+      "loss": 0.293,
+      "step": 9037
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.391553328561203e-05,
+      "loss": 0.2749,
+      "step": 9038
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3913743736578385e-05,
+      "loss": 0.3564,
+      "step": 9039
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.391195418754474e-05,
+      "loss": 0.2081,
+      "step": 9040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.39101646385111e-05,
+      "loss": 0.1654,
+      "step": 9041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390837508947745e-05,
+      "loss": 0.2602,
+      "step": 9042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3906585540443806e-05,
+      "loss": 0.4026,
+      "step": 9043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390479599141017e-05,
+      "loss": 0.2095,
+      "step": 9044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390300644237652e-05,
+      "loss": 0.2921,
+      "step": 9045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390121689334288e-05,
+      "loss": 0.1353,
+      "step": 9046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.389942734430924e-05,
+      "loss": 0.2413,
+      "step": 9047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3897637795275595e-05,
+      "loss": 0.1077,
+      "step": 9048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.389584824624195e-05,
+      "loss": 0.1282,
+      "step": 9049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3894058697208306e-05,
+      "loss": 0.3135,
+      "step": 9050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.389226914817466e-05,
+      "loss": 0.2265,
+      "step": 9051
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3890479599141016e-05,
+      "loss": 0.1415,
+      "step": 9052
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.388869005010737e-05,
+      "loss": 0.1002,
+      "step": 9053
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.388690050107373e-05,
+      "loss": 0.1164,
+      "step": 9054
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.388511095204009e-05,
+      "loss": 0.1389,
+      "step": 9055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3883321403006444e-05,
+      "loss": 0.1394,
+      "step": 9056
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.38815318539728e-05,
+      "loss": 0.1776,
+      "step": 9057
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3879742304939154e-05,
+      "loss": 0.2139,
+      "step": 9058
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.387795275590551e-05,
+      "loss": 0.1995,
+      "step": 9059
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.387616320687187e-05,
+      "loss": 0.2657,
+      "step": 9060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3874373657838226e-05,
+      "loss": 0.2055,
+      "step": 9061
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.387258410880458e-05,
+      "loss": 0.2511,
+      "step": 9062
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3870794559770944e-05,
+      "loss": 0.3047,
+      "step": 9063
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.38690050107373e-05,
+      "loss": 0.2628,
+      "step": 9064
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.386721546170365e-05,
+      "loss": 0.1035,
+      "step": 9065
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.386542591267001e-05,
+      "loss": 0.0971,
+      "step": 9066
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3863636363636364e-05,
+      "loss": 0.1112,
+      "step": 9067
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.386184681460272e-05,
+      "loss": 0.2191,
+      "step": 9068
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.386005726556908e-05,
+      "loss": 0.2429,
+      "step": 9069
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.385826771653544e-05,
+      "loss": 0.1163,
+      "step": 9070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.385647816750179e-05,
+      "loss": 0.2269,
+      "step": 9071
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.385468861846815e-05,
+      "loss": 0.2296,
+      "step": 9072
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.38528990694345e-05,
+      "loss": 0.2802,
+      "step": 9073
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.385110952040086e-05,
+      "loss": 0.2392,
+      "step": 9074
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.384931997136721e-05,
+      "loss": 0.2261,
+      "step": 9075
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3847530422333575e-05,
+      "loss": 0.0829,
+      "step": 9076
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.384574087329993e-05,
+      "loss": 0.1173,
+      "step": 9077
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3843951324266285e-05,
+      "loss": 0.3413,
+      "step": 9078
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.384216177523265e-05,
+      "loss": 0.1622,
+      "step": 9079
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3840372226199e-05,
+      "loss": 0.1677,
+      "step": 9080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.383858267716535e-05,
+      "loss": 0.1911,
+      "step": 9081
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.383679312813171e-05,
+      "loss": 0.2,
+      "step": 9082
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.383500357909807e-05,
+      "loss": 0.2083,
+      "step": 9083
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.383321403006442e-05,
+      "loss": 0.1691,
+      "step": 9084
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3831424481030785e-05,
+      "loss": 0.0877,
+      "step": 9085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.382963493199714e-05,
+      "loss": 0.1435,
+      "step": 9086
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3827845382963495e-05,
+      "loss": 0.2927,
+      "step": 9087
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.382605583392985e-05,
+      "loss": 0.2692,
+      "step": 9088
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3824266284896206e-05,
+      "loss": 0.3349,
+      "step": 9089
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.382247673586256e-05,
+      "loss": 0.1322,
+      "step": 9090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.382068718682892e-05,
+      "loss": 0.1683,
+      "step": 9091
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.381889763779528e-05,
+      "loss": 0.1494,
+      "step": 9092
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.381710808876163e-05,
+      "loss": 0.0975,
+      "step": 9093
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.381531853972799e-05,
+      "loss": 0.224,
+      "step": 9094
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.381352899069435e-05,
+      "loss": 0.2982,
+      "step": 9095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3811739441660706e-05,
+      "loss": 0.1766,
+      "step": 9096
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.380994989262706e-05,
+      "loss": 0.2211,
+      "step": 9097
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3808160343593416e-05,
+      "loss": 0.331,
+      "step": 9098
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.380637079455977e-05,
+      "loss": 0.1851,
+      "step": 9099
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3804581245526126e-05,
+      "loss": 0.2729,
+      "step": 9100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.380279169649249e-05,
+      "loss": 0.1011,
+      "step": 9101
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3801002147458844e-05,
+      "loss": 0.2304,
+      "step": 9102
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37992125984252e-05,
+      "loss": 0.2882,
+      "step": 9103
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3797423049391554e-05,
+      "loss": 0.1967,
+      "step": 9104
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.379563350035791e-05,
+      "loss": 0.0652,
+      "step": 9105
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3793843951324264e-05,
+      "loss": 0.1823,
+      "step": 9106
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3792054402290626e-05,
+      "loss": 0.2705,
+      "step": 9107
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.379026485325698e-05,
+      "loss": 0.1947,
+      "step": 9108
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.378847530422334e-05,
+      "loss": 0.0791,
+      "step": 9109
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37866857551897e-05,
+      "loss": 0.1031,
+      "step": 9110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3784896206156054e-05,
+      "loss": 0.1159,
+      "step": 9111
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37831066571224e-05,
+      "loss": 0.2141,
+      "step": 9112
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3781317108088764e-05,
+      "loss": 0.0939,
+      "step": 9113
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377952755905512e-05,
+      "loss": 0.1055,
+      "step": 9114
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3777738010021475e-05,
+      "loss": 0.1406,
+      "step": 9115
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377594846098783e-05,
+      "loss": 0.2041,
+      "step": 9116
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377415891195419e-05,
+      "loss": 0.1844,
+      "step": 9117
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377236936292055e-05,
+      "loss": 0.176,
+      "step": 9118
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37705798138869e-05,
+      "loss": 0.2192,
+      "step": 9119
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.376879026485326e-05,
+      "loss": 0.2127,
+      "step": 9120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.376700071581961e-05,
+      "loss": 0.1362,
+      "step": 9121
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.376521116678597e-05,
+      "loss": 0.2465,
+      "step": 9122
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.376342161775233e-05,
+      "loss": 0.1401,
+      "step": 9123
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3761632068718685e-05,
+      "loss": 0.2071,
+      "step": 9124
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.375984251968504e-05,
+      "loss": 0.1556,
+      "step": 9125
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37580529706514e-05,
+      "loss": 0.2897,
+      "step": 9126
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.375626342161776e-05,
+      "loss": 0.0602,
+      "step": 9127
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3754473872584106e-05,
+      "loss": 0.1302,
+      "step": 9128
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.375268432355047e-05,
+      "loss": 0.2433,
+      "step": 9129
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.375089477451682e-05,
+      "loss": 0.1669,
+      "step": 9130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.374910522548318e-05,
+      "loss": 0.1074,
+      "step": 9131
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.374731567644954e-05,
+      "loss": 0.1517,
+      "step": 9132
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3745526127415895e-05,
+      "loss": 0.1023,
+      "step": 9133
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.374373657838225e-05,
+      "loss": 0.1094,
+      "step": 9134
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3741947029348606e-05,
+      "loss": 0.1807,
+      "step": 9135
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.374015748031496e-05,
+      "loss": 0.147,
+      "step": 9136
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3738367931281316e-05,
+      "loss": 0.1607,
+      "step": 9137
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.373657838224767e-05,
+      "loss": 0.1394,
+      "step": 9138
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.373478883321403e-05,
+      "loss": 0.2057,
+      "step": 9139
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.373299928418039e-05,
+      "loss": 0.1064,
+      "step": 9140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3731209735146744e-05,
+      "loss": 0.2148,
+      "step": 9141
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3729420186113105e-05,
+      "loss": 0.2225,
+      "step": 9142
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.372763063707946e-05,
+      "loss": 0.1472,
+      "step": 9143
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.372584108804581e-05,
+      "loss": 0.0907,
+      "step": 9144
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.372405153901217e-05,
+      "loss": 0.1184,
+      "step": 9145
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3722261989978526e-05,
+      "loss": 0.2093,
+      "step": 9146
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.372047244094488e-05,
+      "loss": 0.1263,
+      "step": 9147
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3718682891911243e-05,
+      "loss": 0.1453,
+      "step": 9148
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37168933428776e-05,
+      "loss": 0.0916,
+      "step": 9149
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3715103793843954e-05,
+      "loss": 0.1818,
+      "step": 9150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.371331424481031e-05,
+      "loss": 0.2023,
+      "step": 9151
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3711524695776664e-05,
+      "loss": 0.2574,
+      "step": 9152
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370973514674302e-05,
+      "loss": 0.1646,
+      "step": 9153
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370794559770938e-05,
+      "loss": 0.1151,
+      "step": 9154
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3706156048675737e-05,
+      "loss": 0.1642,
+      "step": 9155
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370436649964209e-05,
+      "loss": 0.1028,
+      "step": 9156
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370257695060845e-05,
+      "loss": 0.0708,
+      "step": 9157
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370078740157481e-05,
+      "loss": 0.1066,
+      "step": 9158
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369899785254116e-05,
+      "loss": 0.1695,
+      "step": 9159
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369720830350752e-05,
+      "loss": 0.395,
+      "step": 9160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3695418754473874e-05,
+      "loss": 0.2319,
+      "step": 9161
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369362920544023e-05,
+      "loss": 0.1465,
+      "step": 9162
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3691839656406585e-05,
+      "loss": 0.1923,
+      "step": 9163
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369005010737295e-05,
+      "loss": 0.1936,
+      "step": 9164
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.36882605583393e-05,
+      "loss": 0.1636,
+      "step": 9165
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368647100930566e-05,
+      "loss": 0.1947,
+      "step": 9166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368468146027201e-05,
+      "loss": 0.1913,
+      "step": 9167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368289191123837e-05,
+      "loss": 0.1721,
+      "step": 9168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368110236220472e-05,
+      "loss": 0.136,
+      "step": 9169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3679312813171085e-05,
+      "loss": 0.1863,
+      "step": 9170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.367752326413744e-05,
+      "loss": 0.1773,
+      "step": 9171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3675733715103795e-05,
+      "loss": 0.1856,
+      "step": 9172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.367394416607016e-05,
+      "loss": 0.1535,
+      "step": 9173
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.367215461703651e-05,
+      "loss": 0.0796,
+      "step": 9174
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.367036506800286e-05,
+      "loss": 0.2396,
+      "step": 9175
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.366857551896922e-05,
+      "loss": 0.2088,
+      "step": 9176
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.366678596993558e-05,
+      "loss": 0.2631,
+      "step": 9177
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.366499642090193e-05,
+      "loss": 0.1557,
+      "step": 9178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.366320687186829e-05,
+      "loss": 0.0906,
+      "step": 9179
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.366141732283465e-05,
+      "loss": 0.2208,
+      "step": 9180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3659627773801005e-05,
+      "loss": 0.16,
+      "step": 9181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.365783822476736e-05,
+      "loss": 0.2399,
+      "step": 9182
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3656048675733716e-05,
+      "loss": 0.1593,
+      "step": 9183
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.365425912670007e-05,
+      "loss": 0.1106,
+      "step": 9184
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3652469577666426e-05,
+      "loss": 0.1961,
+      "step": 9185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.365068002863279e-05,
+      "loss": 0.0957,
+      "step": 9186
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.364889047959914e-05,
+      "loss": 0.2185,
+      "step": 9187
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.36471009305655e-05,
+      "loss": 0.1141,
+      "step": 9188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.364531138153186e-05,
+      "loss": 0.1433,
+      "step": 9189
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.364352183249821e-05,
+      "loss": 0.185,
+      "step": 9190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3641732283464564e-05,
+      "loss": 0.1807,
+      "step": 9191
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3639942734430926e-05,
+      "loss": 0.1829,
+      "step": 9192
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.363815318539728e-05,
+      "loss": 0.1759,
+      "step": 9193
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3636363636363636e-05,
+      "loss": 0.2456,
+      "step": 9194
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.363457408733e-05,
+      "loss": 0.1565,
+      "step": 9195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3632784538296354e-05,
+      "loss": 0.1719,
+      "step": 9196
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.363099498926271e-05,
+      "loss": 0.2691,
+      "step": 9197
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3629205440229064e-05,
+      "loss": 0.1914,
+      "step": 9198
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.362741589119542e-05,
+      "loss": 0.0771,
+      "step": 9199
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3625626342161774e-05,
+      "loss": 0.1172,
+      "step": 9200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.362383679312813e-05,
+      "loss": 0.083,
+      "step": 9201
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.362204724409449e-05,
+      "loss": 0.0972,
+      "step": 9202
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.362025769506085e-05,
+      "loss": 0.1494,
+      "step": 9203
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.36184681460272e-05,
+      "loss": 0.2379,
+      "step": 9204
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3616678596993564e-05,
+      "loss": 0.2046,
+      "step": 9205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.361488904795991e-05,
+      "loss": 0.1906,
+      "step": 9206
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.361309949892627e-05,
+      "loss": 0.1444,
+      "step": 9207
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.361130994989263e-05,
+      "loss": 0.1278,
+      "step": 9208
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3609520400858985e-05,
+      "loss": 0.081,
+      "step": 9209
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360773085182534e-05,
+      "loss": 0.2016,
+      "step": 9210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.36059413027917e-05,
+      "loss": 0.1026,
+      "step": 9211
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360415175375806e-05,
+      "loss": 0.1725,
+      "step": 9212
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360236220472441e-05,
+      "loss": 0.1713,
+      "step": 9213
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360057265569077e-05,
+      "loss": 0.1352,
+      "step": 9214
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359878310665712e-05,
+      "loss": 0.204,
+      "step": 9215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359699355762348e-05,
+      "loss": 0.131,
+      "step": 9216
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359520400858984e-05,
+      "loss": 0.4866,
+      "step": 9217
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3593414459556195e-05,
+      "loss": 0.2035,
+      "step": 9218
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359162491052255e-05,
+      "loss": 0.2258,
+      "step": 9219
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3589835361488905e-05,
+      "loss": 0.2338,
+      "step": 9220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.358804581245527e-05,
+      "loss": 0.1961,
+      "step": 9221
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3586256263421616e-05,
+      "loss": 0.1717,
+      "step": 9222
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.358446671438797e-05,
+      "loss": 0.1486,
+      "step": 9223
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.358267716535433e-05,
+      "loss": 0.2435,
+      "step": 9224
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.358088761632069e-05,
+      "loss": 0.0849,
+      "step": 9225
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.357909806728704e-05,
+      "loss": 0.1158,
+      "step": 9226
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3577308518253405e-05,
+      "loss": 0.2143,
+      "step": 9227
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.357551896921976e-05,
+      "loss": 0.1526,
+      "step": 9228
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3573729420186116e-05,
+      "loss": 0.0974,
+      "step": 9229
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.357193987115247e-05,
+      "loss": 0.0609,
+      "step": 9230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3570150322118826e-05,
+      "loss": 0.0753,
+      "step": 9231
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.356836077308518e-05,
+      "loss": 0.1424,
+      "step": 9232
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.356657122405154e-05,
+      "loss": 0.3917,
+      "step": 9233
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.35647816750179e-05,
+      "loss": 0.1302,
+      "step": 9234
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3562992125984254e-05,
+      "loss": 0.1659,
+      "step": 9235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3561202576950616e-05,
+      "loss": 0.1718,
+      "step": 9236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3559413027916964e-05,
+      "loss": 0.1115,
+      "step": 9237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.355762347888332e-05,
+      "loss": 0.0921,
+      "step": 9238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.355583392984968e-05,
+      "loss": 0.1419,
+      "step": 9239
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3554044380816036e-05,
+      "loss": 0.1269,
+      "step": 9240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.355225483178239e-05,
+      "loss": 0.1962,
+      "step": 9241
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.355046528274875e-05,
+      "loss": 0.1744,
+      "step": 9242
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.354867573371511e-05,
+      "loss": 0.1068,
+      "step": 9243
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3546886184681464e-05,
+      "loss": 0.144,
+      "step": 9244
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.354509663564782e-05,
+      "loss": 0.1216,
+      "step": 9245
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3543307086614174e-05,
+      "loss": 0.1251,
+      "step": 9246
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.354151753758053e-05,
+      "loss": 0.1541,
+      "step": 9247
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3539727988546885e-05,
+      "loss": 0.2089,
+      "step": 9248
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.353793843951325e-05,
+      "loss": 0.0775,
+      "step": 9249
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.35361488904796e-05,
+      "loss": 0.1078,
+      "step": 9250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.353435934144596e-05,
+      "loss": 0.2863,
+      "step": 9251
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.353256979241232e-05,
+      "loss": 0.1336,
+      "step": 9252
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.353078024337867e-05,
+      "loss": 0.0784,
+      "step": 9253
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.352899069434502e-05,
+      "loss": 0.1483,
+      "step": 9254
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3527201145311385e-05,
+      "loss": 0.1167,
+      "step": 9255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.352541159627774e-05,
+      "loss": 0.1995,
+      "step": 9256
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3523622047244095e-05,
+      "loss": 0.3095,
+      "step": 9257
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.352183249821046e-05,
+      "loss": 0.082,
+      "step": 9258
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.352004294917681e-05,
+      "loss": 0.1989,
+      "step": 9259
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.351825340014317e-05,
+      "loss": 0.1191,
+      "step": 9260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.351646385110952e-05,
+      "loss": 0.2598,
+      "step": 9261
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.351467430207588e-05,
+      "loss": 0.2488,
+      "step": 9262
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.351288475304223e-05,
+      "loss": 0.1942,
+      "step": 9263
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.351109520400859e-05,
+      "loss": 0.0827,
+      "step": 9264
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.350930565497495e-05,
+      "loss": 0.2149,
+      "step": 9265
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3507516105941305e-05,
+      "loss": 0.2626,
+      "step": 9266
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.350572655690766e-05,
+      "loss": 0.1838,
+      "step": 9267
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.350393700787402e-05,
+      "loss": 0.2101,
+      "step": 9268
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.350214745884037e-05,
+      "loss": 0.216,
+      "step": 9269
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3500357909806726e-05,
+      "loss": 0.175,
+      "step": 9270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349856836077309e-05,
+      "loss": 0.1237,
+      "step": 9271
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349677881173944e-05,
+      "loss": 0.0843,
+      "step": 9272
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.34949892627058e-05,
+      "loss": 0.1443,
+      "step": 9273
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349319971367216e-05,
+      "loss": 0.1527,
+      "step": 9274
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3491410164638516e-05,
+      "loss": 0.2165,
+      "step": 9275
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3489620615604864e-05,
+      "loss": 0.3093,
+      "step": 9276
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3487831066571226e-05,
+      "loss": 0.3005,
+      "step": 9277
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.348604151753758e-05,
+      "loss": 0.3048,
+      "step": 9278
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3484251968503936e-05,
+      "loss": 0.1851,
+      "step": 9279
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.34824624194703e-05,
+      "loss": 0.1322,
+      "step": 9280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3480672870436653e-05,
+      "loss": 0.234,
+      "step": 9281
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.347888332140301e-05,
+      "loss": 0.0995,
+      "step": 9282
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3477093772369364e-05,
+      "loss": 0.1172,
+      "step": 9283
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.347530422333572e-05,
+      "loss": 0.1563,
+      "step": 9284
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3473514674302074e-05,
+      "loss": 0.1822,
+      "step": 9285
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.347172512526843e-05,
+      "loss": 0.0468,
+      "step": 9286
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.346993557623479e-05,
+      "loss": 0.1693,
+      "step": 9287
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3468146027201147e-05,
+      "loss": 0.2381,
+      "step": 9288
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.34663564781675e-05,
+      "loss": 0.1834,
+      "step": 9289
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3464566929133864e-05,
+      "loss": 0.1417,
+      "step": 9290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.346277738010022e-05,
+      "loss": 0.2023,
+      "step": 9291
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.346098783106657e-05,
+      "loss": 0.1986,
+      "step": 9292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.345919828203293e-05,
+      "loss": 0.2336,
+      "step": 9293
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3457408732999285e-05,
+      "loss": 0.15,
+      "step": 9294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.345561918396564e-05,
+      "loss": 0.1297,
+      "step": 9295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3453829634932e-05,
+      "loss": 0.153,
+      "step": 9296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.345204008589836e-05,
+      "loss": 0.1606,
+      "step": 9297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.345025053686471e-05,
+      "loss": 0.1529,
+      "step": 9298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.344846098783107e-05,
+      "loss": 0.2113,
+      "step": 9299
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.344667143879742e-05,
+      "loss": 0.0808,
+      "step": 9300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.344488188976378e-05,
+      "loss": 0.1636,
+      "step": 9301
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.344309234073014e-05,
+      "loss": 0.1623,
+      "step": 9302
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3441302791696495e-05,
+      "loss": 0.0989,
+      "step": 9303
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.343951324266285e-05,
+      "loss": 0.1442,
+      "step": 9304
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3437723693629205e-05,
+      "loss": 0.1049,
+      "step": 9305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.343593414459557e-05,
+      "loss": 0.1376,
+      "step": 9306
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.343414459556192e-05,
+      "loss": 0.1426,
+      "step": 9307
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.343235504652828e-05,
+      "loss": 0.293,
+      "step": 9308
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.343056549749463e-05,
+      "loss": 0.0587,
+      "step": 9309
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.342877594846099e-05,
+      "loss": 0.1725,
+      "step": 9310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.342698639942734e-05,
+      "loss": 0.2107,
+      "step": 9311
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3425196850393705e-05,
+      "loss": 0.1847,
+      "step": 9312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.342340730136006e-05,
+      "loss": 0.1931,
+      "step": 9313
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3421617752326416e-05,
+      "loss": 0.2278,
+      "step": 9314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.341982820329278e-05,
+      "loss": 0.1967,
+      "step": 9315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3418038654259126e-05,
+      "loss": 0.2474,
+      "step": 9316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.341624910522548e-05,
+      "loss": 0.0725,
+      "step": 9317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.341445955619184e-05,
+      "loss": 0.1433,
+      "step": 9318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.34126700071582e-05,
+      "loss": 0.1728,
+      "step": 9319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3410880458124553e-05,
+      "loss": 0.0861,
+      "step": 9320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3409090909090915e-05,
+      "loss": 0.1389,
+      "step": 9321
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340730136005727e-05,
+      "loss": 0.2235,
+      "step": 9322
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340551181102362e-05,
+      "loss": 0.16,
+      "step": 9323
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340372226198998e-05,
+      "loss": 0.1708,
+      "step": 9324
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3401932712956336e-05,
+      "loss": 0.1694,
+      "step": 9325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340014316392269e-05,
+      "loss": 0.3729,
+      "step": 9326
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3398353614889047e-05,
+      "loss": 0.2286,
+      "step": 9327
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.339656406585541e-05,
+      "loss": 0.1459,
+      "step": 9328
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3394774516821764e-05,
+      "loss": 0.0567,
+      "step": 9329
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.339298496778812e-05,
+      "loss": 0.1893,
+      "step": 9330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3391195418754474e-05,
+      "loss": 0.1323,
+      "step": 9331
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.338940586972083e-05,
+      "loss": 0.133,
+      "step": 9332
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3387616320687184e-05,
+      "loss": 0.133,
+      "step": 9333
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3385826771653546e-05,
+      "loss": 0.2147,
+      "step": 9334
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.33840372226199e-05,
+      "loss": 0.065,
+      "step": 9335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.338224767358626e-05,
+      "loss": 0.2065,
+      "step": 9336
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.338045812455262e-05,
+      "loss": 0.1977,
+      "step": 9337
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3378668575518974e-05,
+      "loss": 0.2155,
+      "step": 9338
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.337687902648532e-05,
+      "loss": 0.1327,
+      "step": 9339
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3375089477451684e-05,
+      "loss": 0.1052,
+      "step": 9340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.337329992841804e-05,
+      "loss": 0.1869,
+      "step": 9341
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3371510379384395e-05,
+      "loss": 0.1828,
+      "step": 9342
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336972083035076e-05,
+      "loss": 0.2831,
+      "step": 9343
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336793128131711e-05,
+      "loss": 0.1092,
+      "step": 9344
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336614173228347e-05,
+      "loss": 0.0813,
+      "step": 9345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336435218324982e-05,
+      "loss": 0.1184,
+      "step": 9346
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336256263421618e-05,
+      "loss": 0.0445,
+      "step": 9347
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336077308518253e-05,
+      "loss": 0.1127,
+      "step": 9348
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335898353614889e-05,
+      "loss": 0.1578,
+      "step": 9349
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335719398711525e-05,
+      "loss": 0.3819,
+      "step": 9350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3355404438081605e-05,
+      "loss": 0.1578,
+      "step": 9351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335361488904796e-05,
+      "loss": 0.1346,
+      "step": 9352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335182534001432e-05,
+      "loss": 0.0874,
+      "step": 9353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.335003579098068e-05,
+      "loss": 0.1978,
+      "step": 9354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3348246241947026e-05,
+      "loss": 0.1812,
+      "step": 9355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.334645669291339e-05,
+      "loss": 0.2426,
+      "step": 9356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.334466714387974e-05,
+      "loss": 0.1563,
+      "step": 9357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.33428775948461e-05,
+      "loss": 0.2035,
+      "step": 9358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.334108804581246e-05,
+      "loss": 0.2386,
+      "step": 9359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3339298496778815e-05,
+      "loss": 0.1921,
+      "step": 9360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.333750894774517e-05,
+      "loss": 0.2,
+      "step": 9361
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3335719398711526e-05,
+      "loss": 0.2029,
+      "step": 9362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.333392984967788e-05,
+      "loss": 0.248,
+      "step": 9363
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3332140300644236e-05,
+      "loss": 0.0996,
+      "step": 9364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.33303507516106e-05,
+      "loss": 0.1596,
+      "step": 9365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.332856120257695e-05,
+      "loss": 0.1817,
+      "step": 9366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.332677165354331e-05,
+      "loss": 0.2103,
+      "step": 9367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3324982104509664e-05,
+      "loss": 0.1273,
+      "step": 9368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3323192555476026e-05,
+      "loss": 0.1747,
+      "step": 9369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3321403006442374e-05,
+      "loss": 0.1052,
+      "step": 9370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.331961345740873e-05,
+      "loss": 0.2157,
+      "step": 9371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.331782390837509e-05,
+      "loss": 0.1906,
+      "step": 9372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3316034359341446e-05,
+      "loss": 0.0859,
+      "step": 9373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.33142448103078e-05,
+      "loss": 0.1843,
+      "step": 9374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3312455261274164e-05,
+      "loss": 0.2808,
+      "step": 9375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.331066571224052e-05,
+      "loss": 0.1273,
+      "step": 9376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3308876163206874e-05,
+      "loss": 0.275,
+      "step": 9377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.330708661417323e-05,
+      "loss": 0.1605,
+      "step": 9378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3305297065139584e-05,
+      "loss": 0.3549,
+      "step": 9379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.330350751610594e-05,
+      "loss": 0.0848,
+      "step": 9380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.33017179670723e-05,
+      "loss": 0.1602,
+      "step": 9381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329992841803866e-05,
+      "loss": 0.184,
+      "step": 9382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329813886900501e-05,
+      "loss": 0.1732,
+      "step": 9383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3296349319971374e-05,
+      "loss": 0.2091,
+      "step": 9384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329455977093773e-05,
+      "loss": 0.1096,
+      "step": 9385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329277022190408e-05,
+      "loss": 0.0713,
+      "step": 9386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329098067287044e-05,
+      "loss": 0.1867,
+      "step": 9387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3289191123836795e-05,
+      "loss": 0.2822,
+      "step": 9388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.328740157480315e-05,
+      "loss": 0.0812,
+      "step": 9389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3285612025769505e-05,
+      "loss": 0.1545,
+      "step": 9390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.328382247673587e-05,
+      "loss": 0.1079,
+      "step": 9391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.328203292770222e-05,
+      "loss": 0.1934,
+      "step": 9392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.328024337866858e-05,
+      "loss": 0.2491,
+      "step": 9393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.327845382963493e-05,
+      "loss": 0.1745,
+      "step": 9394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.327666428060129e-05,
+      "loss": 0.2165,
+      "step": 9395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.327487473156764e-05,
+      "loss": 0.2089,
+      "step": 9396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3273085182534005e-05,
+      "loss": 0.0825,
+      "step": 9397
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.327129563350036e-05,
+      "loss": 0.1849,
+      "step": 9398
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3269506084466715e-05,
+      "loss": 0.2354,
+      "step": 9399
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.326771653543308e-05,
+      "loss": 0.113,
+      "step": 9400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.326592698639943e-05,
+      "loss": 0.1582,
+      "step": 9401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.326413743736578e-05,
+      "loss": 0.2913,
+      "step": 9402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.326234788833214e-05,
+      "loss": 0.1302,
+      "step": 9403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.32605583392985e-05,
+      "loss": 0.1983,
+      "step": 9404
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.325876879026485e-05,
+      "loss": 0.1384,
+      "step": 9405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3256979241231215e-05,
+      "loss": 0.1103,
+      "step": 9406
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.325518969219757e-05,
+      "loss": 0.2077,
+      "step": 9407
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3253400143163926e-05,
+      "loss": 0.2589,
+      "step": 9408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.325161059413028e-05,
+      "loss": 0.1406,
+      "step": 9409
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3249821045096636e-05,
+      "loss": 0.1471,
+      "step": 9410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.324803149606299e-05,
+      "loss": 0.2705,
+      "step": 9411
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3246241947029346e-05,
+      "loss": 0.0813,
+      "step": 9412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.324445239799571e-05,
+      "loss": 0.201,
+      "step": 9413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3242662848962064e-05,
+      "loss": 0.24,
+      "step": 9414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.324087329992842e-05,
+      "loss": 0.2308,
+      "step": 9415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.323908375089478e-05,
+      "loss": 0.2337,
+      "step": 9416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.323729420186113e-05,
+      "loss": 0.1062,
+      "step": 9417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3235504652827484e-05,
+      "loss": 0.1816,
+      "step": 9418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3233715103793846e-05,
+      "loss": 0.2016,
+      "step": 9419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.32319255547602e-05,
+      "loss": 0.1788,
+      "step": 9420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.323013600572656e-05,
+      "loss": 0.2175,
+      "step": 9421
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.322834645669292e-05,
+      "loss": 0.1919,
+      "step": 9422
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3226556907659274e-05,
+      "loss": 0.2998,
+      "step": 9423
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.322476735862563e-05,
+      "loss": 0.2453,
+      "step": 9424
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3222977809591984e-05,
+      "loss": 0.1988,
+      "step": 9425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.322118826055834e-05,
+      "loss": 0.3316,
+      "step": 9426
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3219398711524695e-05,
+      "loss": 0.2775,
+      "step": 9427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3217609162491057e-05,
+      "loss": 0.1116,
+      "step": 9428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.321581961345741e-05,
+      "loss": 0.1575,
+      "step": 9429
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.321403006442377e-05,
+      "loss": 0.2374,
+      "step": 9430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.321224051539012e-05,
+      "loss": 0.2613,
+      "step": 9431
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3210450966356484e-05,
+      "loss": 0.1779,
+      "step": 9432
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.320866141732283e-05,
+      "loss": 0.1974,
+      "step": 9433
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.320687186828919e-05,
+      "loss": 0.1765,
+      "step": 9434
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.320508231925555e-05,
+      "loss": 0.1209,
+      "step": 9435
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3203292770221905e-05,
+      "loss": 0.2535,
+      "step": 9436
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.320150322118826e-05,
+      "loss": 0.1507,
+      "step": 9437
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319971367215462e-05,
+      "loss": 0.1409,
+      "step": 9438
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319792412312098e-05,
+      "loss": 0.1856,
+      "step": 9439
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319613457408733e-05,
+      "loss": 0.133,
+      "step": 9440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319434502505369e-05,
+      "loss": 0.1159,
+      "step": 9441
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319255547602004e-05,
+      "loss": 0.1715,
+      "step": 9442
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31907659269864e-05,
+      "loss": 0.1803,
+      "step": 9443
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318897637795276e-05,
+      "loss": 0.2406,
+      "step": 9444
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3187186828919115e-05,
+      "loss": 0.2805,
+      "step": 9445
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318539727988547e-05,
+      "loss": 0.1499,
+      "step": 9446
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3183607730851826e-05,
+      "loss": 0.1486,
+      "step": 9447
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318181818181819e-05,
+      "loss": 0.2058,
+      "step": 9448
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3180028632784536e-05,
+      "loss": 0.168,
+      "step": 9449
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31782390837509e-05,
+      "loss": 0.2114,
+      "step": 9450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.317644953471725e-05,
+      "loss": 0.1707,
+      "step": 9451
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.317465998568361e-05,
+      "loss": 0.2278,
+      "step": 9452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3172870436649964e-05,
+      "loss": 0.0836,
+      "step": 9453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3171080887616325e-05,
+      "loss": 0.2158,
+      "step": 9454
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.316929133858268e-05,
+      "loss": 0.1862,
+      "step": 9455
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3167501789549036e-05,
+      "loss": 0.2236,
+      "step": 9456
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.316571224051539e-05,
+      "loss": 0.1881,
+      "step": 9457
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3163922691481746e-05,
+      "loss": 0.184,
+      "step": 9458
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31621331424481e-05,
+      "loss": 0.1061,
+      "step": 9459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3160343593414463e-05,
+      "loss": 0.1539,
+      "step": 9460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.315855404438082e-05,
+      "loss": 0.1092,
+      "step": 9461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3156764495347174e-05,
+      "loss": 0.1428,
+      "step": 9462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3154974946313536e-05,
+      "loss": 0.1382,
+      "step": 9463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3153185397279884e-05,
+      "loss": 0.3321,
+      "step": 9464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.315139584824624e-05,
+      "loss": 0.1501,
+      "step": 9465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31496062992126e-05,
+      "loss": 0.1239,
+      "step": 9466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3147816750178957e-05,
+      "loss": 0.2529,
+      "step": 9467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.314602720114531e-05,
+      "loss": 0.3493,
+      "step": 9468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3144237652111674e-05,
+      "loss": 0.1172,
+      "step": 9469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.314244810307803e-05,
+      "loss": 0.2428,
+      "step": 9470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3140658554044384e-05,
+      "loss": 0.1815,
+      "step": 9471
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.313886900501074e-05,
+      "loss": 0.124,
+      "step": 9472
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3137079455977094e-05,
+      "loss": 0.113,
+      "step": 9473
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.313528990694345e-05,
+      "loss": 0.0838,
+      "step": 9474
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3133500357909805e-05,
+      "loss": 0.2749,
+      "step": 9475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.313171080887617e-05,
+      "loss": 0.2306,
+      "step": 9476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312992125984252e-05,
+      "loss": 0.1067,
+      "step": 9477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312813171080888e-05,
+      "loss": 0.1779,
+      "step": 9478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312634216177524e-05,
+      "loss": 0.2681,
+      "step": 9479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312455261274159e-05,
+      "loss": 0.19,
+      "step": 9480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.312276306370794e-05,
+      "loss": 0.1184,
+      "step": 9481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3120973514674305e-05,
+      "loss": 0.15,
+      "step": 9482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.311918396564066e-05,
+      "loss": 0.0433,
+      "step": 9483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3117394416607015e-05,
+      "loss": 0.2456,
+      "step": 9484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.311560486757338e-05,
+      "loss": 0.2027,
+      "step": 9485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.311381531853973e-05,
+      "loss": 0.1823,
+      "step": 9486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.311202576950609e-05,
+      "loss": 0.3775,
+      "step": 9487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.311023622047244e-05,
+      "loss": 0.154,
+      "step": 9488
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.31084466714388e-05,
+      "loss": 0.1699,
+      "step": 9489
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.310665712240515e-05,
+      "loss": 0.2072,
+      "step": 9490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3104867573371515e-05,
+      "loss": 0.177,
+      "step": 9491
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.310307802433787e-05,
+      "loss": 0.2592,
+      "step": 9492
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3101288475304225e-05,
+      "loss": 0.1691,
+      "step": 9493
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309949892627058e-05,
+      "loss": 0.1782,
+      "step": 9494
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309770937723694e-05,
+      "loss": 0.1994,
+      "step": 9495
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309591982820329e-05,
+      "loss": 0.2044,
+      "step": 9496
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3094130279169646e-05,
+      "loss": 0.2205,
+      "step": 9497
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309234073013601e-05,
+      "loss": 0.1563,
+      "step": 9498
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309055118110236e-05,
+      "loss": 0.1682,
+      "step": 9499
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.308876163206872e-05,
+      "loss": 0.172,
+      "step": 9500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.308697208303508e-05,
+      "loss": 0.1544,
+      "step": 9501
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3085182534001436e-05,
+      "loss": 0.0992,
+      "step": 9502
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3083392984967784e-05,
+      "loss": 0.2037,
+      "step": 9503
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3081603435934146e-05,
+      "loss": 0.1573,
+      "step": 9504
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.30798138869005e-05,
+      "loss": 0.2252,
+      "step": 9505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3078024337866856e-05,
+      "loss": 0.1695,
+      "step": 9506
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.307623478883322e-05,
+      "loss": 0.0909,
+      "step": 9507
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3074445239799574e-05,
+      "loss": 0.2375,
+      "step": 9508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.307265569076593e-05,
+      "loss": 0.148,
+      "step": 9509
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3070866141732284e-05,
+      "loss": 0.267,
+      "step": 9510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.306907659269864e-05,
+      "loss": 0.0808,
+      "step": 9511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3067287043664994e-05,
+      "loss": 0.1608,
+      "step": 9512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3065497494631356e-05,
+      "loss": 0.2486,
+      "step": 9513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.306370794559771e-05,
+      "loss": 0.1692,
+      "step": 9514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.306191839656407e-05,
+      "loss": 0.2849,
+      "step": 9515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.306012884753042e-05,
+      "loss": 0.1827,
+      "step": 9516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3058339298496784e-05,
+      "loss": 0.1554,
+      "step": 9517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.305654974946314e-05,
+      "loss": 0.2787,
+      "step": 9518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3054760200429494e-05,
+      "loss": 0.0921,
+      "step": 9519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.305297065139585e-05,
+      "loss": 0.2189,
+      "step": 9520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3051181102362205e-05,
+      "loss": 0.1719,
+      "step": 9521
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304939155332856e-05,
+      "loss": 0.0581,
+      "step": 9522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304760200429492e-05,
+      "loss": 0.146,
+      "step": 9523
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304581245526128e-05,
+      "loss": 0.1864,
+      "step": 9524
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304402290622763e-05,
+      "loss": 0.2013,
+      "step": 9525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3042233357193994e-05,
+      "loss": 0.2323,
+      "step": 9526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.304044380816034e-05,
+      "loss": 0.1688,
+      "step": 9527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.30386542591267e-05,
+      "loss": 0.1819,
+      "step": 9528
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.303686471009306e-05,
+      "loss": 0.2369,
+      "step": 9529
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3035075161059415e-05,
+      "loss": 0.2295,
+      "step": 9530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.303328561202577e-05,
+      "loss": 0.1572,
+      "step": 9531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.303149606299213e-05,
+      "loss": 0.1211,
+      "step": 9532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302970651395849e-05,
+      "loss": 0.143,
+      "step": 9533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302791696492484e-05,
+      "loss": 0.215,
+      "step": 9534
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.30261274158912e-05,
+      "loss": 0.1492,
+      "step": 9535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302433786685755e-05,
+      "loss": 0.1937,
+      "step": 9536
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302254831782391e-05,
+      "loss": 0.101,
+      "step": 9537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.302075876879026e-05,
+      "loss": 0.2281,
+      "step": 9538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3018969219756625e-05,
+      "loss": 0.0883,
+      "step": 9539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.301717967072298e-05,
+      "loss": 0.1654,
+      "step": 9540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3015390121689336e-05,
+      "loss": 0.1291,
+      "step": 9541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.301360057265569e-05,
+      "loss": 0.1113,
+      "step": 9542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3011811023622046e-05,
+      "loss": 0.1964,
+      "step": 9543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.30100214745884e-05,
+      "loss": 0.2196,
+      "step": 9544
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.300823192555476e-05,
+      "loss": 0.1632,
+      "step": 9545
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.300644237652112e-05,
+      "loss": 0.2094,
+      "step": 9546
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3004652827487474e-05,
+      "loss": 0.3158,
+      "step": 9547
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3002863278453836e-05,
+      "loss": 0.1563,
+      "step": 9548
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.300107372942019e-05,
+      "loss": 0.1214,
+      "step": 9549
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.299928418038654e-05,
+      "loss": 0.1012,
+      "step": 9550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.29974946313529e-05,
+      "loss": 0.221,
+      "step": 9551
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2995705082319256e-05,
+      "loss": 0.1106,
+      "step": 9552
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.299391553328561e-05,
+      "loss": 0.1262,
+      "step": 9553
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2992125984251974e-05,
+      "loss": 0.1319,
+      "step": 9554
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.299033643521833e-05,
+      "loss": 0.0654,
+      "step": 9555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2988546886184684e-05,
+      "loss": 0.1571,
+      "step": 9556
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.298675733715104e-05,
+      "loss": 0.1972,
+      "step": 9557
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2984967788117394e-05,
+      "loss": 0.1587,
+      "step": 9558
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.298317823908375e-05,
+      "loss": 0.179,
+      "step": 9559
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2981388690050105e-05,
+      "loss": 0.085,
+      "step": 9560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297959914101647e-05,
+      "loss": 0.215,
+      "step": 9561
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297780959198282e-05,
+      "loss": 0.0708,
+      "step": 9562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297602004294918e-05,
+      "loss": 0.2305,
+      "step": 9563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297423049391554e-05,
+      "loss": 0.3472,
+      "step": 9564
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2972440944881894e-05,
+      "loss": 0.3675,
+      "step": 9565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297065139584824e-05,
+      "loss": 0.1028,
+      "step": 9566
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2968861846814605e-05,
+      "loss": 0.1224,
+      "step": 9567
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.296707229778096e-05,
+      "loss": 0.1215,
+      "step": 9568
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2965282748747315e-05,
+      "loss": 0.1388,
+      "step": 9569
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.296349319971368e-05,
+      "loss": 0.1715,
+      "step": 9570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.296170365068003e-05,
+      "loss": 0.1158,
+      "step": 9571
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.295991410164639e-05,
+      "loss": 0.3734,
+      "step": 9572
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.295812455261274e-05,
+      "loss": 0.1373,
+      "step": 9573
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.29563350035791e-05,
+      "loss": 0.3204,
+      "step": 9574
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 0.1421,
+      "step": 9575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2952755905511815e-05,
+      "loss": 0.1709,
+      "step": 9576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.295096635647817e-05,
+      "loss": 0.3087,
+      "step": 9577
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2949176807444525e-05,
+      "loss": 0.1328,
+      "step": 9578
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.294738725841088e-05,
+      "loss": 0.1043,
+      "step": 9579
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.294559770937724e-05,
+      "loss": 0.1622,
+      "step": 9580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.29438081603436e-05,
+      "loss": 0.0875,
+      "step": 9581
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2942018611309946e-05,
+      "loss": 0.1324,
+      "step": 9582
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.294022906227631e-05,
+      "loss": 0.1283,
+      "step": 9583
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.293843951324266e-05,
+      "loss": 0.0569,
+      "step": 9584
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.293664996420902e-05,
+      "loss": 0.1611,
+      "step": 9585
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.293486041517538e-05,
+      "loss": 0.2398,
+      "step": 9586
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2933070866141736e-05,
+      "loss": 0.1566,
+      "step": 9587
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.293128131710809e-05,
+      "loss": 0.1639,
+      "step": 9588
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2929491768074446e-05,
+      "loss": 0.1804,
+      "step": 9589
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.29277022190408e-05,
+      "loss": 0.3488,
+      "step": 9590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2925912670007156e-05,
+      "loss": 0.0783,
+      "step": 9591
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.292412312097352e-05,
+      "loss": 0.1345,
+      "step": 9592
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2922333571939873e-05,
+      "loss": 0.0913,
+      "step": 9593
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.292054402290623e-05,
+      "loss": 0.1075,
+      "step": 9594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.291875447387259e-05,
+      "loss": 0.1558,
+      "step": 9595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2916964924838946e-05,
+      "loss": 0.1765,
+      "step": 9596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2915175375805294e-05,
+      "loss": 0.1883,
+      "step": 9597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2913385826771656e-05,
+      "loss": 0.2339,
+      "step": 9598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.291159627773801e-05,
+      "loss": 0.0937,
+      "step": 9599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2909806728704367e-05,
+      "loss": 0.1808,
+      "step": 9600
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.290801717967072e-05,
+      "loss": 0.1113,
+      "step": 9601
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2906227630637084e-05,
+      "loss": 0.2923,
+      "step": 9602
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.290443808160344e-05,
+      "loss": 0.2536,
+      "step": 9603
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2902648532569794e-05,
+      "loss": 0.2136,
+      "step": 9604
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.290085898353615e-05,
+      "loss": 0.4707,
+      "step": 9605
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2899069434502505e-05,
+      "loss": 0.2326,
+      "step": 9606
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289727988546886e-05,
+      "loss": 0.157,
+      "step": 9607
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289549033643522e-05,
+      "loss": 0.1471,
+      "step": 9608
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289370078740158e-05,
+      "loss": 0.1602,
+      "step": 9609
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289191123836793e-05,
+      "loss": 0.1683,
+      "step": 9610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2890121689334294e-05,
+      "loss": 0.1025,
+      "step": 9611
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.288833214030065e-05,
+      "loss": 0.0502,
+      "step": 9612
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2886542591267e-05,
+      "loss": 0.1346,
+      "step": 9613
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.288475304223336e-05,
+      "loss": 0.2551,
+      "step": 9614
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2882963493199715e-05,
+      "loss": 0.0562,
+      "step": 9615
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.288117394416607e-05,
+      "loss": 0.2285,
+      "step": 9616
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287938439513243e-05,
+      "loss": 0.1954,
+      "step": 9617
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287759484609879e-05,
+      "loss": 0.1596,
+      "step": 9618
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287580529706514e-05,
+      "loss": 0.1855,
+      "step": 9619
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.28740157480315e-05,
+      "loss": 0.1952,
+      "step": 9620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287222619899785e-05,
+      "loss": 0.1132,
+      "step": 9621
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287043664996421e-05,
+      "loss": 0.1375,
+      "step": 9622
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.286864710093056e-05,
+      "loss": 0.1939,
+      "step": 9623
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2866857551896925e-05,
+      "loss": 0.2481,
+      "step": 9624
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.286506800286328e-05,
+      "loss": 0.2216,
+      "step": 9625
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2863278453829635e-05,
+      "loss": 0.2264,
+      "step": 9626
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2861488904796e-05,
+      "loss": 0.1587,
+      "step": 9627
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2859699355762346e-05,
+      "loss": 0.2394,
+      "step": 9628
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.28579098067287e-05,
+      "loss": 0.282,
+      "step": 9629
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.285612025769506e-05,
+      "loss": 0.1303,
+      "step": 9630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.285433070866142e-05,
+      "loss": 0.2712,
+      "step": 9631
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2852541159627773e-05,
+      "loss": 0.133,
+      "step": 9632
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2850751610594135e-05,
+      "loss": 0.0957,
+      "step": 9633
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.284896206156049e-05,
+      "loss": 0.1835,
+      "step": 9634
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2847172512526846e-05,
+      "loss": 0.2758,
+      "step": 9635
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.28453829634932e-05,
+      "loss": 0.1647,
+      "step": 9636
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2843593414459556e-05,
+      "loss": 0.0913,
+      "step": 9637
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.284180386542591e-05,
+      "loss": 0.0835,
+      "step": 9638
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.284001431639227e-05,
+      "loss": 0.1607,
+      "step": 9639
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.283822476735863e-05,
+      "loss": 0.1836,
+      "step": 9640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2836435218324984e-05,
+      "loss": 0.1739,
+      "step": 9641
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.283464566929134e-05,
+      "loss": 0.1745,
+      "step": 9642
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.28328561202577e-05,
+      "loss": 0.2549,
+      "step": 9643
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.283106657122405e-05,
+      "loss": 0.2245,
+      "step": 9644
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2829277022190404e-05,
+      "loss": 0.1342,
+      "step": 9645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2827487473156766e-05,
+      "loss": 0.1361,
+      "step": 9646
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.282569792412312e-05,
+      "loss": 0.178,
+      "step": 9647
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.282390837508948e-05,
+      "loss": 0.1784,
+      "step": 9648
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.282211882605584e-05,
+      "loss": 0.1127,
+      "step": 9649
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2820329277022194e-05,
+      "loss": 0.1435,
+      "step": 9650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.281853972798855e-05,
+      "loss": 0.1895,
+      "step": 9651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2816750178954904e-05,
+      "loss": 0.2481,
+      "step": 9652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.281496062992126e-05,
+      "loss": 0.291,
+      "step": 9653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2813171080887615e-05,
+      "loss": 0.2615,
+      "step": 9654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.281138153185398e-05,
+      "loss": 0.2215,
+      "step": 9655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.280959198282033e-05,
+      "loss": 0.2373,
+      "step": 9656
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.280780243378669e-05,
+      "loss": 0.2232,
+      "step": 9657
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.280601288475304e-05,
+      "loss": 0.1091,
+      "step": 9658
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2804223335719404e-05,
+      "loss": 0.1286,
+      "step": 9659
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.280243378668575e-05,
+      "loss": 0.0731,
+      "step": 9660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2800644237652115e-05,
+      "loss": 0.2026,
+      "step": 9661
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.279885468861847e-05,
+      "loss": 0.1467,
+      "step": 9662
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2797065139584825e-05,
+      "loss": 0.1819,
+      "step": 9663
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.279527559055118e-05,
+      "loss": 0.1524,
+      "step": 9664
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.279348604151754e-05,
+      "loss": 0.1474,
+      "step": 9665
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.27916964924839e-05,
+      "loss": 0.2185,
+      "step": 9666
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278990694345025e-05,
+      "loss": 0.1655,
+      "step": 9667
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278811739441661e-05,
+      "loss": 0.2802,
+      "step": 9668
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278632784538296e-05,
+      "loss": 0.1569,
+      "step": 9669
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278453829634932e-05,
+      "loss": 0.4594,
+      "step": 9670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278274874731568e-05,
+      "loss": 0.1899,
+      "step": 9671
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2780959198282035e-05,
+      "loss": 0.1135,
+      "step": 9672
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277916964924839e-05,
+      "loss": 0.1064,
+      "step": 9673
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277738010021475e-05,
+      "loss": 0.1286,
+      "step": 9674
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.27755905511811e-05,
+      "loss": 0.3377,
+      "step": 9675
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2773801002147456e-05,
+      "loss": 0.1435,
+      "step": 9676
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277201145311382e-05,
+      "loss": 0.1258,
+      "step": 9677
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277022190408017e-05,
+      "loss": 0.2508,
+      "step": 9678
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.276843235504653e-05,
+      "loss": 0.2619,
+      "step": 9679
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.276664280601289e-05,
+      "loss": 0.162,
+      "step": 9680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2764853256979246e-05,
+      "loss": 0.2095,
+      "step": 9681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.27630637079456e-05,
+      "loss": 0.2031,
+      "step": 9682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2761274158911956e-05,
+      "loss": 0.274,
+      "step": 9683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.275948460987831e-05,
+      "loss": 0.11,
+      "step": 9684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2757695060844666e-05,
+      "loss": 0.1086,
+      "step": 9685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.275590551181102e-05,
+      "loss": 0.1877,
+      "step": 9686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2754115962777384e-05,
+      "loss": 0.1757,
+      "step": 9687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.275232641374374e-05,
+      "loss": 0.2049,
+      "step": 9688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2750536864710094e-05,
+      "loss": 0.1896,
+      "step": 9689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2748747315676456e-05,
+      "loss": 0.1333,
+      "step": 9690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2746957766642804e-05,
+      "loss": 0.0969,
+      "step": 9691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.274516821760916e-05,
+      "loss": 0.1028,
+      "step": 9692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.274337866857552e-05,
+      "loss": 0.1964,
+      "step": 9693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.274158911954188e-05,
+      "loss": 0.1809,
+      "step": 9694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.273979957050823e-05,
+      "loss": 0.1722,
+      "step": 9695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2738010021474594e-05,
+      "loss": 0.1188,
+      "step": 9696
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.273622047244095e-05,
+      "loss": 0.1397,
+      "step": 9697
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2734430923407304e-05,
+      "loss": 0.1264,
+      "step": 9698
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.273264137437366e-05,
+      "loss": 0.16,
+      "step": 9699
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2730851825340015e-05,
+      "loss": 0.2,
+      "step": 9700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.272906227630637e-05,
+      "loss": 0.2349,
+      "step": 9701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 0.0841,
+      "step": 9702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.272548317823909e-05,
+      "loss": 0.2236,
+      "step": 9703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.272369362920544e-05,
+      "loss": 0.085,
+      "step": 9704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.27219040801718e-05,
+      "loss": 0.1831,
+      "step": 9705
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.272011453113816e-05,
+      "loss": 0.2135,
+      "step": 9706
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.271832498210451e-05,
+      "loss": 0.1072,
+      "step": 9707
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.271653543307086e-05,
+      "loss": 0.3805,
+      "step": 9708
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2714745884037225e-05,
+      "loss": 0.0932,
+      "step": 9709
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.271295633500358e-05,
+      "loss": 0.1785,
+      "step": 9710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2711166785969935e-05,
+      "loss": 0.1129,
+      "step": 9711
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.27093772369363e-05,
+      "loss": 0.2099,
+      "step": 9712
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270758768790265e-05,
+      "loss": 0.1554,
+      "step": 9713
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270579813886901e-05,
+      "loss": 0.3101,
+      "step": 9714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270400858983536e-05,
+      "loss": 0.1124,
+      "step": 9715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270221904080172e-05,
+      "loss": 0.104,
+      "step": 9716
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270042949176807e-05,
+      "loss": 0.1733,
+      "step": 9717
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2698639942734435e-05,
+      "loss": 0.151,
+      "step": 9718
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.269685039370079e-05,
+      "loss": 0.1193,
+      "step": 9719
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2695060844667146e-05,
+      "loss": 0.2909,
+      "step": 9720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.26932712956335e-05,
+      "loss": 0.134,
+      "step": 9721
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2691481746599856e-05,
+      "loss": 0.1496,
+      "step": 9722
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268969219756621e-05,
+      "loss": 0.1508,
+      "step": 9723
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268790264853257e-05,
+      "loss": 0.3374,
+      "step": 9724
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268611309949893e-05,
+      "loss": 0.2255,
+      "step": 9725
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2684323550465284e-05,
+      "loss": 0.0833,
+      "step": 9726
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268253400143164e-05,
+      "loss": 0.1266,
+      "step": 9727
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2680744452398e-05,
+      "loss": 0.1297,
+      "step": 9728
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2678954903364356e-05,
+      "loss": 0.0768,
+      "step": 9729
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2677165354330704e-05,
+      "loss": 0.2076,
+      "step": 9730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2675375805297066e-05,
+      "loss": 0.1954,
+      "step": 9731
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.267358625626342e-05,
+      "loss": 0.0509,
+      "step": 9732
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.267179670722978e-05,
+      "loss": 0.1864,
+      "step": 9733
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.267000715819614e-05,
+      "loss": 0.1763,
+      "step": 9734
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2668217609162494e-05,
+      "loss": 0.1667,
+      "step": 9735
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.266642806012885e-05,
+      "loss": 0.118,
+      "step": 9736
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.266463851109521e-05,
+      "loss": 0.0877,
+      "step": 9737
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.266284896206156e-05,
+      "loss": 0.1554,
+      "step": 9738
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2661059413027915e-05,
+      "loss": 0.1529,
+      "step": 9739
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2659269863994277e-05,
+      "loss": 0.1691,
+      "step": 9740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.265748031496063e-05,
+      "loss": 0.2197,
+      "step": 9741
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.265569076592699e-05,
+      "loss": 0.2478,
+      "step": 9742
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.265390121689335e-05,
+      "loss": 0.1098,
+      "step": 9743
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2652111667859704e-05,
+      "loss": 0.1132,
+      "step": 9744
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.265032211882606e-05,
+      "loss": 0.2329,
+      "step": 9745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2648532569792414e-05,
+      "loss": 0.2173,
+      "step": 9746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.264674302075877e-05,
+      "loss": 0.1949,
+      "step": 9747
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2644953471725125e-05,
+      "loss": 0.2421,
+      "step": 9748
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.264316392269148e-05,
+      "loss": 0.1717,
+      "step": 9749
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.264137437365784e-05,
+      "loss": 0.0505,
+      "step": 9750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.26395848246242e-05,
+      "loss": 0.1574,
+      "step": 9751
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.263779527559055e-05,
+      "loss": 0.1923,
+      "step": 9752
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2636005726556914e-05,
+      "loss": 0.2315,
+      "step": 9753
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.263421617752326e-05,
+      "loss": 0.1115,
+      "step": 9754
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.263242662848962e-05,
+      "loss": 0.1145,
+      "step": 9755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.263063707945598e-05,
+      "loss": 0.1772,
+      "step": 9756
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2628847530422335e-05,
+      "loss": 0.2011,
+      "step": 9757
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.262705798138869e-05,
+      "loss": 0.0784,
+      "step": 9758
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.262526843235505e-05,
+      "loss": 0.1566,
+      "step": 9759
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.262347888332141e-05,
+      "loss": 0.1697,
+      "step": 9760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2621689334287756e-05,
+      "loss": 0.0922,
+      "step": 9761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.261989978525412e-05,
+      "loss": 0.1293,
+      "step": 9762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.261811023622047e-05,
+      "loss": 0.1384,
+      "step": 9763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.261632068718683e-05,
+      "loss": 0.1999,
+      "step": 9764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.261453113815319e-05,
+      "loss": 0.18,
+      "step": 9765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2612741589119545e-05,
+      "loss": 0.2018,
+      "step": 9766
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.26109520400859e-05,
+      "loss": 0.227,
+      "step": 9767
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2609162491052256e-05,
+      "loss": 0.3978,
+      "step": 9768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.260737294201861e-05,
+      "loss": 0.1331,
+      "step": 9769
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2605583392984966e-05,
+      "loss": 0.0893,
+      "step": 9770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.260379384395132e-05,
+      "loss": 0.1634,
+      "step": 9771
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2602004294917683e-05,
+      "loss": 0.2089,
+      "step": 9772
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.260021474588404e-05,
+      "loss": 0.1145,
+      "step": 9773
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2598425196850394e-05,
+      "loss": 0.2663,
+      "step": 9774
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2596635647816756e-05,
+      "loss": 0.1241,
+      "step": 9775
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.259484609878311e-05,
+      "loss": 0.2254,
+      "step": 9776
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.259305654974946e-05,
+      "loss": 0.2769,
+      "step": 9777
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.259126700071582e-05,
+      "loss": 0.1743,
+      "step": 9778
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2589477451682177e-05,
+      "loss": 0.2644,
+      "step": 9779
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258768790264853e-05,
+      "loss": 0.141,
+      "step": 9780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2585898353614894e-05,
+      "loss": 0.1551,
+      "step": 9781
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258410880458125e-05,
+      "loss": 0.294,
+      "step": 9782
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2582319255547604e-05,
+      "loss": 0.2229,
+      "step": 9783
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258052970651396e-05,
+      "loss": 0.1726,
+      "step": 9784
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2578740157480314e-05,
+      "loss": 0.1906,
+      "step": 9785
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257695060844667e-05,
+      "loss": 0.1335,
+      "step": 9786
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257516105941303e-05,
+      "loss": 0.2055,
+      "step": 9787
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257337151037939e-05,
+      "loss": 0.2422,
+      "step": 9788
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257158196134574e-05,
+      "loss": 0.1727,
+      "step": 9789
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.25697924123121e-05,
+      "loss": 0.3047,
+      "step": 9790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.256800286327846e-05,
+      "loss": 0.0749,
+      "step": 9791
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2566213314244814e-05,
+      "loss": 0.2784,
+      "step": 9792
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.256442376521116e-05,
+      "loss": 0.1516,
+      "step": 9793
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2562634216177525e-05,
+      "loss": 0.1106,
+      "step": 9794
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.256084466714388e-05,
+      "loss": 0.1293,
+      "step": 9795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2559055118110235e-05,
+      "loss": 0.1218,
+      "step": 9796
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.25572655690766e-05,
+      "loss": 0.2286,
+      "step": 9797
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.255547602004295e-05,
+      "loss": 0.1394,
+      "step": 9798
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.255368647100931e-05,
+      "loss": 0.1896,
+      "step": 9799
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.255189692197567e-05,
+      "loss": 0.1436,
+      "step": 9800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.255010737294202e-05,
+      "loss": 0.0792,
+      "step": 9801
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.254831782390837e-05,
+      "loss": 0.068,
+      "step": 9802
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2546528274874735e-05,
+      "loss": 0.1171,
+      "step": 9803
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.254473872584109e-05,
+      "loss": 0.1733,
+      "step": 9804
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2542949176807445e-05,
+      "loss": 0.1627,
+      "step": 9805
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.254115962777381e-05,
+      "loss": 0.2014,
+      "step": 9806
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253937007874016e-05,
+      "loss": 0.1667,
+      "step": 9807
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253758052970651e-05,
+      "loss": 0.0992,
+      "step": 9808
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253579098067287e-05,
+      "loss": 0.3071,
+      "step": 9809
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253400143163923e-05,
+      "loss": 0.1408,
+      "step": 9810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253221188260558e-05,
+      "loss": 0.2703,
+      "step": 9811
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.253042233357194e-05,
+      "loss": 0.1936,
+      "step": 9812
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.25286327845383e-05,
+      "loss": 0.1736,
+      "step": 9813
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2526843235504656e-05,
+      "loss": 0.1815,
+      "step": 9814
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.252505368647101e-05,
+      "loss": 0.0852,
+      "step": 9815
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2523264137437366e-05,
+      "loss": 0.1234,
+      "step": 9816
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.252147458840372e-05,
+      "loss": 0.1621,
+      "step": 9817
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2519685039370076e-05,
+      "loss": 0.2724,
+      "step": 9818
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.251789549033644e-05,
+      "loss": 0.198,
+      "step": 9819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2516105941302794e-05,
+      "loss": 0.1646,
+      "step": 9820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.251431639226915e-05,
+      "loss": 0.2889,
+      "step": 9821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.251252684323551e-05,
+      "loss": 0.0754,
+      "step": 9822
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2510737294201866e-05,
+      "loss": 0.1249,
+      "step": 9823
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2508947745168214e-05,
+      "loss": 0.1752,
+      "step": 9824
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2507158196134576e-05,
+      "loss": 0.2646,
+      "step": 9825
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.250536864710093e-05,
+      "loss": 0.3785,
+      "step": 9826
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.250357909806729e-05,
+      "loss": 0.2832,
+      "step": 9827
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.250178954903365e-05,
+      "loss": 0.1575,
+      "step": 9828
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.2194,
+      "step": 9829
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.249821045096636e-05,
+      "loss": 0.1388,
+      "step": 9830
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2496420901932714e-05,
+      "loss": 0.1571,
+      "step": 9831
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.249463135289907e-05,
+      "loss": 0.1364,
+      "step": 9832
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2492841803865425e-05,
+      "loss": 0.2733,
+      "step": 9833
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.249105225483178e-05,
+      "loss": 0.119,
+      "step": 9834
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.248926270579814e-05,
+      "loss": 0.1879,
+      "step": 9835
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.24874731567645e-05,
+      "loss": 0.164,
+      "step": 9836
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.248568360773085e-05,
+      "loss": 0.1986,
+      "step": 9837
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2483894058697214e-05,
+      "loss": 0.0812,
+      "step": 9838
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.248210450966357e-05,
+      "loss": 0.1895,
+      "step": 9839
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.248031496062992e-05,
+      "loss": 0.1507,
+      "step": 9840
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247852541159628e-05,
+      "loss": 0.1185,
+      "step": 9841
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2476735862562635e-05,
+      "loss": 0.1909,
+      "step": 9842
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247494631352899e-05,
+      "loss": 0.1021,
+      "step": 9843
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247315676449535e-05,
+      "loss": 0.2859,
+      "step": 9844
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247136721546171e-05,
+      "loss": 0.1339,
+      "step": 9845
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.246957766642806e-05,
+      "loss": 0.2278,
+      "step": 9846
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.246778811739442e-05,
+      "loss": 0.1531,
+      "step": 9847
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.246599856836077e-05,
+      "loss": 0.1569,
+      "step": 9848
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.246420901932713e-05,
+      "loss": 0.3555,
+      "step": 9849
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.246241947029349e-05,
+      "loss": 0.1867,
+      "step": 9850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2460629921259845e-05,
+      "loss": 0.1139,
+      "step": 9851
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.24588403722262e-05,
+      "loss": 0.0947,
+      "step": 9852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2457050823192556e-05,
+      "loss": 0.1846,
+      "step": 9853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.245526127415892e-05,
+      "loss": 0.1606,
+      "step": 9854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2453471725125266e-05,
+      "loss": 0.1527,
+      "step": 9855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.245168217609162e-05,
+      "loss": 0.0904,
+      "step": 9856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.244989262705798e-05,
+      "loss": 0.1134,
+      "step": 9857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.244810307802434e-05,
+      "loss": 0.0717,
+      "step": 9858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2446313528990694e-05,
+      "loss": 0.0791,
+      "step": 9859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2444523979957056e-05,
+      "loss": 0.236,
+      "step": 9860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.244273443092341e-05,
+      "loss": 0.1481,
+      "step": 9861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2440944881889766e-05,
+      "loss": 0.1112,
+      "step": 9862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.243915533285612e-05,
+      "loss": 0.1121,
+      "step": 9863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2437365783822476e-05,
+      "loss": 0.1729,
+      "step": 9864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.243557623478883e-05,
+      "loss": 0.1479,
+      "step": 9865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2433786685755194e-05,
+      "loss": 0.1809,
+      "step": 9866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.243199713672155e-05,
+      "loss": 0.1937,
+      "step": 9867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2430207587687904e-05,
+      "loss": 0.2061,
+      "step": 9868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.242841803865426e-05,
+      "loss": 0.1261,
+      "step": 9869
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.242662848962062e-05,
+      "loss": 0.228,
+      "step": 9870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.242483894058697e-05,
+      "loss": 0.0878,
+      "step": 9871
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.242304939155333e-05,
+      "loss": 0.1278,
+      "step": 9872
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.242125984251969e-05,
+      "loss": 0.1699,
+      "step": 9873
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.241947029348604e-05,
+      "loss": 0.1455,
+      "step": 9874
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.24176807444524e-05,
+      "loss": 0.1081,
+      "step": 9875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.241589119541876e-05,
+      "loss": 0.1275,
+      "step": 9876
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2414101646385114e-05,
+      "loss": 0.1796,
+      "step": 9877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.241231209735147e-05,
+      "loss": 0.0721,
+      "step": 9878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2410522548317825e-05,
+      "loss": 0.1204,
+      "step": 9879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.240873299928418e-05,
+      "loss": 0.3113,
+      "step": 9880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2406943450250535e-05,
+      "loss": 0.0834,
+      "step": 9881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.24051539012169e-05,
+      "loss": 0.1579,
+      "step": 9882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.240336435218325e-05,
+      "loss": 0.1802,
+      "step": 9883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.240157480314961e-05,
+      "loss": 0.0508,
+      "step": 9884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.239978525411597e-05,
+      "loss": 0.1544,
+      "step": 9885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2397995705082324e-05,
+      "loss": 0.2947,
+      "step": 9886
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.239620615604867e-05,
+      "loss": 0.2507,
+      "step": 9887
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2394416607015035e-05,
+      "loss": 0.1767,
+      "step": 9888
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.239262705798139e-05,
+      "loss": 0.18,
+      "step": 9889
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2390837508947745e-05,
+      "loss": 0.1629,
+      "step": 9890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238904795991411e-05,
+      "loss": 0.1245,
+      "step": 9891
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238725841088046e-05,
+      "loss": 0.0334,
+      "step": 9892
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238546886184682e-05,
+      "loss": 0.2421,
+      "step": 9893
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238367931281317e-05,
+      "loss": 0.2081,
+      "step": 9894
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238188976377953e-05,
+      "loss": 0.3511,
+      "step": 9895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238010021474588e-05,
+      "loss": 0.3979,
+      "step": 9896
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.237831066571224e-05,
+      "loss": 0.3059,
+      "step": 9897
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.23765211166786e-05,
+      "loss": 0.1815,
+      "step": 9898
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2374731567644956e-05,
+      "loss": 0.0788,
+      "step": 9899
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.237294201861131e-05,
+      "loss": 0.0735,
+      "step": 9900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.237115246957767e-05,
+      "loss": 0.171,
+      "step": 9901
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.236936292054402e-05,
+      "loss": 0.0788,
+      "step": 9902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2367573371510376e-05,
+      "loss": 0.1616,
+      "step": 9903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.236578382247674e-05,
+      "loss": 0.1375,
+      "step": 9904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2363994273443093e-05,
+      "loss": 0.3109,
+      "step": 9905
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.236220472440945e-05,
+      "loss": 0.2063,
+      "step": 9906
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.236041517537581e-05,
+      "loss": 0.1945,
+      "step": 9907
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2358625626342166e-05,
+      "loss": 0.1658,
+      "step": 9908
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.235683607730852e-05,
+      "loss": 0.183,
+      "step": 9909
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2355046528274876e-05,
+      "loss": 0.3946,
+      "step": 9910
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.235325697924123e-05,
+      "loss": 0.1952,
+      "step": 9911
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2351467430207587e-05,
+      "loss": 0.134,
+      "step": 9912
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.234967788117395e-05,
+      "loss": 0.1146,
+      "step": 9913
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2347888332140304e-05,
+      "loss": 0.081,
+      "step": 9914
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.234609878310666e-05,
+      "loss": 0.188,
+      "step": 9915
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2344309234073014e-05,
+      "loss": 0.1351,
+      "step": 9916
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2342519685039376e-05,
+      "loss": 0.1601,
+      "step": 9917
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2340730136005725e-05,
+      "loss": 0.2405,
+      "step": 9918
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.233894058697208e-05,
+      "loss": 0.1372,
+      "step": 9919
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.233715103793844e-05,
+      "loss": 0.0618,
+      "step": 9920
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.23353614889048e-05,
+      "loss": 0.2011,
+      "step": 9921
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.233357193987115e-05,
+      "loss": 0.1539,
+      "step": 9922
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2331782390837514e-05,
+      "loss": 0.0882,
+      "step": 9923
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.232999284180387e-05,
+      "loss": 0.147,
+      "step": 9924
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2328203292770224e-05,
+      "loss": 0.1657,
+      "step": 9925
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.232641374373658e-05,
+      "loss": 0.1481,
+      "step": 9926
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2324624194702935e-05,
+      "loss": 0.2111,
+      "step": 9927
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.232283464566929e-05,
+      "loss": 0.1419,
+      "step": 9928
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.232104509663565e-05,
+      "loss": 0.1739,
+      "step": 9929
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231925554760201e-05,
+      "loss": 0.1672,
+      "step": 9930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231746599856836e-05,
+      "loss": 0.1432,
+      "step": 9931
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231567644953472e-05,
+      "loss": 0.2349,
+      "step": 9932
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231388690050108e-05,
+      "loss": 0.1409,
+      "step": 9933
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231209735146743e-05,
+      "loss": 0.1626,
+      "step": 9934
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.231030780243379e-05,
+      "loss": 0.0634,
+      "step": 9935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2308518253400145e-05,
+      "loss": 0.1916,
+      "step": 9936
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.23067287043665e-05,
+      "loss": 0.1723,
+      "step": 9937
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2304939155332855e-05,
+      "loss": 0.156,
+      "step": 9938
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.230314960629922e-05,
+      "loss": 0.3581,
+      "step": 9939
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.230136005726557e-05,
+      "loss": 0.0896,
+      "step": 9940
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.229957050823192e-05,
+      "loss": 0.1022,
+      "step": 9941
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.229778095919828e-05,
+      "loss": 0.1767,
+      "step": 9942
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.229599141016464e-05,
+      "loss": 0.1357,
+      "step": 9943
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2294201861130993e-05,
+      "loss": 0.2032,
+      "step": 9944
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2292412312097355e-05,
+      "loss": 0.3002,
+      "step": 9945
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.229062276306371e-05,
+      "loss": 0.1327,
+      "step": 9946
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2288833214030066e-05,
+      "loss": 0.2545,
+      "step": 9947
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228704366499643e-05,
+      "loss": 0.1102,
+      "step": 9948
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2285254115962776e-05,
+      "loss": 0.1903,
+      "step": 9949
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228346456692913e-05,
+      "loss": 0.2304,
+      "step": 9950
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228167501789549e-05,
+      "loss": 0.0765,
+      "step": 9951
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227988546886185e-05,
+      "loss": 0.1192,
+      "step": 9952
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2278095919828204e-05,
+      "loss": 0.1572,
+      "step": 9953
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2276306370794566e-05,
+      "loss": 0.1455,
+      "step": 9954
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227451682176092e-05,
+      "loss": 0.0791,
+      "step": 9955
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2272727272727276e-05,
+      "loss": 0.3465,
+      "step": 9956
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227093772369363e-05,
+      "loss": 0.2523,
+      "step": 9957
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2269148174659986e-05,
+      "loss": 0.1158,
+      "step": 9958
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.226735862562634e-05,
+      "loss": 0.1977,
+      "step": 9959
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.22655690765927e-05,
+      "loss": 0.064,
+      "step": 9960
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.226377952755906e-05,
+      "loss": 0.2029,
+      "step": 9961
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2261989978525414e-05,
+      "loss": 0.1615,
+      "step": 9962
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.226020042949177e-05,
+      "loss": 0.1261,
+      "step": 9963
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.225841088045813e-05,
+      "loss": 0.0758,
+      "step": 9964
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.225662133142448e-05,
+      "loss": 0.0887,
+      "step": 9965
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2254831782390835e-05,
+      "loss": 0.1897,
+      "step": 9966
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.22530422333572e-05,
+      "loss": 0.2475,
+      "step": 9967
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.225125268432355e-05,
+      "loss": 0.3073,
+      "step": 9968
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.224946313528991e-05,
+      "loss": 0.2258,
+      "step": 9969
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.224767358625627e-05,
+      "loss": 0.1338,
+      "step": 9970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2245884037222624e-05,
+      "loss": 0.1549,
+      "step": 9971
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.224409448818898e-05,
+      "loss": 0.0523,
+      "step": 9972
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2242304939155335e-05,
+      "loss": 0.2163,
+      "step": 9973
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.224051539012169e-05,
+      "loss": 0.3316,
+      "step": 9974
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2238725841088045e-05,
+      "loss": 0.1281,
+      "step": 9975
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.223693629205441e-05,
+      "loss": 0.0565,
+      "step": 9976
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.223514674302076e-05,
+      "loss": 0.0932,
+      "step": 9977
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.223335719398712e-05,
+      "loss": 0.1606,
+      "step": 9978
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.223156764495347e-05,
+      "loss": 0.1038,
+      "step": 9979
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.222977809591983e-05,
+      "loss": 0.1081,
+      "step": 9980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.222798854688618e-05,
+      "loss": 0.2615,
+      "step": 9981
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.222619899785254e-05,
+      "loss": 0.1627,
+      "step": 9982
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.22244094488189e-05,
+      "loss": 0.0919,
+      "step": 9983
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2222619899785255e-05,
+      "loss": 0.1632,
+      "step": 9984
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.222083035075161e-05,
+      "loss": 0.3887,
+      "step": 9985
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.221904080171797e-05,
+      "loss": 0.2438,
+      "step": 9986
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.221725125268433e-05,
+      "loss": 0.1047,
+      "step": 9987
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2215461703650676e-05,
+      "loss": 0.1624,
+      "step": 9988
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.221367215461704e-05,
+      "loss": 0.1829,
+      "step": 9989
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.221188260558339e-05,
+      "loss": 0.2471,
+      "step": 9990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.221009305654975e-05,
+      "loss": 0.2261,
+      "step": 9991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.220830350751611e-05,
+      "loss": 0.0768,
+      "step": 9992
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2206513958482466e-05,
+      "loss": 0.1939,
+      "step": 9993
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.220472440944882e-05,
+      "loss": 0.2229,
+      "step": 9994
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2202934860415176e-05,
+      "loss": 0.1757,
+      "step": 9995
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.220114531138153e-05,
+      "loss": 0.182,
+      "step": 9996
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2199355762347886e-05,
+      "loss": 0.1897,
+      "step": 9997
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.219756621331425e-05,
+      "loss": 0.1938,
+      "step": 9998
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2195776664280604e-05,
+      "loss": 0.2101,
+      "step": 9999
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.219398711524696e-05,
+      "loss": 0.1621,
+      "step": 10000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2192197566213314e-05,
+      "loss": 0.2126,
+      "step": 10001
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2190408017179676e-05,
+      "loss": 0.1303,
+      "step": 10002
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218861846814603e-05,
+      "loss": 0.1679,
+      "step": 10003
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218682891911238e-05,
+      "loss": 0.1641,
+      "step": 10004
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218503937007874e-05,
+      "loss": 0.1116,
+      "step": 10005
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.21832498210451e-05,
+      "loss": 0.1409,
+      "step": 10006
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218146027201145e-05,
+      "loss": 0.0983,
+      "step": 10007
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2179670722977814e-05,
+      "loss": 0.1113,
+      "step": 10008
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.217788117394417e-05,
+      "loss": 0.0906,
+      "step": 10009
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2176091624910524e-05,
+      "loss": 0.1701,
+      "step": 10010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2174302075876886e-05,
+      "loss": 0.2555,
+      "step": 10011
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2172512526843235e-05,
+      "loss": 0.1262,
+      "step": 10012
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.217072297780959e-05,
+      "loss": 0.1079,
+      "step": 10013
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216893342877595e-05,
+      "loss": 0.1049,
+      "step": 10014
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216714387974231e-05,
+      "loss": 0.2045,
+      "step": 10015
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216535433070866e-05,
+      "loss": 0.0692,
+      "step": 10016
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216356478167502e-05,
+      "loss": 0.2238,
+      "step": 10017
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216177523264138e-05,
+      "loss": 0.1332,
+      "step": 10018
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2159985683607735e-05,
+      "loss": 0.1455,
+      "step": 10019
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.215819613457409e-05,
+      "loss": 0.1369,
+      "step": 10020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2156406585540445e-05,
+      "loss": 0.405,
+      "step": 10021
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.21546170365068e-05,
+      "loss": 0.138,
+      "step": 10022
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2152827487473155e-05,
+      "loss": 0.0712,
+      "step": 10023
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.215103793843952e-05,
+      "loss": 0.2712,
+      "step": 10024
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214924838940587e-05,
+      "loss": 0.1647,
+      "step": 10025
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214745884037223e-05,
+      "loss": 0.0719,
+      "step": 10026
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214566929133858e-05,
+      "loss": 0.2316,
+      "step": 10027
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214387974230494e-05,
+      "loss": 0.0971,
+      "step": 10028
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214209019327129e-05,
+      "loss": 0.1136,
+      "step": 10029
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2140300644237655e-05,
+      "loss": 0.1295,
+      "step": 10030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.213851109520401e-05,
+      "loss": 0.1042,
+      "step": 10031
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2136721546170366e-05,
+      "loss": 0.2584,
+      "step": 10032
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.213493199713673e-05,
+      "loss": 0.0862,
+      "step": 10033
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.213314244810308e-05,
+      "loss": 0.2129,
+      "step": 10034
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.213135289906943e-05,
+      "loss": 0.0909,
+      "step": 10035
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.212956335003579e-05,
+      "loss": 0.2986,
+      "step": 10036
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.212777380100215e-05,
+      "loss": 0.1017,
+      "step": 10037
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2125984251968504e-05,
+      "loss": 0.1143,
+      "step": 10038
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2124194702934865e-05,
+      "loss": 0.2785,
+      "step": 10039
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.212240515390122e-05,
+      "loss": 0.071,
+      "step": 10040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2120615604867576e-05,
+      "loss": 0.2425,
+      "step": 10041
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.211882605583393e-05,
+      "loss": 0.2061,
+      "step": 10042
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2117036506800286e-05,
+      "loss": 0.1575,
+      "step": 10043
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.211524695776664e-05,
+      "loss": 0.0944,
+      "step": 10044
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2113457408733e-05,
+      "loss": 0.1554,
+      "step": 10045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.211166785969936e-05,
+      "loss": 0.1444,
+      "step": 10046
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2109878310665714e-05,
+      "loss": 0.1409,
+      "step": 10047
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.210808876163207e-05,
+      "loss": 0.1192,
+      "step": 10048
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.210629921259843e-05,
+      "loss": 0.3112,
+      "step": 10049
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2104509663564786e-05,
+      "loss": 0.1616,
+      "step": 10050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2102720114531135e-05,
+      "loss": 0.0693,
+      "step": 10051
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2100930565497497e-05,
+      "loss": 0.175,
+      "step": 10052
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209914101646385e-05,
+      "loss": 0.0928,
+      "step": 10053
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209735146743021e-05,
+      "loss": 0.1004,
+      "step": 10054
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209556191839657e-05,
+      "loss": 0.0932,
+      "step": 10055
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2093772369362924e-05,
+      "loss": 0.1883,
+      "step": 10056
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209198282032928e-05,
+      "loss": 0.2851,
+      "step": 10057
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2090193271295634e-05,
+      "loss": 0.1715,
+      "step": 10058
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208840372226199e-05,
+      "loss": 0.1543,
+      "step": 10059
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2086614173228345e-05,
+      "loss": 0.1982,
+      "step": 10060
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208482462419471e-05,
+      "loss": 0.1929,
+      "step": 10061
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208303507516106e-05,
+      "loss": 0.0988,
+      "step": 10062
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208124552612742e-05,
+      "loss": 0.2459,
+      "step": 10063
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.207945597709377e-05,
+      "loss": 0.1955,
+      "step": 10064
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2077666428060134e-05,
+      "loss": 0.2097,
+      "step": 10065
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.207587687902649e-05,
+      "loss": 0.3166,
+      "step": 10066
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.207408732999284e-05,
+      "loss": 0.1336,
+      "step": 10067
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.20722977809592e-05,
+      "loss": 0.1241,
+      "step": 10068
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2070508231925555e-05,
+      "loss": 0.3191,
+      "step": 10069
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206871868289191e-05,
+      "loss": 0.0621,
+      "step": 10070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206692913385827e-05,
+      "loss": 0.1161,
+      "step": 10071
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206513958482463e-05,
+      "loss": 0.1233,
+      "step": 10072
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206335003579098e-05,
+      "loss": 0.1143,
+      "step": 10073
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206156048675734e-05,
+      "loss": 0.1114,
+      "step": 10074
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.205977093772369e-05,
+      "loss": 0.3105,
+      "step": 10075
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.205798138869005e-05,
+      "loss": 0.1944,
+      "step": 10076
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.205619183965641e-05,
+      "loss": 0.1953,
+      "step": 10077
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2054402290622765e-05,
+      "loss": 0.1846,
+      "step": 10078
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.205261274158912e-05,
+      "loss": 0.0839,
+      "step": 10079
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2050823192555476e-05,
+      "loss": 0.2233,
+      "step": 10080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.204903364352184e-05,
+      "loss": 0.2295,
+      "step": 10081
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2047244094488186e-05,
+      "loss": 0.1635,
+      "step": 10082
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.204545454545455e-05,
+      "loss": 0.0845,
+      "step": 10083
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.20436649964209e-05,
+      "loss": 0.1651,
+      "step": 10084
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.204187544738726e-05,
+      "loss": 0.2229,
+      "step": 10085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2040085898353614e-05,
+      "loss": 0.0719,
+      "step": 10086
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2038296349319976e-05,
+      "loss": 0.255,
+      "step": 10087
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.203650680028633e-05,
+      "loss": 0.0721,
+      "step": 10088
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2034717251252686e-05,
+      "loss": 0.2042,
+      "step": 10089
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.203292770221904e-05,
+      "loss": 0.2582,
+      "step": 10090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2031138153185397e-05,
+      "loss": 0.0996,
+      "step": 10091
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.202934860415175e-05,
+      "loss": 0.1427,
+      "step": 10092
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2027559055118114e-05,
+      "loss": 0.1406,
+      "step": 10093
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.202576950608447e-05,
+      "loss": 0.1147,
+      "step": 10094
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2023979957050824e-05,
+      "loss": 0.169,
+      "step": 10095
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2022190408017186e-05,
+      "loss": 0.1874,
+      "step": 10096
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.202040085898354e-05,
+      "loss": 0.2442,
+      "step": 10097
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.201861130994989e-05,
+      "loss": 0.1473,
+      "step": 10098
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.201682176091625e-05,
+      "loss": 0.1165,
+      "step": 10099
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.201503221188261e-05,
+      "loss": 0.1867,
+      "step": 10100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.201324266284896e-05,
+      "loss": 0.1508,
+      "step": 10101
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2011453113815324e-05,
+      "loss": 0.1802,
+      "step": 10102
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.200966356478168e-05,
+      "loss": 0.1076,
+      "step": 10103
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2007874015748034e-05,
+      "loss": 0.1563,
+      "step": 10104
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.200608446671439e-05,
+      "loss": 0.1799,
+      "step": 10105
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2004294917680745e-05,
+      "loss": 0.0857,
+      "step": 10106
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.20025053686471e-05,
+      "loss": 0.2416,
+      "step": 10107
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2000715819613455e-05,
+      "loss": 0.0481,
+      "step": 10108
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199892627057982e-05,
+      "loss": 0.2026,
+      "step": 10109
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199713672154617e-05,
+      "loss": 0.1086,
+      "step": 10110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199534717251253e-05,
+      "loss": 0.1032,
+      "step": 10111
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199355762347889e-05,
+      "loss": 0.1987,
+      "step": 10112
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199176807444524e-05,
+      "loss": 0.1857,
+      "step": 10113
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.198997852541159e-05,
+      "loss": 0.2238,
+      "step": 10114
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1988188976377955e-05,
+      "loss": 0.1041,
+      "step": 10115
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.198639942734431e-05,
+      "loss": 0.1838,
+      "step": 10116
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1984609878310665e-05,
+      "loss": 0.1848,
+      "step": 10117
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.198282032927703e-05,
+      "loss": 0.1379,
+      "step": 10118
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.198103078024338e-05,
+      "loss": 0.1897,
+      "step": 10119
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197924123120974e-05,
+      "loss": 0.1659,
+      "step": 10120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197745168217609e-05,
+      "loss": 0.2275,
+      "step": 10121
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197566213314245e-05,
+      "loss": 0.1507,
+      "step": 10122
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.19738725841088e-05,
+      "loss": 0.1934,
+      "step": 10123
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1972083035075165e-05,
+      "loss": 0.1383,
+      "step": 10124
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197029348604152e-05,
+      "loss": 0.2824,
+      "step": 10125
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1968503937007876e-05,
+      "loss": 0.0627,
+      "step": 10126
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.196671438797423e-05,
+      "loss": 0.2676,
+      "step": 10127
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.196492483894059e-05,
+      "loss": 0.1897,
+      "step": 10128
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.196313528990694e-05,
+      "loss": 0.2218,
+      "step": 10129
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1961345740873296e-05,
+      "loss": 0.0898,
+      "step": 10130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.195955619183966e-05,
+      "loss": 0.1041,
+      "step": 10131
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1957766642806014e-05,
+      "loss": 0.2251,
+      "step": 10132
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.195597709377237e-05,
+      "loss": 0.3663,
+      "step": 10133
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.195418754473873e-05,
+      "loss": 0.1649,
+      "step": 10134
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1952397995705086e-05,
+      "loss": 0.1073,
+      "step": 10135
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.195060844667144e-05,
+      "loss": 0.0542,
+      "step": 10136
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1948818897637796e-05,
+      "loss": 0.1536,
+      "step": 10137
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.194702934860415e-05,
+      "loss": 0.139,
+      "step": 10138
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.194523979957051e-05,
+      "loss": 0.1555,
+      "step": 10139
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.194345025053687e-05,
+      "loss": 0.1352,
+      "step": 10140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1941660701503224e-05,
+      "loss": 0.1716,
+      "step": 10141
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.193987115246958e-05,
+      "loss": 0.1174,
+      "step": 10142
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1938081603435934e-05,
+      "loss": 0.1182,
+      "step": 10143
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1936292054402296e-05,
+      "loss": 0.0813,
+      "step": 10144
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1934502505368645e-05,
+      "loss": 0.243,
+      "step": 10145
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.193271295633501e-05,
+      "loss": 0.1824,
+      "step": 10146
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.193092340730136e-05,
+      "loss": 0.157,
+      "step": 10147
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.192913385826772e-05,
+      "loss": 0.1585,
+      "step": 10148
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.192734430923407e-05,
+      "loss": 0.4534,
+      "step": 10149
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1925554760200434e-05,
+      "loss": 0.2467,
+      "step": 10150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.192376521116679e-05,
+      "loss": 0.2037,
+      "step": 10151
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1921975662133145e-05,
+      "loss": 0.1522,
+      "step": 10152
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.19201861130995e-05,
+      "loss": 0.1346,
+      "step": 10153
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1918396564065855e-05,
+      "loss": 0.2354,
+      "step": 10154
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.191660701503221e-05,
+      "loss": 0.0569,
+      "step": 10155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.191481746599857e-05,
+      "loss": 0.1809,
+      "step": 10156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.191302791696493e-05,
+      "loss": 0.2353,
+      "step": 10157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.191123836793128e-05,
+      "loss": 0.1483,
+      "step": 10158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1909448818897644e-05,
+      "loss": 0.1848,
+      "step": 10159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.190765926986399e-05,
+      "loss": 0.126,
+      "step": 10160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.190586972083035e-05,
+      "loss": 0.2482,
+      "step": 10161
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.190408017179671e-05,
+      "loss": 0.2377,
+      "step": 10162
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1902290622763065e-05,
+      "loss": 0.2705,
+      "step": 10163
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.190050107372942e-05,
+      "loss": 0.2087,
+      "step": 10164
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189871152469578e-05,
+      "loss": 0.1033,
+      "step": 10165
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189692197566214e-05,
+      "loss": 0.2022,
+      "step": 10166
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189513242662849e-05,
+      "loss": 0.1214,
+      "step": 10167
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189334287759485e-05,
+      "loss": 0.162,
+      "step": 10168
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.18915533285612e-05,
+      "loss": 0.2534,
+      "step": 10169
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.188976377952756e-05,
+      "loss": 0.1461,
+      "step": 10170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1887974230493914e-05,
+      "loss": 0.0985,
+      "step": 10171
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1886184681460276e-05,
+      "loss": 0.2114,
+      "step": 10172
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.188439513242663e-05,
+      "loss": 0.1764,
+      "step": 10173
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1882605583392986e-05,
+      "loss": 0.1758,
+      "step": 10174
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.188081603435935e-05,
+      "loss": 0.134,
+      "step": 10175
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1879026485325696e-05,
+      "loss": 0.0949,
+      "step": 10176
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.187723693629205e-05,
+      "loss": 0.2335,
+      "step": 10177
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1875447387258413e-05,
+      "loss": 0.2862,
+      "step": 10178
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.187365783822477e-05,
+      "loss": 0.1286,
+      "step": 10179
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1871868289191124e-05,
+      "loss": 0.2187,
+      "step": 10180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1870078740157486e-05,
+      "loss": 0.1828,
+      "step": 10181
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.186828919112384e-05,
+      "loss": 0.0515,
+      "step": 10182
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1866499642090196e-05,
+      "loss": 0.2137,
+      "step": 10183
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.186471009305655e-05,
+      "loss": 0.1868,
+      "step": 10184
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.186292054402291e-05,
+      "loss": 0.2646,
+      "step": 10185
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.186113099498926e-05,
+      "loss": 0.1327,
+      "step": 10186
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1859341445955624e-05,
+      "loss": 0.1366,
+      "step": 10187
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.185755189692198e-05,
+      "loss": 0.1179,
+      "step": 10188
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1855762347888334e-05,
+      "loss": 0.1163,
+      "step": 10189
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.185397279885469e-05,
+      "loss": 0.1183,
+      "step": 10190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.185218324982105e-05,
+      "loss": 0.1686,
+      "step": 10191
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.18503937007874e-05,
+      "loss": 0.0978,
+      "step": 10192
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1848604151753755e-05,
+      "loss": 0.1655,
+      "step": 10193
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.184681460272012e-05,
+      "loss": 0.1631,
+      "step": 10194
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.184502505368647e-05,
+      "loss": 0.2409,
+      "step": 10195
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.184323550465283e-05,
+      "loss": 0.2052,
+      "step": 10196
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.184144595561919e-05,
+      "loss": 0.1995,
+      "step": 10197
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1839656406585544e-05,
+      "loss": 0.2162,
+      "step": 10198
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.183786685755189e-05,
+      "loss": 0.0886,
+      "step": 10199
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1836077308518255e-05,
+      "loss": 0.1589,
+      "step": 10200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.183428775948461e-05,
+      "loss": 0.4095,
+      "step": 10201
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1832498210450965e-05,
+      "loss": 0.216,
+      "step": 10202
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.183070866141733e-05,
+      "loss": 0.0392,
+      "step": 10203
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.182891911238368e-05,
+      "loss": 0.1798,
+      "step": 10204
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.182712956335004e-05,
+      "loss": 0.2261,
+      "step": 10205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.182534001431639e-05,
+      "loss": 0.0462,
+      "step": 10206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.182355046528275e-05,
+      "loss": 0.243,
+      "step": 10207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.18217609162491e-05,
+      "loss": 0.2335,
+      "step": 10208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1819971367215465e-05,
+      "loss": 0.2822,
+      "step": 10209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.1672,
+      "step": 10210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1816392269148176e-05,
+      "loss": 0.1372,
+      "step": 10211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.181460272011453e-05,
+      "loss": 0.3175,
+      "step": 10212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.181281317108089e-05,
+      "loss": 0.1382,
+      "step": 10213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.181102362204725e-05,
+      "loss": 0.2993,
+      "step": 10214
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1809234073013596e-05,
+      "loss": 0.2119,
+      "step": 10215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.180744452397996e-05,
+      "loss": 0.3128,
+      "step": 10216
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1805654974946313e-05,
+      "loss": 0.3378,
+      "step": 10217
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.180386542591267e-05,
+      "loss": 0.1713,
+      "step": 10218
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.180207587687903e-05,
+      "loss": 0.1334,
+      "step": 10219
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1800286327845386e-05,
+      "loss": 0.2203,
+      "step": 10220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.179849677881174e-05,
+      "loss": 0.1364,
+      "step": 10221
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.17967072297781e-05,
+      "loss": 0.2066,
+      "step": 10222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.179491768074445e-05,
+      "loss": 0.1895,
+      "step": 10223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1793128131710807e-05,
+      "loss": 0.1078,
+      "step": 10224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.179133858267717e-05,
+      "loss": 0.1575,
+      "step": 10225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1789549033643524e-05,
+      "loss": 0.1152,
+      "step": 10226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.178775948460988e-05,
+      "loss": 0.1088,
+      "step": 10227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1785969935576234e-05,
+      "loss": 0.14,
+      "step": 10228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1784180386542596e-05,
+      "loss": 0.2095,
+      "step": 10229
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.178239083750895e-05,
+      "loss": 0.2289,
+      "step": 10230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1780601288475306e-05,
+      "loss": 0.189,
+      "step": 10231
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177881173944166e-05,
+      "loss": 0.107,
+      "step": 10232
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177702219040802e-05,
+      "loss": 0.1593,
+      "step": 10233
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177523264137437e-05,
+      "loss": 0.165,
+      "step": 10234
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1773443092340734e-05,
+      "loss": 0.1385,
+      "step": 10235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177165354330709e-05,
+      "loss": 0.0892,
+      "step": 10236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1769863994273444e-05,
+      "loss": 0.1386,
+      "step": 10237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1768074445239806e-05,
+      "loss": 0.2991,
+      "step": 10238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1766284896206155e-05,
+      "loss": 0.2099,
+      "step": 10239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.176449534717251e-05,
+      "loss": 0.2164,
+      "step": 10240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.176270579813887e-05,
+      "loss": 0.1421,
+      "step": 10241
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.176091624910523e-05,
+      "loss": 0.1049,
+      "step": 10242
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.175912670007158e-05,
+      "loss": 0.1386,
+      "step": 10243
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1757337151037944e-05,
+      "loss": 0.1669,
+      "step": 10244
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.17555476020043e-05,
+      "loss": 0.2036,
+      "step": 10245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.175375805297065e-05,
+      "loss": 0.1313,
+      "step": 10246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.175196850393701e-05,
+      "loss": 0.1587,
+      "step": 10247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1750178954903365e-05,
+      "loss": 0.0899,
+      "step": 10248
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.174838940586972e-05,
+      "loss": 0.1504,
+      "step": 10249
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.174659985683608e-05,
+      "loss": 0.0632,
+      "step": 10250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.174481030780244e-05,
+      "loss": 0.1573,
+      "step": 10251
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.174302075876879e-05,
+      "loss": 0.1164,
+      "step": 10252
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.174123120973515e-05,
+      "loss": 0.1318,
+      "step": 10253
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.17394416607015e-05,
+      "loss": 0.3254,
+      "step": 10254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.173765211166786e-05,
+      "loss": 0.1559,
+      "step": 10255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1735862562634213e-05,
+      "loss": 0.1708,
+      "step": 10256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1734073013600575e-05,
+      "loss": 0.2484,
+      "step": 10257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.173228346456693e-05,
+      "loss": 0.3582,
+      "step": 10258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1730493915533286e-05,
+      "loss": 0.2451,
+      "step": 10259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.172870436649965e-05,
+      "loss": 0.077,
+      "step": 10260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1726914817466e-05,
+      "loss": 0.0967,
+      "step": 10261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.172512526843235e-05,
+      "loss": 0.1573,
+      "step": 10262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.172333571939871e-05,
+      "loss": 0.0994,
+      "step": 10263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.172154617036507e-05,
+      "loss": 0.073,
+      "step": 10264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1719756621331424e-05,
+      "loss": 0.283,
+      "step": 10265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1717967072297786e-05,
+      "loss": 0.1775,
+      "step": 10266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.171617752326414e-05,
+      "loss": 0.1777,
+      "step": 10267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1714387974230496e-05,
+      "loss": 0.1285,
+      "step": 10268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.171259842519685e-05,
+      "loss": 0.1511,
+      "step": 10269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1710808876163206e-05,
+      "loss": 0.1725,
+      "step": 10270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.170901932712956e-05,
+      "loss": 0.219,
+      "step": 10271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1707229778095924e-05,
+      "loss": 0.0779,
+      "step": 10272
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.170544022906228e-05,
+      "loss": 0.1507,
+      "step": 10273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1703650680028634e-05,
+      "loss": 0.0743,
+      "step": 10274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.170186113099499e-05,
+      "loss": 0.2897,
+      "step": 10275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.170007158196135e-05,
+      "loss": 0.1307,
+      "step": 10276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1698282032927706e-05,
+      "loss": 0.1375,
+      "step": 10277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1696492483894055e-05,
+      "loss": 0.2161,
+      "step": 10278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.169470293486042e-05,
+      "loss": 0.1209,
+      "step": 10279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.169291338582677e-05,
+      "loss": 0.1274,
+      "step": 10280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.169112383679313e-05,
+      "loss": 0.1667,
+      "step": 10281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.168933428775949e-05,
+      "loss": 0.1861,
+      "step": 10282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1687544738725844e-05,
+      "loss": 0.2154,
+      "step": 10283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.16857551896922e-05,
+      "loss": 0.1699,
+      "step": 10284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.168396564065856e-05,
+      "loss": 0.1079,
+      "step": 10285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.168217609162491e-05,
+      "loss": 0.1006,
+      "step": 10286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1680386542591265e-05,
+      "loss": 0.1193,
+      "step": 10287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167859699355763e-05,
+      "loss": 0.1403,
+      "step": 10288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167680744452398e-05,
+      "loss": 0.1594,
+      "step": 10289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167501789549034e-05,
+      "loss": 0.0863,
+      "step": 10290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167322834645669e-05,
+      "loss": 0.1778,
+      "step": 10291
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1671438797423055e-05,
+      "loss": 0.2092,
+      "step": 10292
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.16696492483894e-05,
+      "loss": 0.2903,
+      "step": 10293
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1667859699355765e-05,
+      "loss": 0.2207,
+      "step": 10294
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166607015032212e-05,
+      "loss": 0.1347,
+      "step": 10295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1664280601288475e-05,
+      "loss": 0.1369,
+      "step": 10296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166249105225483e-05,
+      "loss": 0.1642,
+      "step": 10297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166070150322119e-05,
+      "loss": 0.1017,
+      "step": 10298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.165891195418755e-05,
+      "loss": 0.1031,
+      "step": 10299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.16571224051539e-05,
+      "loss": 0.1635,
+      "step": 10300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.165533285612026e-05,
+      "loss": 0.142,
+      "step": 10301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.165354330708661e-05,
+      "loss": 0.1123,
+      "step": 10302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.165175375805297e-05,
+      "loss": 0.2317,
+      "step": 10303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.164996420901933e-05,
+      "loss": 0.2939,
+      "step": 10304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1648174659985686e-05,
+      "loss": 0.1451,
+      "step": 10305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.164638511095204e-05,
+      "loss": 0.1166,
+      "step": 10306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.16445955619184e-05,
+      "loss": 0.1089,
+      "step": 10307
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.164280601288476e-05,
+      "loss": 0.0867,
+      "step": 10308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1641016463851106e-05,
+      "loss": 0.1516,
+      "step": 10309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.163922691481747e-05,
+      "loss": 0.1494,
+      "step": 10310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1637437365783824e-05,
+      "loss": 0.2714,
+      "step": 10311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.163564781675018e-05,
+      "loss": 0.1588,
+      "step": 10312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.163385826771654e-05,
+      "loss": 0.2215,
+      "step": 10313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1632068718682896e-05,
+      "loss": 0.1672,
+      "step": 10314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.163027916964925e-05,
+      "loss": 0.1906,
+      "step": 10315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1628489620615606e-05,
+      "loss": 0.1378,
+      "step": 10316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.162670007158196e-05,
+      "loss": 0.1315,
+      "step": 10317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.162491052254832e-05,
+      "loss": 0.3137,
+      "step": 10318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.162312097351467e-05,
+      "loss": 0.154,
+      "step": 10319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1621331424481034e-05,
+      "loss": 0.2025,
+      "step": 10320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.161954187544739e-05,
+      "loss": 0.1576,
+      "step": 10321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1617752326413744e-05,
+      "loss": 0.2167,
+      "step": 10322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1615962777380106e-05,
+      "loss": 0.1436,
+      "step": 10323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.161417322834646e-05,
+      "loss": 0.1532,
+      "step": 10324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.161238367931281e-05,
+      "loss": 0.2481,
+      "step": 10325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.161059413027917e-05,
+      "loss": 0.2245,
+      "step": 10326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.160880458124553e-05,
+      "loss": 0.1535,
+      "step": 10327
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.160701503221188e-05,
+      "loss": 0.3085,
+      "step": 10328
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1605225483178244e-05,
+      "loss": 0.1493,
+      "step": 10329
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.16034359341446e-05,
+      "loss": 0.2111,
+      "step": 10330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1601646385110955e-05,
+      "loss": 0.2096,
+      "step": 10331
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159985683607731e-05,
+      "loss": 0.137,
+      "step": 10332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1598067287043665e-05,
+      "loss": 0.1236,
+      "step": 10333
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159627773801002e-05,
+      "loss": 0.1755,
+      "step": 10334
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159448818897638e-05,
+      "loss": 0.1798,
+      "step": 10335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159269863994274e-05,
+      "loss": 0.1528,
+      "step": 10336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159090909090909e-05,
+      "loss": 0.1582,
+      "step": 10337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158911954187545e-05,
+      "loss": 0.1265,
+      "step": 10338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158732999284181e-05,
+      "loss": 0.1176,
+      "step": 10339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158554044380816e-05,
+      "loss": 0.1717,
+      "step": 10340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158375089477451e-05,
+      "loss": 0.211,
+      "step": 10341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1581961345740875e-05,
+      "loss": 0.1217,
+      "step": 10342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158017179670723e-05,
+      "loss": 0.1242,
+      "step": 10343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1578382247673586e-05,
+      "loss": 0.1533,
+      "step": 10344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157659269863995e-05,
+      "loss": 0.0452,
+      "step": 10345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.15748031496063e-05,
+      "loss": 0.3378,
+      "step": 10346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157301360057266e-05,
+      "loss": 0.3004,
+      "step": 10347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157122405153901e-05,
+      "loss": 0.0913,
+      "step": 10348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.156943450250537e-05,
+      "loss": 0.1146,
+      "step": 10349
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1567644953471724e-05,
+      "loss": 0.469,
+      "step": 10350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1565855404438085e-05,
+      "loss": 0.1489,
+      "step": 10351
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.156406585540444e-05,
+      "loss": 0.1674,
+      "step": 10352
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1562276306370796e-05,
+      "loss": 0.2032,
+      "step": 10353
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.156048675733715e-05,
+      "loss": 0.2352,
+      "step": 10354
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.155869720830351e-05,
+      "loss": 0.0775,
+      "step": 10355
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.155690765926986e-05,
+      "loss": 0.2021,
+      "step": 10356
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1555118110236223e-05,
+      "loss": 0.2156,
+      "step": 10357
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.155332856120258e-05,
+      "loss": 0.2412,
+      "step": 10358
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1551539012168934e-05,
+      "loss": 0.2343,
+      "step": 10359
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.154974946313529e-05,
+      "loss": 0.0742,
+      "step": 10360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.154795991410165e-05,
+      "loss": 0.1725,
+      "step": 10361
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1546170365068006e-05,
+      "loss": 0.1034,
+      "step": 10362
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.154438081603436e-05,
+      "loss": 0.1897,
+      "step": 10363
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1542591267000717e-05,
+      "loss": 0.1186,
+      "step": 10364
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.154080171796707e-05,
+      "loss": 0.1158,
+      "step": 10365
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.153901216893343e-05,
+      "loss": 0.0902,
+      "step": 10366
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.153722261989979e-05,
+      "loss": 0.2255,
+      "step": 10367
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1535433070866144e-05,
+      "loss": 0.1076,
+      "step": 10368
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.15336435218325e-05,
+      "loss": 0.21,
+      "step": 10369
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.153185397279886e-05,
+      "loss": 0.1236,
+      "step": 10370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1530064423765216e-05,
+      "loss": 0.2753,
+      "step": 10371
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1528274874731565e-05,
+      "loss": 0.0688,
+      "step": 10372
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.152648532569793e-05,
+      "loss": 0.1896,
+      "step": 10373
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.152469577666428e-05,
+      "loss": 0.1065,
+      "step": 10374
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.152290622763064e-05,
+      "loss": 0.1662,
+      "step": 10375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.152111667859699e-05,
+      "loss": 0.126,
+      "step": 10376
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1519327129563354e-05,
+      "loss": 0.1585,
+      "step": 10377
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.151753758052971e-05,
+      "loss": 0.1878,
+      "step": 10378
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1515748031496065e-05,
+      "loss": 0.1602,
+      "step": 10379
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.151395848246242e-05,
+      "loss": 0.1089,
+      "step": 10380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1512168933428775e-05,
+      "loss": 0.3991,
+      "step": 10381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.151037938439513e-05,
+      "loss": 0.0866,
+      "step": 10382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.150858983536149e-05,
+      "loss": 0.1636,
+      "step": 10383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.150680028632785e-05,
+      "loss": 0.1613,
+      "step": 10384
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.15050107372942e-05,
+      "loss": 0.2658,
+      "step": 10385
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1503221188260565e-05,
+      "loss": 0.1587,
+      "step": 10386
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.150143163922691e-05,
+      "loss": 0.2206,
+      "step": 10387
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.149964209019327e-05,
+      "loss": 0.1086,
+      "step": 10388
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.149785254115963e-05,
+      "loss": 0.1057,
+      "step": 10389
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1496062992125985e-05,
+      "loss": 0.2039,
+      "step": 10390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.149427344309234e-05,
+      "loss": 0.0854,
+      "step": 10391
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.14924838940587e-05,
+      "loss": 0.0876,
+      "step": 10392
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.149069434502506e-05,
+      "loss": 0.1476,
+      "step": 10393
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148890479599141e-05,
+      "loss": 0.1143,
+      "step": 10394
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148711524695777e-05,
+      "loss": 0.2328,
+      "step": 10395
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148532569792412e-05,
+      "loss": 0.3347,
+      "step": 10396
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148353614889048e-05,
+      "loss": 0.0899,
+      "step": 10397
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148174659985684e-05,
+      "loss": 0.1701,
+      "step": 10398
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1479957050823196e-05,
+      "loss": 0.0959,
+      "step": 10399
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147816750178955e-05,
+      "loss": 0.2777,
+      "step": 10400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1476377952755906e-05,
+      "loss": 0.1365,
+      "step": 10401
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147458840372227e-05,
+      "loss": 0.1035,
+      "step": 10402
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1472798854688617e-05,
+      "loss": 0.1417,
+      "step": 10403
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147100930565497e-05,
+      "loss": 0.1862,
+      "step": 10404
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1469219756621334e-05,
+      "loss": 0.2013,
+      "step": 10405
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.146743020758769e-05,
+      "loss": 0.2176,
+      "step": 10406
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1465640658554044e-05,
+      "loss": 0.2658,
+      "step": 10407
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1463851109520406e-05,
+      "loss": 0.2473,
+      "step": 10408
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.146206156048676e-05,
+      "loss": 0.2007,
+      "step": 10409
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1460272011453116e-05,
+      "loss": 0.2335,
+      "step": 10410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.145848246241947e-05,
+      "loss": 0.126,
+      "step": 10411
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.145669291338583e-05,
+      "loss": 0.0965,
+      "step": 10412
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.145490336435218e-05,
+      "loss": 0.0757,
+      "step": 10413
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1453113815318544e-05,
+      "loss": 0.1154,
+      "step": 10414
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.14513242662849e-05,
+      "loss": 0.0925,
+      "step": 10415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1449534717251254e-05,
+      "loss": 0.1774,
+      "step": 10416
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.144774516821761e-05,
+      "loss": 0.1006,
+      "step": 10417
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1445955619183965e-05,
+      "loss": 0.1636,
+      "step": 10418
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.144416607015032e-05,
+      "loss": 0.2563,
+      "step": 10419
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.144237652111668e-05,
+      "loss": 0.1311,
+      "step": 10420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.144058697208304e-05,
+      "loss": 0.1733,
+      "step": 10421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.143879742304939e-05,
+      "loss": 0.2317,
+      "step": 10422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.143700787401575e-05,
+      "loss": 0.4351,
+      "step": 10423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.143521832498211e-05,
+      "loss": 0.1134,
+      "step": 10424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1433428775948465e-05,
+      "loss": 0.1097,
+      "step": 10425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.143163922691481e-05,
+      "loss": 0.1432,
+      "step": 10426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1429849677881175e-05,
+      "loss": 0.1598,
+      "step": 10427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.142806012884753e-05,
+      "loss": 0.2938,
+      "step": 10428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1426270579813885e-05,
+      "loss": 0.2597,
+      "step": 10429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.142448103078025e-05,
+      "loss": 0.1874,
+      "step": 10430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.14226914817466e-05,
+      "loss": 0.1502,
+      "step": 10431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.142090193271296e-05,
+      "loss": 0.1525,
+      "step": 10432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.141911238367932e-05,
+      "loss": 0.3233,
+      "step": 10433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.141732283464567e-05,
+      "loss": 0.239,
+      "step": 10434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.141553328561202e-05,
+      "loss": 0.263,
+      "step": 10435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1413743736578385e-05,
+      "loss": 0.1877,
+      "step": 10436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.141195418754474e-05,
+      "loss": 0.2733,
+      "step": 10437
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1410164638511096e-05,
+      "loss": 0.076,
+      "step": 10438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.140837508947745e-05,
+      "loss": 0.1614,
+      "step": 10439
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.140658554044381e-05,
+      "loss": 0.1277,
+      "step": 10440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.140479599141017e-05,
+      "loss": 0.2152,
+      "step": 10441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.140300644237652e-05,
+      "loss": 0.2453,
+      "step": 10442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.140121689334288e-05,
+      "loss": 0.1678,
+      "step": 10443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1399427344309234e-05,
+      "loss": 0.1371,
+      "step": 10444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139763779527559e-05,
+      "loss": 0.199,
+      "step": 10445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139584824624195e-05,
+      "loss": 0.2486,
+      "step": 10446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1394058697208306e-05,
+      "loss": 0.1303,
+      "step": 10447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139226914817466e-05,
+      "loss": 0.2957,
+      "step": 10448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139047959914102e-05,
+      "loss": 0.2657,
+      "step": 10449
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.138869005010737e-05,
+      "loss": 0.2214,
+      "step": 10450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.138690050107373e-05,
+      "loss": 0.1663,
+      "step": 10451
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.138511095204009e-05,
+      "loss": 0.129,
+      "step": 10452
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1383321403006444e-05,
+      "loss": 0.0974,
+      "step": 10453
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13815318539728e-05,
+      "loss": 0.1362,
+      "step": 10454
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137974230493916e-05,
+      "loss": 0.1907,
+      "step": 10455
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1377952755905516e-05,
+      "loss": 0.1004,
+      "step": 10456
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137616320687187e-05,
+      "loss": 0.0814,
+      "step": 10457
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137437365783823e-05,
+      "loss": 0.1651,
+      "step": 10458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137258410880458e-05,
+      "loss": 0.1027,
+      "step": 10459
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137079455977094e-05,
+      "loss": 0.1135,
+      "step": 10460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13690050107373e-05,
+      "loss": 0.1074,
+      "step": 10461
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1367215461703654e-05,
+      "loss": 0.2501,
+      "step": 10462
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.136542591267001e-05,
+      "loss": 0.1683,
+      "step": 10463
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1363636363636365e-05,
+      "loss": 0.1271,
+      "step": 10464
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.136184681460272e-05,
+      "loss": 0.1228,
+      "step": 10465
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1360057265569075e-05,
+      "loss": 0.3168,
+      "step": 10466
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.135826771653543e-05,
+      "loss": 0.1593,
+      "step": 10467
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.135647816750179e-05,
+      "loss": 0.274,
+      "step": 10468
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.135468861846815e-05,
+      "loss": 0.2501,
+      "step": 10469
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13528990694345e-05,
+      "loss": 0.1167,
+      "step": 10470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1351109520400864e-05,
+      "loss": 0.3299,
+      "step": 10471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.134931997136722e-05,
+      "loss": 0.1359,
+      "step": 10472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.134753042233357e-05,
+      "loss": 0.2701,
+      "step": 10473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.134574087329993e-05,
+      "loss": 0.2593,
+      "step": 10474
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1343951324266285e-05,
+      "loss": 0.0979,
+      "step": 10475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.134216177523264e-05,
+      "loss": 0.0955,
+      "step": 10476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1340372226199e-05,
+      "loss": 0.1036,
+      "step": 10477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133858267716536e-05,
+      "loss": 0.1601,
+      "step": 10478
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133679312813171e-05,
+      "loss": 0.2106,
+      "step": 10479
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133500357909807e-05,
+      "loss": 0.2553,
+      "step": 10480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133321403006442e-05,
+      "loss": 0.2985,
+      "step": 10481
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.133142448103078e-05,
+      "loss": 0.207,
+      "step": 10482
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.132963493199714e-05,
+      "loss": 0.3322,
+      "step": 10483
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1327845382963496e-05,
+      "loss": 0.1497,
+      "step": 10484
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.132605583392985e-05,
+      "loss": 0.1513,
+      "step": 10485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1324266284896206e-05,
+      "loss": 0.2108,
+      "step": 10486
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.132247673586257e-05,
+      "loss": 0.2137,
+      "step": 10487
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.132068718682892e-05,
+      "loss": 0.3004,
+      "step": 10488
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.131889763779527e-05,
+      "loss": 0.1449,
+      "step": 10489
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1317108088761633e-05,
+      "loss": 0.2807,
+      "step": 10490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.131531853972799e-05,
+      "loss": 0.1778,
+      "step": 10491
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1313528990694344e-05,
+      "loss": 0.1528,
+      "step": 10492
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1311739441660706e-05,
+      "loss": 0.2813,
+      "step": 10493
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.130994989262706e-05,
+      "loss": 0.065,
+      "step": 10494
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1308160343593416e-05,
+      "loss": 0.1424,
+      "step": 10495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.130637079455978e-05,
+      "loss": 0.2061,
+      "step": 10496
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1304581245526127e-05,
+      "loss": 0.1995,
+      "step": 10497
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.130279169649248e-05,
+      "loss": 0.1118,
+      "step": 10498
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1301002147458844e-05,
+      "loss": 0.1572,
+      "step": 10499
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.12992125984252e-05,
+      "loss": 0.1097,
+      "step": 10500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1297423049391554e-05,
+      "loss": 0.3584,
+      "step": 10501
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.129563350035791e-05,
+      "loss": 0.1107,
+      "step": 10502
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.129384395132427e-05,
+      "loss": 0.2486,
+      "step": 10503
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1292054402290627e-05,
+      "loss": 0.2864,
+      "step": 10504
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.129026485325698e-05,
+      "loss": 0.1972,
+      "step": 10505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.128847530422334e-05,
+      "loss": 0.2295,
+      "step": 10506
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.128668575518969e-05,
+      "loss": 0.133,
+      "step": 10507
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.128489620615605e-05,
+      "loss": 0.2347,
+      "step": 10508
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.128310665712241e-05,
+      "loss": 0.2214,
+      "step": 10509
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1281317108088764e-05,
+      "loss": 0.1088,
+      "step": 10510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.127952755905512e-05,
+      "loss": 0.2804,
+      "step": 10511
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1277738010021475e-05,
+      "loss": 0.2352,
+      "step": 10512
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.127594846098783e-05,
+      "loss": 0.1584,
+      "step": 10513
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1274158911954185e-05,
+      "loss": 0.2572,
+      "step": 10514
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.127236936292055e-05,
+      "loss": 0.18,
+      "step": 10515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.12705798138869e-05,
+      "loss": 0.1417,
+      "step": 10516
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.126879026485326e-05,
+      "loss": 0.0784,
+      "step": 10517
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.126700071581962e-05,
+      "loss": 0.2069,
+      "step": 10518
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1265211166785975e-05,
+      "loss": 0.17,
+      "step": 10519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.126342161775232e-05,
+      "loss": 0.17,
+      "step": 10520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1261632068718685e-05,
+      "loss": 0.0647,
+      "step": 10521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.125984251968504e-05,
+      "loss": 0.2061,
+      "step": 10522
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1258052970651396e-05,
+      "loss": 0.1334,
+      "step": 10523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.125626342161776e-05,
+      "loss": 0.221,
+      "step": 10524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.125447387258411e-05,
+      "loss": 0.1817,
+      "step": 10525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.125268432355047e-05,
+      "loss": 0.136,
+      "step": 10526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.125089477451682e-05,
+      "loss": 0.0819,
+      "step": 10527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.124910522548318e-05,
+      "loss": 0.1059,
+      "step": 10528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1247315676449533e-05,
+      "loss": 0.2123,
+      "step": 10529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.124552612741589e-05,
+      "loss": 0.1224,
+      "step": 10530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.124373657838225e-05,
+      "loss": 0.1752,
+      "step": 10531
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1241947029348606e-05,
+      "loss": 0.1265,
+      "step": 10532
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.124015748031496e-05,
+      "loss": 0.089,
+      "step": 10533
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.123836793128132e-05,
+      "loss": 0.1586,
+      "step": 10534
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.123657838224768e-05,
+      "loss": 0.1752,
+      "step": 10535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1234788833214027e-05,
+      "loss": 0.1306,
+      "step": 10536
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.123299928418039e-05,
+      "loss": 0.1911,
+      "step": 10537
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1231209735146744e-05,
+      "loss": 0.0898,
+      "step": 10538
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.12294201861131e-05,
+      "loss": 0.1711,
+      "step": 10539
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.122763063707946e-05,
+      "loss": 0.1103,
+      "step": 10540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1225841088045816e-05,
+      "loss": 0.2541,
+      "step": 10541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.122405153901217e-05,
+      "loss": 0.1174,
+      "step": 10542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1222261989978526e-05,
+      "loss": 0.3639,
+      "step": 10543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.122047244094488e-05,
+      "loss": 0.2094,
+      "step": 10544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.121868289191124e-05,
+      "loss": 0.2287,
+      "step": 10545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.12168933428776e-05,
+      "loss": 0.342,
+      "step": 10546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1215103793843954e-05,
+      "loss": 0.2288,
+      "step": 10547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.121331424481031e-05,
+      "loss": 0.1855,
+      "step": 10548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1211524695776664e-05,
+      "loss": 0.1573,
+      "step": 10549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1209735146743026e-05,
+      "loss": 0.1841,
+      "step": 10550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1207945597709375e-05,
+      "loss": 0.1853,
+      "step": 10551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.120615604867573e-05,
+      "loss": 0.1142,
+      "step": 10552
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.120436649964209e-05,
+      "loss": 0.0981,
+      "step": 10553
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.120257695060845e-05,
+      "loss": 0.2018,
+      "step": 10554
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.12007874015748e-05,
+      "loss": 0.2782,
+      "step": 10555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1198997852541164e-05,
+      "loss": 0.1213,
+      "step": 10556
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.119720830350752e-05,
+      "loss": 0.0754,
+      "step": 10557
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1195418754473875e-05,
+      "loss": 0.1425,
+      "step": 10558
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.119362920544023e-05,
+      "loss": 0.08,
+      "step": 10559
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1191839656406585e-05,
+      "loss": 0.1368,
+      "step": 10560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.119005010737294e-05,
+      "loss": 0.2204,
+      "step": 10561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.11882605583393e-05,
+      "loss": 0.2816,
+      "step": 10562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118647100930566e-05,
+      "loss": 0.1968,
+      "step": 10563
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118468146027201e-05,
+      "loss": 0.1863,
+      "step": 10564
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118289191123837e-05,
+      "loss": 0.1379,
+      "step": 10565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118110236220473e-05,
+      "loss": 0.1207,
+      "step": 10566
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117931281317108e-05,
+      "loss": 0.1733,
+      "step": 10567
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117752326413744e-05,
+      "loss": 0.1488,
+      "step": 10568
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1175733715103795e-05,
+      "loss": 0.0876,
+      "step": 10569
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117394416607015e-05,
+      "loss": 0.1176,
+      "step": 10570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1172154617036506e-05,
+      "loss": 0.0805,
+      "step": 10571
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117036506800287e-05,
+      "loss": 0.2138,
+      "step": 10572
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.116857551896922e-05,
+      "loss": 0.0795,
+      "step": 10573
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.116678596993558e-05,
+      "loss": 0.1111,
+      "step": 10574
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.116499642090193e-05,
+      "loss": 0.1001,
+      "step": 10575
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.116320687186829e-05,
+      "loss": 0.0991,
+      "step": 10576
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1161417322834644e-05,
+      "loss": 0.1426,
+      "step": 10577
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1159627773801006e-05,
+      "loss": 0.0918,
+      "step": 10578
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.115783822476736e-05,
+      "loss": 0.2882,
+      "step": 10579
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1156048675733716e-05,
+      "loss": 0.1476,
+      "step": 10580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.115425912670008e-05,
+      "loss": 0.2232,
+      "step": 10581
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.115246957766643e-05,
+      "loss": 0.163,
+      "step": 10582
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.115068002863278e-05,
+      "loss": 0.1449,
+      "step": 10583
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1148890479599144e-05,
+      "loss": 0.2271,
+      "step": 10584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.11471009305655e-05,
+      "loss": 0.0587,
+      "step": 10585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1145311381531854e-05,
+      "loss": 0.2987,
+      "step": 10586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.114352183249821e-05,
+      "loss": 0.0857,
+      "step": 10587
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.114173228346457e-05,
+      "loss": 0.1537,
+      "step": 10588
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1139942734430926e-05,
+      "loss": 0.1637,
+      "step": 10589
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113815318539728e-05,
+      "loss": 0.0937,
+      "step": 10590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113636363636364e-05,
+      "loss": 0.2946,
+      "step": 10591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113457408732999e-05,
+      "loss": 0.2212,
+      "step": 10592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113278453829635e-05,
+      "loss": 0.0741,
+      "step": 10593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.113099498926271e-05,
+      "loss": 0.1814,
+      "step": 10594
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1129205440229064e-05,
+      "loss": 0.2071,
+      "step": 10595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.112741589119542e-05,
+      "loss": 0.0563,
+      "step": 10596
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.112562634216178e-05,
+      "loss": 0.1342,
+      "step": 10597
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.112383679312813e-05,
+      "loss": 0.047,
+      "step": 10598
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1122047244094485e-05,
+      "loss": 0.2873,
+      "step": 10599
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.112025769506085e-05,
+      "loss": 0.0734,
+      "step": 10600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.11184681460272e-05,
+      "loss": 0.2424,
+      "step": 10601
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.111667859699356e-05,
+      "loss": 0.1131,
+      "step": 10602
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.111488904795992e-05,
+      "loss": 0.1899,
+      "step": 10603
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1113099498926275e-05,
+      "loss": 0.2558,
+      "step": 10604
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.111130994989263e-05,
+      "loss": 0.1684,
+      "step": 10605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1109520400858985e-05,
+      "loss": 0.0982,
+      "step": 10606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.110773085182534e-05,
+      "loss": 0.1904,
+      "step": 10607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1105941302791695e-05,
+      "loss": 0.0423,
+      "step": 10608
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.110415175375806e-05,
+      "loss": 0.1986,
+      "step": 10609
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.110236220472441e-05,
+      "loss": 0.1806,
+      "step": 10610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.110057265569077e-05,
+      "loss": 0.1221,
+      "step": 10611
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109878310665712e-05,
+      "loss": 0.0945,
+      "step": 10612
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1096993557623485e-05,
+      "loss": 0.1997,
+      "step": 10613
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109520400858983e-05,
+      "loss": 0.1664,
+      "step": 10614
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109341445955619e-05,
+      "loss": 0.2524,
+      "step": 10615
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109162491052255e-05,
+      "loss": 0.1305,
+      "step": 10616
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1089835361488906e-05,
+      "loss": 0.2173,
+      "step": 10617
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108804581245526e-05,
+      "loss": 0.0882,
+      "step": 10618
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108625626342162e-05,
+      "loss": 0.1304,
+      "step": 10619
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108446671438798e-05,
+      "loss": 0.0881,
+      "step": 10620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108267716535433e-05,
+      "loss": 0.1564,
+      "step": 10621
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108088761632069e-05,
+      "loss": 0.2996,
+      "step": 10622
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1079098067287044e-05,
+      "loss": 0.0771,
+      "step": 10623
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.10773085182534e-05,
+      "loss": 0.3237,
+      "step": 10624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.107551896921976e-05,
+      "loss": 0.1736,
+      "step": 10625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1073729420186116e-05,
+      "loss": 0.2042,
+      "step": 10626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.107193987115247e-05,
+      "loss": 0.1507,
+      "step": 10627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1070150322118826e-05,
+      "loss": 0.0801,
+      "step": 10628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.106836077308519e-05,
+      "loss": 0.2495,
+      "step": 10629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.106657122405154e-05,
+      "loss": 0.1122,
+      "step": 10630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.10647816750179e-05,
+      "loss": 0.1654,
+      "step": 10631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1062992125984254e-05,
+      "loss": 0.1335,
+      "step": 10632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.106120257695061e-05,
+      "loss": 0.238,
+      "step": 10633
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1059413027916964e-05,
+      "loss": 0.1687,
+      "step": 10634
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1057623478883326e-05,
+      "loss": 0.1502,
+      "step": 10635
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.105583392984968e-05,
+      "loss": 0.2366,
+      "step": 10636
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1054044380816037e-05,
+      "loss": 0.2669,
+      "step": 10637
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.105225483178239e-05,
+      "loss": 0.1516,
+      "step": 10638
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.105046528274875e-05,
+      "loss": 0.1327,
+      "step": 10639
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.10486757337151e-05,
+      "loss": 0.1236,
+      "step": 10640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1046886184681464e-05,
+      "loss": 0.1563,
+      "step": 10641
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.104509663564782e-05,
+      "loss": 0.07,
+      "step": 10642
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1043307086614175e-05,
+      "loss": 0.2769,
+      "step": 10643
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1041517537580536e-05,
+      "loss": 0.0788,
+      "step": 10644
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1039727988546885e-05,
+      "loss": 0.1875,
+      "step": 10645
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.103793843951324e-05,
+      "loss": 0.1215,
+      "step": 10646
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.10361488904796e-05,
+      "loss": 0.1217,
+      "step": 10647
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.103435934144596e-05,
+      "loss": 0.2207,
+      "step": 10648
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.103256979241231e-05,
+      "loss": 0.2384,
+      "step": 10649
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.103078024337867e-05,
+      "loss": 0.2061,
+      "step": 10650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.102899069434503e-05,
+      "loss": 0.1665,
+      "step": 10651
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1027201145311385e-05,
+      "loss": 0.164,
+      "step": 10652
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.102541159627774e-05,
+      "loss": 0.2325,
+      "step": 10653
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1023622047244095e-05,
+      "loss": 0.1686,
+      "step": 10654
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.102183249821045e-05,
+      "loss": 0.1046,
+      "step": 10655
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1020042949176806e-05,
+      "loss": 0.164,
+      "step": 10656
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.101825340014317e-05,
+      "loss": 0.2465,
+      "step": 10657
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.101646385110952e-05,
+      "loss": 0.2336,
+      "step": 10658
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.101467430207588e-05,
+      "loss": 0.3485,
+      "step": 10659
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.101288475304224e-05,
+      "loss": 0.2716,
+      "step": 10660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.101109520400859e-05,
+      "loss": 0.2834,
+      "step": 10661
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1009305654974944e-05,
+      "loss": 0.1834,
+      "step": 10662
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1007516105941305e-05,
+      "loss": 0.1214,
+      "step": 10663
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.100572655690766e-05,
+      "loss": 0.2938,
+      "step": 10664
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1003937007874016e-05,
+      "loss": 0.1134,
+      "step": 10665
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100214745884038e-05,
+      "loss": 0.16,
+      "step": 10666
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100035790980673e-05,
+      "loss": 0.1386,
+      "step": 10667
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.099856836077309e-05,
+      "loss": 0.1819,
+      "step": 10668
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0996778811739443e-05,
+      "loss": 0.1726,
+      "step": 10669
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09949892627058e-05,
+      "loss": 0.1244,
+      "step": 10670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0993199713672154e-05,
+      "loss": 0.1077,
+      "step": 10671
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0991410164638516e-05,
+      "loss": 0.2711,
+      "step": 10672
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098962061560487e-05,
+      "loss": 0.2311,
+      "step": 10673
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0987831066571226e-05,
+      "loss": 0.1626,
+      "step": 10674
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098604151753758e-05,
+      "loss": 0.2219,
+      "step": 10675
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098425196850394e-05,
+      "loss": 0.2911,
+      "step": 10676
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098246241947029e-05,
+      "loss": 0.1795,
+      "step": 10677
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098067287043665e-05,
+      "loss": 0.2135,
+      "step": 10678
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.097888332140301e-05,
+      "loss": 0.2201,
+      "step": 10679
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0977093772369364e-05,
+      "loss": 0.1566,
+      "step": 10680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.097530422333572e-05,
+      "loss": 0.0975,
+      "step": 10681
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.097351467430208e-05,
+      "loss": 0.2281,
+      "step": 10682
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0971725125268436e-05,
+      "loss": 0.2377,
+      "step": 10683
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0969935576234785e-05,
+      "loss": 0.1428,
+      "step": 10684
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.096814602720115e-05,
+      "loss": 0.0876,
+      "step": 10685
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09663564781675e-05,
+      "loss": 0.1844,
+      "step": 10686
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.096456692913386e-05,
+      "loss": 0.3436,
+      "step": 10687
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.096277738010022e-05,
+      "loss": 0.0747,
+      "step": 10688
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0960987831066574e-05,
+      "loss": 0.14,
+      "step": 10689
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.095919828203293e-05,
+      "loss": 0.1794,
+      "step": 10690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0957408732999285e-05,
+      "loss": 0.1331,
+      "step": 10691
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.095561918396564e-05,
+      "loss": 0.1521,
+      "step": 10692
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0953829634931995e-05,
+      "loss": 0.1473,
+      "step": 10693
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.095204008589836e-05,
+      "loss": 0.1354,
+      "step": 10694
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.095025053686471e-05,
+      "loss": 0.3192,
+      "step": 10695
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.094846098783107e-05,
+      "loss": 0.2591,
+      "step": 10696
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.094667143879742e-05,
+      "loss": 0.1916,
+      "step": 10697
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0944881889763785e-05,
+      "loss": 0.1215,
+      "step": 10698
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.094309234073014e-05,
+      "loss": 0.219,
+      "step": 10699
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.094130279169649e-05,
+      "loss": 0.1898,
+      "step": 10700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.093951324266285e-05,
+      "loss": 0.1394,
+      "step": 10701
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0937723693629205e-05,
+      "loss": 0.0873,
+      "step": 10702
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.093593414459556e-05,
+      "loss": 0.2198,
+      "step": 10703
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.093414459556192e-05,
+      "loss": 0.1453,
+      "step": 10704
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.093235504652828e-05,
+      "loss": 0.077,
+      "step": 10705
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.093056549749463e-05,
+      "loss": 0.1624,
+      "step": 10706
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0928775948460995e-05,
+      "loss": 0.239,
+      "step": 10707
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.092698639942734e-05,
+      "loss": 0.0777,
+      "step": 10708
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09251968503937e-05,
+      "loss": 0.1314,
+      "step": 10709
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.092340730136006e-05,
+      "loss": 0.1663,
+      "step": 10710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0921617752326416e-05,
+      "loss": 0.2005,
+      "step": 10711
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.091982820329277e-05,
+      "loss": 0.1375,
+      "step": 10712
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0918038654259126e-05,
+      "loss": 0.2721,
+      "step": 10713
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.091624910522549e-05,
+      "loss": 0.1613,
+      "step": 10714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.091445955619184e-05,
+      "loss": 0.1254,
+      "step": 10715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09126700071582e-05,
+      "loss": 0.1483,
+      "step": 10716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0910880458124554e-05,
+      "loss": 0.1271,
+      "step": 10717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.090909090909091e-05,
+      "loss": 0.0654,
+      "step": 10718
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0907301360057264e-05,
+      "loss": 0.2123,
+      "step": 10719
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0905511811023626e-05,
+      "loss": 0.1041,
+      "step": 10720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.090372226198998e-05,
+      "loss": 0.0783,
+      "step": 10721
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0901932712956336e-05,
+      "loss": 0.0918,
+      "step": 10722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.09001431639227e-05,
+      "loss": 0.1861,
+      "step": 10723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.089835361488905e-05,
+      "loss": 0.142,
+      "step": 10724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08965640658554e-05,
+      "loss": 0.1757,
+      "step": 10725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0894774516821764e-05,
+      "loss": 0.1417,
+      "step": 10726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.089298496778812e-05,
+      "loss": 0.1645,
+      "step": 10727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0891195418754474e-05,
+      "loss": 0.1207,
+      "step": 10728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0889405869720836e-05,
+      "loss": 0.2004,
+      "step": 10729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.088761632068719e-05,
+      "loss": 0.1896,
+      "step": 10730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.088582677165354e-05,
+      "loss": 0.1359,
+      "step": 10731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08840372226199e-05,
+      "loss": 0.1644,
+      "step": 10732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.088224767358626e-05,
+      "loss": 0.2899,
+      "step": 10733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.088045812455261e-05,
+      "loss": 0.2419,
+      "step": 10734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0878668575518974e-05,
+      "loss": 0.1299,
+      "step": 10735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.087687902648533e-05,
+      "loss": 0.1596,
+      "step": 10736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0875089477451685e-05,
+      "loss": 0.1892,
+      "step": 10737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.087329992841804e-05,
+      "loss": 0.1341,
+      "step": 10738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0871510379384395e-05,
+      "loss": 0.145,
+      "step": 10739
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086972083035075e-05,
+      "loss": 0.1766,
+      "step": 10740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0867931281317105e-05,
+      "loss": 0.1522,
+      "step": 10741
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086614173228347e-05,
+      "loss": 0.0578,
+      "step": 10742
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086435218324982e-05,
+      "loss": 0.1615,
+      "step": 10743
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086256263421618e-05,
+      "loss": 0.2519,
+      "step": 10744
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086077308518254e-05,
+      "loss": 0.0981,
+      "step": 10745
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0858983536148895e-05,
+      "loss": 0.3268,
+      "step": 10746
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.085719398711524e-05,
+      "loss": 0.1338,
+      "step": 10747
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0855404438081605e-05,
+      "loss": 0.2574,
+      "step": 10748
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.085361488904796e-05,
+      "loss": 0.1252,
+      "step": 10749
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0851825340014316e-05,
+      "loss": 0.1691,
+      "step": 10750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.085003579098068e-05,
+      "loss": 0.2381,
+      "step": 10751
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.084824624194703e-05,
+      "loss": 0.215,
+      "step": 10752
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.084645669291339e-05,
+      "loss": 0.1479,
+      "step": 10753
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.084466714387974e-05,
+      "loss": 0.2269,
+      "step": 10754
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08428775948461e-05,
+      "loss": 0.2407,
+      "step": 10755
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0841088045812454e-05,
+      "loss": 0.1486,
+      "step": 10756
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0839298496778816e-05,
+      "loss": 0.1109,
+      "step": 10757
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.083750894774517e-05,
+      "loss": 0.1843,
+      "step": 10758
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0835719398711526e-05,
+      "loss": 0.1086,
+      "step": 10759
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.083392984967788e-05,
+      "loss": 0.1598,
+      "step": 10760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.083214030064424e-05,
+      "loss": 0.1523,
+      "step": 10761
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08303507516106e-05,
+      "loss": 0.1245,
+      "step": 10762
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.082856120257695e-05,
+      "loss": 0.1972,
+      "step": 10763
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.082677165354331e-05,
+      "loss": 0.152,
+      "step": 10764
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0824982104509664e-05,
+      "loss": 0.0757,
+      "step": 10765
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.082319255547602e-05,
+      "loss": 0.1722,
+      "step": 10766
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.082140300644238e-05,
+      "loss": 0.089,
+      "step": 10767
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0819613457408736e-05,
+      "loss": 0.1685,
+      "step": 10768
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.081782390837509e-05,
+      "loss": 0.2365,
+      "step": 10769
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.081603435934145e-05,
+      "loss": 0.2868,
+      "step": 10770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08142448103078e-05,
+      "loss": 0.2107,
+      "step": 10771
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.081245526127416e-05,
+      "loss": 0.0995,
+      "step": 10772
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.081066571224052e-05,
+      "loss": 0.2395,
+      "step": 10773
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0808876163206874e-05,
+      "loss": 0.1356,
+      "step": 10774
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.080708661417323e-05,
+      "loss": 0.0945,
+      "step": 10775
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0805297065139585e-05,
+      "loss": 0.1358,
+      "step": 10776
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0803507516105947e-05,
+      "loss": 0.0641,
+      "step": 10777
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0801717967072295e-05,
+      "loss": 0.0607,
+      "step": 10778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079992841803866e-05,
+      "loss": 0.1321,
+      "step": 10779
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079813886900501e-05,
+      "loss": 0.2231,
+      "step": 10780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079634931997137e-05,
+      "loss": 0.0967,
+      "step": 10781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079455977093772e-05,
+      "loss": 0.1447,
+      "step": 10782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0792770221904084e-05,
+      "loss": 0.1544,
+      "step": 10783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079098067287044e-05,
+      "loss": 0.1851,
+      "step": 10784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0789191123836795e-05,
+      "loss": 0.0943,
+      "step": 10785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078740157480315e-05,
+      "loss": 0.209,
+      "step": 10786
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0785612025769505e-05,
+      "loss": 0.2147,
+      "step": 10787
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078382247673586e-05,
+      "loss": 0.0454,
+      "step": 10788
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078203292770222e-05,
+      "loss": 0.1566,
+      "step": 10789
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078024337866858e-05,
+      "loss": 0.153,
+      "step": 10790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.077845382963493e-05,
+      "loss": 0.0745,
+      "step": 10791
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0776664280601295e-05,
+      "loss": 0.2016,
+      "step": 10792
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.077487473156765e-05,
+      "loss": 0.0883,
+      "step": 10793
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0773085182534e-05,
+      "loss": 0.088,
+      "step": 10794
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.077129563350036e-05,
+      "loss": 0.2226,
+      "step": 10795
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0769506084466716e-05,
+      "loss": 0.1318,
+      "step": 10796
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076771653543307e-05,
+      "loss": 0.1398,
+      "step": 10797
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0765926986399426e-05,
+      "loss": 0.1333,
+      "step": 10798
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076413743736579e-05,
+      "loss": 0.0979,
+      "step": 10799
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076234788833214e-05,
+      "loss": 0.1289,
+      "step": 10800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.07605583392985e-05,
+      "loss": 0.1282,
+      "step": 10801
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0758768790264853e-05,
+      "loss": 0.4503,
+      "step": 10802
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.075697924123121e-05,
+      "loss": 0.0756,
+      "step": 10803
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0755189692197564e-05,
+      "loss": 0.1207,
+      "step": 10804
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0753400143163926e-05,
+      "loss": 0.1766,
+      "step": 10805
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.075161059413028e-05,
+      "loss": 0.0859,
+      "step": 10806
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0749821045096636e-05,
+      "loss": 0.0842,
+      "step": 10807
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0748031496063e-05,
+      "loss": 0.1544,
+      "step": 10808
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.074624194702935e-05,
+      "loss": 0.2626,
+      "step": 10809
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.07444523979957e-05,
+      "loss": 0.3532,
+      "step": 10810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0742662848962064e-05,
+      "loss": 0.3831,
+      "step": 10811
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.074087329992842e-05,
+      "loss": 0.0505,
+      "step": 10812
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0739083750894774e-05,
+      "loss": 0.3599,
+      "step": 10813
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0737294201861136e-05,
+      "loss": 0.1276,
+      "step": 10814
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.073550465282749e-05,
+      "loss": 0.1425,
+      "step": 10815
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0733715103793847e-05,
+      "loss": 0.1082,
+      "step": 10816
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.07319255547602e-05,
+      "loss": 0.0625,
+      "step": 10817
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.073013600572656e-05,
+      "loss": 0.159,
+      "step": 10818
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.072834645669291e-05,
+      "loss": 0.2663,
+      "step": 10819
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0726556907659274e-05,
+      "loss": 0.296,
+      "step": 10820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.072476735862563e-05,
+      "loss": 0.0571,
+      "step": 10821
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0722977809591984e-05,
+      "loss": 0.0941,
+      "step": 10822
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.072118826055834e-05,
+      "loss": 0.2599,
+      "step": 10823
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.07193987115247e-05,
+      "loss": 0.1369,
+      "step": 10824
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.071760916249105e-05,
+      "loss": 0.1808,
+      "step": 10825
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0715819613457405e-05,
+      "loss": 0.0768,
+      "step": 10826
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.071403006442377e-05,
+      "loss": 0.1162,
+      "step": 10827
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.071224051539012e-05,
+      "loss": 0.16,
+      "step": 10828
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.071045096635648e-05,
+      "loss": 0.1127,
+      "step": 10829
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.070866141732284e-05,
+      "loss": 0.207,
+      "step": 10830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0706871868289195e-05,
+      "loss": 0.1745,
+      "step": 10831
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.070508231925555e-05,
+      "loss": 0.1405,
+      "step": 10832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0703292770221905e-05,
+      "loss": 0.0695,
+      "step": 10833
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.070150322118826e-05,
+      "loss": 0.1246,
+      "step": 10834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0699713672154615e-05,
+      "loss": 0.1808,
+      "step": 10835
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.069792412312098e-05,
+      "loss": 0.0665,
+      "step": 10836
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.069613457408733e-05,
+      "loss": 0.2657,
+      "step": 10837
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.069434502505369e-05,
+      "loss": 0.1706,
+      "step": 10838
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.069255547602004e-05,
+      "loss": 0.0256,
+      "step": 10839
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0690765926986405e-05,
+      "loss": 0.2268,
+      "step": 10840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0688976377952753e-05,
+      "loss": 0.0424,
+      "step": 10841
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0687186828919115e-05,
+      "loss": 0.1582,
+      "step": 10842
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.068539727988547e-05,
+      "loss": 0.1055,
+      "step": 10843
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0683607730851826e-05,
+      "loss": 0.2085,
+      "step": 10844
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.068181818181818e-05,
+      "loss": 0.099,
+      "step": 10845
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.068002863278454e-05,
+      "loss": 0.0752,
+      "step": 10846
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.06782390837509e-05,
+      "loss": 0.1686,
+      "step": 10847
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.067644953471725e-05,
+      "loss": 0.2182,
+      "step": 10848
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.067465998568361e-05,
+      "loss": 0.1334,
+      "step": 10849
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0672870436649964e-05,
+      "loss": 0.0505,
+      "step": 10850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.067108088761632e-05,
+      "loss": 0.2654,
+      "step": 10851
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066929133858268e-05,
+      "loss": 0.2418,
+      "step": 10852
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0667501789549036e-05,
+      "loss": 0.115,
+      "step": 10853
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066571224051539e-05,
+      "loss": 0.1165,
+      "step": 10854
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066392269148175e-05,
+      "loss": 0.2339,
+      "step": 10855
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066213314244811e-05,
+      "loss": 0.1347,
+      "step": 10856
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066034359341446e-05,
+      "loss": 0.1007,
+      "step": 10857
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.065855404438082e-05,
+      "loss": 0.2022,
+      "step": 10858
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0656764495347174e-05,
+      "loss": 0.2284,
+      "step": 10859
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.065497494631353e-05,
+      "loss": 0.1371,
+      "step": 10860
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0653185397279884e-05,
+      "loss": 0.3125,
+      "step": 10861
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0651395848246246e-05,
+      "loss": 0.2839,
+      "step": 10862
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.06496062992126e-05,
+      "loss": 0.199,
+      "step": 10863
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.064781675017896e-05,
+      "loss": 0.1696,
+      "step": 10864
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.064602720114531e-05,
+      "loss": 0.125,
+      "step": 10865
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.064423765211167e-05,
+      "loss": 0.1257,
+      "step": 10866
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.064244810307802e-05,
+      "loss": 0.2245,
+      "step": 10867
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0640658554044384e-05,
+      "loss": 0.1474,
+      "step": 10868
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.063886900501074e-05,
+      "loss": 0.1567,
+      "step": 10869
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0637079455977095e-05,
+      "loss": 0.1193,
+      "step": 10870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.063528990694346e-05,
+      "loss": 0.2223,
+      "step": 10871
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0633500357909805e-05,
+      "loss": 0.071,
+      "step": 10872
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.063171080887616e-05,
+      "loss": 0.2258,
+      "step": 10873
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.062992125984252e-05,
+      "loss": 0.2007,
+      "step": 10874
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.062813171080888e-05,
+      "loss": 0.0687,
+      "step": 10875
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.062634216177523e-05,
+      "loss": 0.2853,
+      "step": 10876
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0624552612741595e-05,
+      "loss": 0.161,
+      "step": 10877
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.062276306370795e-05,
+      "loss": 0.1,
+      "step": 10878
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0620973514674305e-05,
+      "loss": 0.2264,
+      "step": 10879
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061918396564066e-05,
+      "loss": 0.0927,
+      "step": 10880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0617394416607015e-05,
+      "loss": 0.1217,
+      "step": 10881
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061560486757337e-05,
+      "loss": 0.2041,
+      "step": 10882
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061381531853973e-05,
+      "loss": 0.1545,
+      "step": 10883
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061202576950609e-05,
+      "loss": 0.2526,
+      "step": 10884
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.061023622047244e-05,
+      "loss": 0.0673,
+      "step": 10885
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.06084466714388e-05,
+      "loss": 0.2343,
+      "step": 10886
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.060665712240516e-05,
+      "loss": 0.2275,
+      "step": 10887
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.060486757337151e-05,
+      "loss": 0.0772,
+      "step": 10888
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0603078024337864e-05,
+      "loss": 0.1815,
+      "step": 10889
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0601288475304226e-05,
+      "loss": 0.0883,
+      "step": 10890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059949892627058e-05,
+      "loss": 0.0587,
+      "step": 10891
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0597709377236936e-05,
+      "loss": 0.2099,
+      "step": 10892
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.05959198282033e-05,
+      "loss": 0.2635,
+      "step": 10893
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059413027916965e-05,
+      "loss": 0.1829,
+      "step": 10894
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059234073013601e-05,
+      "loss": 0.2875,
+      "step": 10895
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0590551181102364e-05,
+      "loss": 0.1575,
+      "step": 10896
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.058876163206872e-05,
+      "loss": 0.312,
+      "step": 10897
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0586972083035074e-05,
+      "loss": 0.0738,
+      "step": 10898
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0585182534001436e-05,
+      "loss": 0.0826,
+      "step": 10899
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.058339298496779e-05,
+      "loss": 0.0585,
+      "step": 10900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0581603435934146e-05,
+      "loss": 0.3725,
+      "step": 10901
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.05798138869005e-05,
+      "loss": 0.0528,
+      "step": 10902
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.057802433786686e-05,
+      "loss": 0.2613,
+      "step": 10903
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.057623478883321e-05,
+      "loss": 0.0575,
+      "step": 10904
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0574445239799574e-05,
+      "loss": 0.0625,
+      "step": 10905
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.057265569076593e-05,
+      "loss": 0.2477,
+      "step": 10906
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0570866141732284e-05,
+      "loss": 0.0919,
+      "step": 10907
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.056907659269864e-05,
+      "loss": 0.1512,
+      "step": 10908
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0567287043665e-05,
+      "loss": 0.1271,
+      "step": 10909
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0565497494631357e-05,
+      "loss": 0.1583,
+      "step": 10910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0563707945597705e-05,
+      "loss": 0.237,
+      "step": 10911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.056191839656407e-05,
+      "loss": 0.0967,
+      "step": 10912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.056012884753042e-05,
+      "loss": 0.0407,
+      "step": 10913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.055833929849678e-05,
+      "loss": 0.1748,
+      "step": 10914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.055654974946314e-05,
+      "loss": 0.094,
+      "step": 10915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0554760200429495e-05,
+      "loss": 0.1586,
+      "step": 10916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.055297065139585e-05,
+      "loss": 0.246,
+      "step": 10917
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.055118110236221e-05,
+      "loss": 0.1849,
+      "step": 10918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054939155332856e-05,
+      "loss": 0.0919,
+      "step": 10919
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0547602004294915e-05,
+      "loss": 0.146,
+      "step": 10920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054581245526128e-05,
+      "loss": 0.1702,
+      "step": 10921
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054402290622763e-05,
+      "loss": 0.171,
+      "step": 10922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054223335719399e-05,
+      "loss": 0.1686,
+      "step": 10923
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.054044380816034e-05,
+      "loss": 0.407,
+      "step": 10924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0538654259126705e-05,
+      "loss": 0.1494,
+      "step": 10925
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.053686471009306e-05,
+      "loss": 0.1474,
+      "step": 10926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0535075161059415e-05,
+      "loss": 0.1716,
+      "step": 10927
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.053328561202577e-05,
+      "loss": 0.1488,
+      "step": 10928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0531496062992126e-05,
+      "loss": 0.2159,
+      "step": 10929
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.052970651395848e-05,
+      "loss": 0.127,
+      "step": 10930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.052791696492484e-05,
+      "loss": 0.0529,
+      "step": 10931
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.05261274158912e-05,
+      "loss": 0.1279,
+      "step": 10932
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.052433786685755e-05,
+      "loss": 0.1979,
+      "step": 10933
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0522548317823915e-05,
+      "loss": 0.2149,
+      "step": 10934
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0520758768790264e-05,
+      "loss": 0.1859,
+      "step": 10935
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.051896921975662e-05,
+      "loss": 0.1176,
+      "step": 10936
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0517179670722977e-05,
+      "loss": 0.1671,
+      "step": 10937
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0515390121689336e-05,
+      "loss": 0.2688,
+      "step": 10938
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0513600572655694e-05,
+      "loss": 0.1615,
+      "step": 10939
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.051181102362205e-05,
+      "loss": 0.1549,
+      "step": 10940
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0510021474588408e-05,
+      "loss": 0.1801,
+      "step": 10941
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0508231925554763e-05,
+      "loss": 0.1092,
+      "step": 10942
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0506442376521115e-05,
+      "loss": 0.1536,
+      "step": 10943
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0504652827487474e-05,
+      "loss": 0.1237,
+      "step": 10944
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.050286327845383e-05,
+      "loss": 0.0839,
+      "step": 10945
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0501073729420188e-05,
+      "loss": 0.1282,
+      "step": 10946
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0499284180386546e-05,
+      "loss": 0.1531,
+      "step": 10947
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.04974946313529e-05,
+      "loss": 0.0895,
+      "step": 10948
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.049570508231926e-05,
+      "loss": 0.1596,
+      "step": 10949
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0493915533285612e-05,
+      "loss": 0.2447,
+      "step": 10950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0492125984251967e-05,
+      "loss": 0.1235,
+      "step": 10951
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0490336435218326e-05,
+      "loss": 0.0632,
+      "step": 10952
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.048854688618468e-05,
+      "loss": 0.1636,
+      "step": 10953
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.048675733715104e-05,
+      "loss": 0.1132,
+      "step": 10954
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0484967788117398e-05,
+      "loss": 0.2142,
+      "step": 10955
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0483178239083753e-05,
+      "loss": 0.1899,
+      "step": 10956
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.048138869005011e-05,
+      "loss": 0.0689,
+      "step": 10957
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0479599141016463e-05,
+      "loss": 0.1555,
+      "step": 10958
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.047780959198282e-05,
+      "loss": 0.1563,
+      "step": 10959
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0476020042949177e-05,
+      "loss": 0.185,
+      "step": 10960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0474230493915536e-05,
+      "loss": 0.1529,
+      "step": 10961
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.047244094488189e-05,
+      "loss": 0.1223,
+      "step": 10962
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.047065139584825e-05,
+      "loss": 0.2724,
+      "step": 10963
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0468861846814605e-05,
+      "loss": 0.292,
+      "step": 10964
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0467072297780963e-05,
+      "loss": 0.2788,
+      "step": 10965
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0465282748747315e-05,
+      "loss": 0.2583,
+      "step": 10966
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.046349319971367e-05,
+      "loss": 0.0646,
+      "step": 10967
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.046170365068003e-05,
+      "loss": 0.116,
+      "step": 10968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0459914101646388e-05,
+      "loss": 0.202,
+      "step": 10969
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0458124552612743e-05,
+      "loss": 0.1895,
+      "step": 10970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.04563350035791e-05,
+      "loss": 0.0928,
+      "step": 10971
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0454545454545456e-05,
+      "loss": 0.2181,
+      "step": 10972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0452755905511815e-05,
+      "loss": 0.2321,
+      "step": 10973
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0450966356478167e-05,
+      "loss": 0.1869,
+      "step": 10974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0449176807444525e-05,
+      "loss": 0.1457,
+      "step": 10975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.044738725841088e-05,
+      "loss": 0.0731,
+      "step": 10976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.044559770937724e-05,
+      "loss": 0.1145,
+      "step": 10977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0443808160343594e-05,
+      "loss": 0.1079,
+      "step": 10978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0442018611309953e-05,
+      "loss": 0.1594,
+      "step": 10979
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0440229062276308e-05,
+      "loss": 0.1576,
+      "step": 10980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0438439513242667e-05,
+      "loss": 0.1165,
+      "step": 10981
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.043664996420902e-05,
+      "loss": 0.2198,
+      "step": 10982
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0434860415175377e-05,
+      "loss": 0.2067,
+      "step": 10983
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0433070866141732e-05,
+      "loss": 0.1435,
+      "step": 10984
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.043128131710809e-05,
+      "loss": 0.1344,
+      "step": 10985
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0429491768074446e-05,
+      "loss": 0.1268,
+      "step": 10986
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0427702219040805e-05,
+      "loss": 0.1447,
+      "step": 10987
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0425912670007163e-05,
+      "loss": 0.1652,
+      "step": 10988
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0424123120973512e-05,
+      "loss": 0.1943,
+      "step": 10989
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.042233357193987e-05,
+      "loss": 0.1185,
+      "step": 10990
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.042054402290623e-05,
+      "loss": 0.1161,
+      "step": 10991
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0418754473872584e-05,
+      "loss": 0.1963,
+      "step": 10992
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0416964924838943e-05,
+      "loss": 0.1132,
+      "step": 10993
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0415175375805298e-05,
+      "loss": 0.2667,
+      "step": 10994
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0413385826771656e-05,
+      "loss": 0.1895,
+      "step": 10995
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0411596277738015e-05,
+      "loss": 0.0982,
+      "step": 10996
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0409806728704367e-05,
+      "loss": 0.1878,
+      "step": 10997
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0408017179670722e-05,
+      "loss": 0.157,
+      "step": 10998
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.040622763063708e-05,
+      "loss": 0.3492,
+      "step": 10999
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0404438081603436e-05,
+      "loss": 0.1901,
+      "step": 11000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0402648532569794e-05,
+      "loss": 0.2641,
+      "step": 11001
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0400858983536153e-05,
+      "loss": 0.1828,
+      "step": 11002
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0399069434502508e-05,
+      "loss": 0.1246,
+      "step": 11003
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0397279885468867e-05,
+      "loss": 0.1758,
+      "step": 11004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.039549033643522e-05,
+      "loss": 0.0725,
+      "step": 11005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0393700787401574e-05,
+      "loss": 0.2143,
+      "step": 11006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0391911238367932e-05,
+      "loss": 0.1168,
+      "step": 11007
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0390121689334287e-05,
+      "loss": 0.2234,
+      "step": 11008
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0388332140300646e-05,
+      "loss": 0.2673,
+      "step": 11009
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0386542591267005e-05,
+      "loss": 0.1717,
+      "step": 11010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.038475304223336e-05,
+      "loss": 0.177,
+      "step": 11011
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.038296349319972e-05,
+      "loss": 0.144,
+      "step": 11012
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.038117394416607e-05,
+      "loss": 0.2324,
+      "step": 11013
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0379384395132425e-05,
+      "loss": 0.1027,
+      "step": 11014
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0377594846098784e-05,
+      "loss": 0.203,
+      "step": 11015
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.037580529706514e-05,
+      "loss": 0.093,
+      "step": 11016
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0374015748031498e-05,
+      "loss": 0.0931,
+      "step": 11017
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0372226198997856e-05,
+      "loss": 0.1239,
+      "step": 11018
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.037043664996421e-05,
+      "loss": 0.2528,
+      "step": 11019
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.036864710093057e-05,
+      "loss": 0.2666,
+      "step": 11020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0366857551896922e-05,
+      "loss": 0.1048,
+      "step": 11021
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0365068002863277e-05,
+      "loss": 0.2216,
+      "step": 11022
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0363278453829636e-05,
+      "loss": 0.1564,
+      "step": 11023
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0361488904795994e-05,
+      "loss": 0.1726,
+      "step": 11024
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.035969935576235e-05,
+      "loss": 0.1718,
+      "step": 11025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0357909806728708e-05,
+      "loss": 0.1994,
+      "step": 11026
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0356120257695063e-05,
+      "loss": 0.0934,
+      "step": 11027
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0354330708661422e-05,
+      "loss": 0.0614,
+      "step": 11028
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0352541159627774e-05,
+      "loss": 0.1598,
+      "step": 11029
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.035075161059413e-05,
+      "loss": 0.2589,
+      "step": 11030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0348962061560487e-05,
+      "loss": 0.1958,
+      "step": 11031
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0347172512526846e-05,
+      "loss": 0.2391,
+      "step": 11032
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.03453829634932e-05,
+      "loss": 0.1063,
+      "step": 11033
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.034359341445956e-05,
+      "loss": 0.0558,
+      "step": 11034
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0341803865425915e-05,
+      "loss": 0.0542,
+      "step": 11035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0340014316392267e-05,
+      "loss": 0.1089,
+      "step": 11036
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0338224767358625e-05,
+      "loss": 0.254,
+      "step": 11037
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0336435218324984e-05,
+      "loss": 0.0549,
+      "step": 11038
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.033464566929134e-05,
+      "loss": 0.2369,
+      "step": 11039
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0332856120257698e-05,
+      "loss": 0.0896,
+      "step": 11040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0331066571224053e-05,
+      "loss": 0.1324,
+      "step": 11041
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.032927702219041e-05,
+      "loss": 0.2504,
+      "step": 11042
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0327487473156767e-05,
+      "loss": 0.058,
+      "step": 11043
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.032569792412312e-05,
+      "loss": 0.1688,
+      "step": 11044
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0323908375089477e-05,
+      "loss": 0.1762,
+      "step": 11045
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0322118826055836e-05,
+      "loss": 0.1568,
+      "step": 11046
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.032032927702219e-05,
+      "loss": 0.139,
+      "step": 11047
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.031853972798855e-05,
+      "loss": 0.0662,
+      "step": 11048
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0316750178954905e-05,
+      "loss": 0.1081,
+      "step": 11049
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0314960629921263e-05,
+      "loss": 0.1418,
+      "step": 11050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0313171080887622e-05,
+      "loss": 0.1467,
+      "step": 11051
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.031138153185397e-05,
+      "loss": 0.1218,
+      "step": 11052
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.030959198282033e-05,
+      "loss": 0.131,
+      "step": 11053
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0307802433786687e-05,
+      "loss": 0.0754,
+      "step": 11054
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0306012884753043e-05,
+      "loss": 0.2669,
+      "step": 11055
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.03042233357194e-05,
+      "loss": 0.2405,
+      "step": 11056
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0302433786685756e-05,
+      "loss": 0.119,
+      "step": 11057
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0300644237652115e-05,
+      "loss": 0.3032,
+      "step": 11058
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0298854688618473e-05,
+      "loss": 0.0993,
+      "step": 11059
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0297065139584825e-05,
+      "loss": 0.0969,
+      "step": 11060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.029527559055118e-05,
+      "loss": 0.2216,
+      "step": 11061
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.029348604151754e-05,
+      "loss": 0.223,
+      "step": 11062
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0291696492483894e-05,
+      "loss": 0.1589,
+      "step": 11063
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0289906943450253e-05,
+      "loss": 0.2346,
+      "step": 11064
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0288117394416608e-05,
+      "loss": 0.3327,
+      "step": 11065
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0286327845382967e-05,
+      "loss": 0.2772,
+      "step": 11066
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0284538296349325e-05,
+      "loss": 0.1314,
+      "step": 11067
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0282748747315677e-05,
+      "loss": 0.1495,
+      "step": 11068
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0280959198282032e-05,
+      "loss": 0.2021,
+      "step": 11069
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.027916964924839e-05,
+      "loss": 0.2351,
+      "step": 11070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0277380100214746e-05,
+      "loss": 0.1551,
+      "step": 11071
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0275590551181105e-05,
+      "loss": 0.1425,
+      "step": 11072
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0273801002147463e-05,
+      "loss": 0.1388,
+      "step": 11073
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.027201145311382e-05,
+      "loss": 0.1748,
+      "step": 11074
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0270221904080177e-05,
+      "loss": 0.2369,
+      "step": 11075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.026843235504653e-05,
+      "loss": 0.1421,
+      "step": 11076
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0266642806012884e-05,
+      "loss": 0.1494,
+      "step": 11077
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0264853256979242e-05,
+      "loss": 0.0636,
+      "step": 11078
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0263063707945598e-05,
+      "loss": 0.2039,
+      "step": 11079
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0261274158911956e-05,
+      "loss": 0.1502,
+      "step": 11080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0259484609878315e-05,
+      "loss": 0.1704,
+      "step": 11081
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.025769506084467e-05,
+      "loss": 0.1644,
+      "step": 11082
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0255905511811022e-05,
+      "loss": 0.1716,
+      "step": 11083
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.025411596277738e-05,
+      "loss": 0.1122,
+      "step": 11084
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0252326413743736e-05,
+      "loss": 0.1187,
+      "step": 11085
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0250536864710094e-05,
+      "loss": 0.2376,
+      "step": 11086
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0248747315676453e-05,
+      "loss": 0.0678,
+      "step": 11087
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0246957766642808e-05,
+      "loss": 0.2422,
+      "step": 11088
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0245168217609167e-05,
+      "loss": 0.2293,
+      "step": 11089
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0243378668575522e-05,
+      "loss": 0.1872,
+      "step": 11090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0241589119541874e-05,
+      "loss": 0.1829,
+      "step": 11091
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0239799570508232e-05,
+      "loss": 0.1001,
+      "step": 11092
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0238010021474587e-05,
+      "loss": 0.1264,
+      "step": 11093
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0236220472440946e-05,
+      "loss": 0.1387,
+      "step": 11094
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0234430923407304e-05,
+      "loss": 0.1788,
+      "step": 11095
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.023264137437366e-05,
+      "loss": 0.0739,
+      "step": 11096
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0230851825340018e-05,
+      "loss": 0.0708,
+      "step": 11097
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0229062276306373e-05,
+      "loss": 0.0903,
+      "step": 11098
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0227272727272725e-05,
+      "loss": 0.2304,
+      "step": 11099
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0225483178239084e-05,
+      "loss": 0.2603,
+      "step": 11100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.022369362920544e-05,
+      "loss": 0.105,
+      "step": 11101
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0221904080171798e-05,
+      "loss": 0.2536,
+      "step": 11102
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0220114531138156e-05,
+      "loss": 0.1712,
+      "step": 11103
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.021832498210451e-05,
+      "loss": 0.2051,
+      "step": 11104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.021653543307087e-05,
+      "loss": 0.0557,
+      "step": 11105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0214745884037225e-05,
+      "loss": 0.173,
+      "step": 11106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0212956335003577e-05,
+      "loss": 0.1523,
+      "step": 11107
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0211166785969936e-05,
+      "loss": 0.0796,
+      "step": 11108
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0209377236936294e-05,
+      "loss": 0.2074,
+      "step": 11109
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.020758768790265e-05,
+      "loss": 0.2181,
+      "step": 11110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0205798138869008e-05,
+      "loss": 0.2432,
+      "step": 11111
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0204008589835363e-05,
+      "loss": 0.213,
+      "step": 11112
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.020221904080172e-05,
+      "loss": 0.1486,
+      "step": 11113
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.020042949176808e-05,
+      "loss": 0.1728,
+      "step": 11114
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.019863994273443e-05,
+      "loss": 0.0831,
+      "step": 11115
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0196850393700787e-05,
+      "loss": 0.0726,
+      "step": 11116
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0195060844667146e-05,
+      "loss": 0.0816,
+      "step": 11117
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.01932712956335e-05,
+      "loss": 0.2931,
+      "step": 11118
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.019148174659986e-05,
+      "loss": 0.2458,
+      "step": 11119
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0189692197566215e-05,
+      "loss": 0.1744,
+      "step": 11120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0187902648532573e-05,
+      "loss": 0.1801,
+      "step": 11121
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0186113099498925e-05,
+      "loss": 0.1335,
+      "step": 11122
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0184323550465284e-05,
+      "loss": 0.1927,
+      "step": 11123
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.018253400143164e-05,
+      "loss": 0.1584,
+      "step": 11124
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0180744452397998e-05,
+      "loss": 0.2189,
+      "step": 11125
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0178954903364353e-05,
+      "loss": 0.1568,
+      "step": 11126
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.017716535433071e-05,
+      "loss": 0.3907,
+      "step": 11127
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0175375805297066e-05,
+      "loss": 0.1857,
+      "step": 11128
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0173586256263425e-05,
+      "loss": 0.1074,
+      "step": 11129
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0171796707229777e-05,
+      "loss": 0.2309,
+      "step": 11130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0170007158196135e-05,
+      "loss": 0.0647,
+      "step": 11131
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.016821760916249e-05,
+      "loss": 0.1609,
+      "step": 11132
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.016642806012885e-05,
+      "loss": 0.0933,
+      "step": 11133
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0164638511095204e-05,
+      "loss": 0.2044,
+      "step": 11134
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0162848962061563e-05,
+      "loss": 0.2083,
+      "step": 11135
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.016105941302792e-05,
+      "loss": 0.0863,
+      "step": 11136
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0159269863994277e-05,
+      "loss": 0.1302,
+      "step": 11137
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.015748031496063e-05,
+      "loss": 0.1157,
+      "step": 11138
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0155690765926987e-05,
+      "loss": 0.1701,
+      "step": 11139
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0153901216893342e-05,
+      "loss": 0.3387,
+      "step": 11140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.01521116678597e-05,
+      "loss": 0.1592,
+      "step": 11141
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0150322118826056e-05,
+      "loss": 0.1416,
+      "step": 11142
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0148532569792415e-05,
+      "loss": 0.3011,
+      "step": 11143
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0146743020758773e-05,
+      "loss": 0.0947,
+      "step": 11144
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.014495347172513e-05,
+      "loss": 0.0778,
+      "step": 11145
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.014316392269148e-05,
+      "loss": 0.1617,
+      "step": 11146
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.014137437365784e-05,
+      "loss": 0.1795,
+      "step": 11147
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0139584824624194e-05,
+      "loss": 0.1785,
+      "step": 11148
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0137795275590553e-05,
+      "loss": 0.1436,
+      "step": 11149
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.013600572655691e-05,
+      "loss": 0.1213,
+      "step": 11150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0134216177523266e-05,
+      "loss": 0.2112,
+      "step": 11151
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0132426628489625e-05,
+      "loss": 0.2076,
+      "step": 11152
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.013063707945598e-05,
+      "loss": 0.2261,
+      "step": 11153
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0128847530422332e-05,
+      "loss": 0.1233,
+      "step": 11154
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.012705798138869e-05,
+      "loss": 0.1019,
+      "step": 11155
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0125268432355046e-05,
+      "loss": 0.0581,
+      "step": 11156
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0123478883321404e-05,
+      "loss": 0.1404,
+      "step": 11157
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0121689334287763e-05,
+      "loss": 0.1205,
+      "step": 11158
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0119899785254118e-05,
+      "loss": 0.2134,
+      "step": 11159
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0118110236220477e-05,
+      "loss": 0.0512,
+      "step": 11160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0116320687186832e-05,
+      "loss": 0.0788,
+      "step": 11161
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0114531138153184e-05,
+      "loss": 0.2913,
+      "step": 11162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0112741589119542e-05,
+      "loss": 0.1513,
+      "step": 11163
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0110952040085897e-05,
+      "loss": 0.1132,
+      "step": 11164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0109162491052256e-05,
+      "loss": 0.1689,
+      "step": 11165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0107372942018615e-05,
+      "loss": 0.1342,
+      "step": 11166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.010558339298497e-05,
+      "loss": 0.1906,
+      "step": 11167
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.010379384395133e-05,
+      "loss": 0.2299,
+      "step": 11168
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.010200429491768e-05,
+      "loss": 0.1904,
+      "step": 11169
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0100214745884035e-05,
+      "loss": 0.1235,
+      "step": 11170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0098425196850394e-05,
+      "loss": 0.1187,
+      "step": 11171
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0096635647816753e-05,
+      "loss": 0.1761,
+      "step": 11172
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0094846098783108e-05,
+      "loss": 0.0701,
+      "step": 11173
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0093056549749466e-05,
+      "loss": 0.1417,
+      "step": 11174
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.009126700071582e-05,
+      "loss": 0.2141,
+      "step": 11175
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.008947745168218e-05,
+      "loss": 0.1218,
+      "step": 11176
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0087687902648532e-05,
+      "loss": 0.1473,
+      "step": 11177
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0085898353614887e-05,
+      "loss": 0.2404,
+      "step": 11178
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0084108804581246e-05,
+      "loss": 0.147,
+      "step": 11179
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0082319255547604e-05,
+      "loss": 0.2674,
+      "step": 11180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.008052970651396e-05,
+      "loss": 0.2847,
+      "step": 11181
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0078740157480318e-05,
+      "loss": 0.1719,
+      "step": 11182
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0076950608446673e-05,
+      "loss": 0.1706,
+      "step": 11183
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0075161059413032e-05,
+      "loss": 0.1885,
+      "step": 11184
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0073371510379384e-05,
+      "loss": 0.1767,
+      "step": 11185
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0071581961345742e-05,
+      "loss": 0.134,
+      "step": 11186
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0069792412312097e-05,
+      "loss": 0.1011,
+      "step": 11187
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0068002863278456e-05,
+      "loss": 0.1524,
+      "step": 11188
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.006621331424481e-05,
+      "loss": 0.1106,
+      "step": 11189
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.006442376521117e-05,
+      "loss": 0.2621,
+      "step": 11190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0062634216177525e-05,
+      "loss": 0.2073,
+      "step": 11191
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0060844667143884e-05,
+      "loss": 0.1101,
+      "step": 11192
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0059055118110235e-05,
+      "loss": 0.2141,
+      "step": 11193
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0057265569076594e-05,
+      "loss": 0.1625,
+      "step": 11194
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.005547602004295e-05,
+      "loss": 0.0633,
+      "step": 11195
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0053686471009308e-05,
+      "loss": 0.108,
+      "step": 11196
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9454486585461163,
+      "eval_loss": 0.15645097196102142,
+      "eval_runtime": 48.8247,
+      "eval_samples_per_second": 1834.46,
+      "eval_steps_per_second": 28.674,
+      "step": 11196
+    }
+  ],
+  "max_steps": 27990,
+  "num_train_epochs": 5,
+  "total_flos": 5861890147014564.0,
+  "trial_name": null,
+  "trial_params": null
+}